The causal impact of discharge to nursing home on readmission and mortality: A propensity score matched survival analysis

Setup

knitr::opts_chunk$set(cache = FALSE, 
                      warning = FALSE, 
                      message = FALSE, 
                      cache.lazy = FALSE)

options(scipen = 999,cache.lazy = FALSE)

set.seed(42)

# Set flag to regenerate stored artifacts
reload = F

# set to "glm_propensity" for sensitivity analysis concerning effect of alternate propensity score model
# set to "noqualityexcl" for sensitivity analysis concerning effect without excluding discharges at risk for documentation errors
# set to "first_patient" for sensitivity analysis concerning effect of including only the first contact for each patient (instead of excluding repeat patients at the matching stage)
sens_analysis <- "first_patient"

path = paste0(getwd(),"/",sens_analysis)
dir.create(path)
knitr::opts_knit$set(root.dir = path)

suppressPackageStartupMessages({

  library(tidycmprsk)
  library(ggsurvfit)
  library(readxl)
  library(lubridate)
  library(boot)
  library(xgboost)
  library(data.table)
  library(knitr)
  library(tidyverse)
  library(MatchIt)
  library(marginaleffects)
  library(survival)
  library(survminer)
  library(Matrix)
  library(cmprsk)
  library(glmnet)
  library(effectsize)
  library(DT)
})

# Define MDC codes based on NBHW definitions https://www.socialstyrelsen.se/statistik-och-data/klassifikationer-och-koder/drg/drg-koder-och-definitioner/



mdc = c("A" = "Nervous System", 
        "B" = "Eye", 
        "C" = "Ear, Nose, Mouth, And Throat", 
        "D" = "Respiratory System", 
        "E" = "Circulatory System", 
        "F" = "Digestive System", 
        "G" = "Hepatobiliary System and Pancreas", 
        "H" = "Musculoskeletal System \n and Connective Tissue", 
        "J" = "Skin, Subcutaneous \nTissue, and Breast", 
        "K" = "Mammary gland diseases", 
        "L" = "Endocrine, Nutritional,\n and Metabolic System", 
        "M" = "Kidney and Urinary Tract", 
        "N" = "Male Reproductive System", 
        "O" = "Female Reproductive System", 
        "P" = "Pregnancy, Childbirth, and Puerperium", 
        "Q" = "Newborn and Other Neonates (Perinatal Period)",
        "R" = "Blood/-Forming and Immun./\nMyeloprolif. Dis. / NS tumors", 
        "S" = "Infectious and Parasitic\n Diseases and Disorders", 
        "T" = "Mental Diseases and Disorders", 
        "U" = "Injuries, Poison, \nand Toxic Effect of Drugs", 
        "V" = "Burns", 
        "W" = "Factors Influencing Health Status", 
        "Z" = "Ungroupable")

mdc_df <- data.frame(mdc = names(mdc),
                     mdc_name = mdc)

## Set parameters

  # Maximum length of stay to include
  max_caredays = 90
  
  # Minimum number of observations from a hospital ward to include
  min_hosp_mvo_obs = 100 
  
  # Time span to search forwards from a contact to identify readmissions
  fu_days = 90
  
  # Time span to search backwards from a contact to identify previous admissions
  prev_days = 365

# Function for getting bootstrap CIs for mean
mean_fun <- function(data,inds){
  return(mean(data[inds]))
}

# Function for calculating number of contacts within a given timeframe
window_dates <- function(d,s,i){
  d = unlist(d)
  l = length(d[d >= s & d <= i])
  return(l)
}


elapsed_months <- function(end_date, start_date) {

  
  sd <- as.POSIXlt(start_date)
  ed <- as.POSIXlt(end_date)
  
  return(12 * (ed$year - sd$year) + (ed$mon - sd$mon))
}

# Function for calculating the quantile of a value

get_q <- function(x,q){sum(x>unlist(q))/(length(unlist(q)))}

# Function for generating sparse matrix for propensity xgboost model

generate_sparsm <- function(data){
  
  cats <- data %>%
    ungroup() %>%
    dplyr::select(id,
           mvo,
           weekday,
           op = interventions,
           prevdiag = prev_last_diag,
           diagprim = diag_prim_last,
           diagsec = diag_sec,
           region,
           muni,
           hosp,
           born,
           civil,
           admit) %>% 
    mutate(mvo = gsub(" ","|",mvo),
           diagprim = gsub(" ","|",diagprim),
           diagsec = gsub(" ","|",diagsec),
           op = gsub(" ","|",op),
           prevdiag = gsub(" ","|",prevdiag),
           muni = gsub(" ","|",muni)
           ) %>%
    mutate(across(everything(),function(x) strsplit(as.character(x),"\\|"))) %>%
    pivot_longer(-id) %>%
    unnest(cols = value,keep_empty = T) %>%
    mutate(id = as.numeric(id),
           name = make.names(paste(name,value,sep = "_")),
           value = 1) %>%
    distinct() %>%
    ungroup()
  
  ids <- unique(cats$id)
  names <- unique(cats$name)
  
  # Map to sparse row/column indices
  id_map <- data.frame(sm_id = seq(1,length(ids)),
                       id = ids)
  
  name_map <- data.frame(sm_d = seq(1,length(names)),
                         name = names)
  
  # Join to data
  cats <- cats %>%
    left_join(id_map) %>%
    left_join(name_map)
  
  
  sprsM_cat <- as(sparseMatrix(i = cats$sm_id,
                               j = cats$sm_d,
                               x = cats$value,
                               dimnames = list(unique(cats$id)[order(unique(cats$sm_id))],
                                               unique(cats$name)[order(unique(cats$sm_d))])), "dgCMatrix")
  
  
  num <- data %>%
    transmute(date = as.numeric(out_date),
              week,
              caredays = caredays,
              prevSince = ifelse(is.na(prev_caredays),fu_days+1,prev_caredays),
              prevCaredays = ifelse(is.na(days_since_prev),366,days_since_prev),
              planned = planned_contact,
              age,
              countHomecareMonths = n_hc,
              countHomeserviceMonths = n_htj,
              countAmbPlanned = count_yr_ov_planned,
              countAmbUnplanned = count_yr_ov_unplanned,
              countDiags = n_diag,
              countInterventions = n_op,
              female = gender,
              countYr = count_yr,
              countYrUnplanned = count_yr_unplanned) 
  
  
  sprsM <- num %>%
    as.matrix() %>%
    Matrix(sparse = T) %>%
    cbind(sprsM_cat)
  
  return(sprsM)
  
}

Load data

Data = hospital data Sol = SOL data

# Load data 

if(file.exists("./data_final.rda") & 
   file.exists("./excl_n.rda") & !reload){
load("data_final.rda")
load("excl_n.rda")
}else{
  
  parent_dir <- dirname(getwd())
  load(paste0(parent_dir,"/data.rda"))
  load(paste0(parent_dir,"/sol.rda"))
  
  d$disch_sabo <- as.numeric(d$discharge == "Särskilt boende")
  
#df indicating which months patient has NH or short term stay 
sabo <- sol %>%
  filter(BOFORM == 2 | KORTTID == 1) %>%
  dplyr::select(lopnr, date_month,shortterm = KORTTID) %>%
  distinct()

#calculate dates for NH at discharge
sabo_dates <- d %>%
  dplyr::select(lopnr,grp, in_date,out_date,admit,discharge) %>%
  left_join(sabo) %>%
  filter(date_month >= out_date - months(3)) %>%
  group_by(lopnr,grp) %>%
  filter(date_month == min(date_month)) %>%
  ungroup() %>%
  distinct() %>%
  mutate(sabo_diff = date_month - out_date)

#calculate dates for NH at admission
sabo_dates_in <- d %>%
  dplyr::select(lopnr,grp, in_date,out_date,admit,discharge) %>%
  left_join(sabo) %>%
  filter(date_month >= in_date - months(3)) %>%
  group_by(lopnr,grp) %>%
  filter(date_month == min(date_month)) %>%
  ungroup() %>%
  distinct() %>%
  mutate(sabo_diff_in = date_month - in_date)

# Get home service / home care dates
htj <- sol %>%
  filter(BOFORM <= 1 & KORTTID != 1,
         HTJ == 1) %>%
  dplyr::select(lopnr, date_month) %>%
  distinct()

hsl <- sol %>%
  filter(BOFORM <= 1 & KORTTID != 1,
         HSL == 1) %>%
  dplyr::select(lopnr, date_month) %>%
  distinct()
  
htj_dates <- d %>%
  dplyr::select(lopnr,grp, in_date,out_date,admit,discharge) %>%
  left_join(htj) %>%
  filter(date_month >= out_date - months(3)) %>%
  group_by(lopnr,grp) %>%
  filter(date_month == min(date_month)) %>%
  ungroup() %>%
  distinct() %>%
  mutate(htj_diff = date_month - out_date)

hsl_dates <- d %>%
  dplyr::select(lopnr,grp, in_date,out_date,admit,discharge) %>%
  left_join(hsl) %>%
  filter(date_month >= out_date - months(3)) %>%
  group_by(lopnr,grp) %>%
  filter(date_month == min(date_month)) %>%
  ungroup() %>%
  distinct() %>%
  mutate(hsl_diff = date_month - out_date)

d <- d %>%
  left_join(dplyr::select(sabo_dates,lopnr,grp,
                          sabo_date = date_month,
                          sabo_diff,shortterm),
            by=c("lopnr","grp")) %>%
  left_join(dplyr::select(sabo_dates_in,lopnr,grp,
                          sabo_date_in = date_month,
                          sabo_diff_in),
            by=c("lopnr","grp")) %>%
  left_join(dplyr::select(htj_dates,lopnr,grp,
                          htj_date = date_month,
                          htj_diff),
            by=c("lopnr","grp")) %>%
  left_join(dplyr::select(hsl_dates,lopnr,grp,
                          hsl_date = date_month,
                          hsl_diff),
            by=c("lopnr","grp"))


d <- d %>%
  #Exclude if admitted from home but has had NH care registered for at least 2 months before discharge (these patients are likely living in NH already)
  mutate(excl_admit = !(admit == "Ordinärt boende" & sabo_diff_in < -30 ) | is.na(sabo_diff),
  #Exclude if discharged to NH but no record of NH care for at least 2 months after discharge
  excl_discharge_sabo = !(discharge == "Särskilt boende" & (sabo_diff > 30 | is.na(sabo_diff))),
  #Exclude if discharged to home but record of NH care within 2 months
  excl_discharge_home = !(discharge == "Ordinärt boende" & (sabo_diff < 30 & !is.na(sabo_diff))),
  shortterm = ifelse(is.na(shortterm),0,shortterm)) %>%
  ungroup() %>%
  select(-dates,-unplanned_dates)

  


# Apply exclusion criteria, and save the number of included records at each step.
excl_n <- list()

excl_n$orig <- nrow(d)

#Exclude cases prior to 2016 (2015 used to calculate propensity score data)
d <- d %>%
 filter(year(out_date) > 2015)

excl_n$year_2015 <- nrow(d)

d <- d %>%
  filter(planned_contact == 0)

excl_n$unplanned <- nrow(d)

d <- d %>%
  arrange(lopnr,grp) %>%
  mutate(days_to_next_unplanned = ifelse(lead(lopnr) == lopnr,lead(in_date) - out_date,NA))

d <- d %>%
 filter(!is.na(source_discharge),
        !is.na(source_admit),
        !is.na(last_hosp))

excl_n$missingdata <- nrow(d)

d <- d %>%
  filter(!source_discharge %in% c(4,1))

excl_n$discharge <- nrow(d)

d <- d %>%
  filter(caredays > 2)

excl_n$shortstay <- nrow(d)

d <- d %>%
  filter(caredays <= max_caredays)

excl_n$longstay <- nrow(d)

d <- d %>%
  filter(out_date < ymd("20200101") - days(fu_days))

excl_n$endofstudy <- nrow(d)

d <- d %>%
  filter(!is.na(last_hosp_mvo))

excl_n$missingmvo <- nrow(d)

d <- d %>%
  filter((days_since_prev >= 0 | is.na(days_since_prev)) & 
           (out_days_to_next >= 0 | is.na(out_days_to_next)))

excl_n$shortgap <- nrow(d)

d <- d %>%
  filter(source_admit == 3)

excl_n$admitfromhome <- nrow(d)

if(sens_analysis != "noqualityexcl"){
  d <- d %>%
    filter(excl_admit & excl_discharge_sabo & excl_discharge_home)
}


excl_n$uncertain_admit_discharge <- nrow(d)

d <- d %>%
  filter((!is.na(htj_diff) & htj_diff < 30) | disch_sabo == 1)

excl_n$admit_home_no_care <- nrow(d)

d <- d %>%
  filter(n_diag > 1)

excl_n$multi_diag <- nrow(d)

if(sens_analysis != "noqualityexcl"){
  d <- d %>%
    group_by(lopnr) %>%
    mutate(first_disch_nh = min(out_date[disch_sabo == 1 & shortterm == 0],na.rm = T)) %>%
    ungroup() %>%
    filter(in_date < first_disch_nh)
}

excl_n$admitfromhome_after_nhdisch  <- nrow(d)

if(sens_analysis == "first_patient"){
  d <- d %>%
    group_by(lopnr) %>%
    filter(grp == min(grp))

}


excl_n$final <- nrow(d)

# Append home care volume data

htj_sum <- htj %>%
     arrange(lopnr, date_month) %>%
     group_by(lopnr) %>%
     mutate(n_htj = row_number()-1) %>%
  ungroup( )%>%
  select(lopnr, discharge_date_month = date_month, n_htj) %>%
  filter(lopnr %in% d$lopnr)

hsl_sum <- hsl %>%
     arrange(lopnr, date_month) %>%
     group_by(lopnr) %>%
     mutate(n_hc = row_number()-1) %>%
  ungroup() %>%
  select(lopnr, discharge_date_month =  date_month, n_hc) %>%
  filter(lopnr %in% d$lopnr)

muni_sums <- full_join(htj_sum,hsl_sum)

d <- d %>%
     left_join(muni_sums,
               by = c("lopnr","discharge_date_month")) %>%
  ungroup()

d$n_htj[is.na(d$n_htj)] <- 0
d$n_hc[is.na(d$n_hc)] <- 0

  save(d, file = "data_final.rda")
  save(excl_n, file = "excl_n.rda")
}

length(unique(d$lopnr))

## [1] 230815

excl_n_per_stage <- as.data.frame(t(as.data.frame(excl_n))) %>%
  transmute(sum = V1,
            excl = lag(sum)-sum,
            description = c("Patients over 65 with 2+ ICD codes", 
                            "Year > 2015", 
                            "Unplanned admission", 
                            "Missing discharge/admission/hospital data", 
                            "Discharge to home or NH", 
                            "Care duration < 3 days", 
                            "Care duration > 90 days", 
                            "Year < 2020", 
                            "Missing ward information", 
                            "Same day readmission", 
                            "Patient admitted from home", 
                            "Agreeement re: admission and discharge between data sources", 
                            "Discharged home without care services",
                            "Single diagnosis only",
                            "Admission from home discharged to NH following discharge to NH",
                            "Final"))


kable(excl_n_per_stage,row.names = F)

sum	excl	description
2852592	NA	Patients over 65 with 2+ ICD codes
2289148	563444	Year > 2015
1856728	432420	Unplanned admission
1856278	450	Missing discharge/admission/hospital data
1704954	151324	Discharge to home or NH
1376882	328072	Care duration < 3 days
1375519	1363	Care duration > 90 days
1288843	86676	Year < 2020
1287612	1231	Missing ward information
1287108	504	Same day readmission
1188724	98384	Patient admitted from home
1001640	187084	Agreeement re: admission and discharge between data sources
434405	567235	Discharged home without care services
413657	20748	Single diagnosis only
413073	584	Admission from home discharged to NH following discharge to NH
230815	182258	Final

# Final n:
nrow(d)

## [1] 230815

# Generate sparse matrix for xgb


if(file.exists("./sm.rda") & !reload){
  load("./sm.rda")
}else{

  sm <- generate_sparsm(d)
  
  save(sm,file = "./sm.rda")
  
}

Train propensity models

if(file.exists("./propensity_xgb.rda") & !reload){
  load("./propensity_xgb.rda")
  load("./propensity_xgb_cv.rda")
  load("./propensity_glm.rda")
  load("./propensity_glm_cv.rda")
}else{

  params <- list(objective = "binary:logistic")


  propensity_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = sm,
                                                 label = as.logical(d$disch_sabo)),
                             nfold = 5,
                             nrounds = 500,
                             prediction = T, 
                             params = params,
                             early_stopping_rounds = 10)

  propensity_xgb <- xgb.train(data = xgb.DMatrix(data  = sm,
                                                 label = as.logical(d$disch_sabo)),
                              params = params,
                            nrounds = propensity_xgb_cv$early_stop$best_iteration)
  

  
  propensity_glm_cv <- cv.glmnet(x=sm,
                                 y = as.logical(d$disch_sabo))
  
  propensity_glm <- glmnet(x=sm,
                           y = as.logical(d$disch_sabo),
                           lambda = propensity_glm_cv$lambda.min)
  
  
  
  save(propensity_xgb_cv,file = "./propensity_xgb_cv.rda")
  save(propensity_xgb,file = "./propensity_xgb.rda")
  save(propensity_glm_cv,file = "./propensity_glm_cv.rda")
  save(propensity_glm,file = "./propensity_glm.rda")
  
}

## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.465612±0.000518 test-logloss:0.466147±0.001737 
## [2]  train-logloss:0.435975±0.000662 test-logloss:0.436956±0.001432 
## [3]  train-logloss:0.416349±0.000827 test-logloss:0.417722±0.001287 
## [4]  train-logloss:0.402860±0.001096 test-logloss:0.404687±0.001209 
## [5]  train-logloss:0.392846±0.000998 test-logloss:0.395187±0.001348 
## [6]  train-logloss:0.385446±0.001159 test-logloss:0.388342±0.001308 
## [7]  train-logloss:0.379472±0.001211 test-logloss:0.382675±0.001266 
## [8]  train-logloss:0.374565±0.001226 test-logloss:0.378092±0.001412 
## [9]  train-logloss:0.370484±0.000846 test-logloss:0.374479±0.001809 
## [10] train-logloss:0.366886±0.000967 test-logloss:0.371201±0.001792 
## [11] train-logloss:0.363854±0.001140 test-logloss:0.368703±0.001667 
## [12] train-logloss:0.360733±0.000874 test-logloss:0.365991±0.001872 
## [13] train-logloss:0.357992±0.001173 test-logloss:0.363556±0.001725 
## [14] train-logloss:0.355773±0.001176 test-logloss:0.361728±0.001787 
## [15] train-logloss:0.353536±0.001030 test-logloss:0.359959±0.001984 
## [16] train-logloss:0.351447±0.000980 test-logloss:0.358350±0.002160 
## [17] train-logloss:0.349676±0.000907 test-logloss:0.357016±0.002260 
## [18] train-logloss:0.347969±0.000840 test-logloss:0.355597±0.002236 
## [19] train-logloss:0.346369±0.000682 test-logloss:0.354335±0.002364 
## [20] train-logloss:0.344866±0.000756 test-logloss:0.353197±0.002347 
## [21] train-logloss:0.343638±0.000688 test-logloss:0.352212±0.002423 
## [22] train-logloss:0.342460±0.000676 test-logloss:0.351419±0.002515 
## [23] train-logloss:0.340917±0.000765 test-logloss:0.350293±0.002352 
## [24] train-logloss:0.339720±0.000698 test-logloss:0.349445±0.002547 
## [25] train-logloss:0.338612±0.000728 test-logloss:0.348735±0.002525 
## [26] train-logloss:0.337511±0.000911 test-logloss:0.347918±0.002381 
## [27] train-logloss:0.336559±0.000879 test-logloss:0.347275±0.002382 
## [28] train-logloss:0.335378±0.000584 test-logloss:0.346437±0.002637 
## [29] train-logloss:0.334496±0.000700 test-logloss:0.345861±0.002569 
## [30] train-logloss:0.333603±0.000627 test-logloss:0.345340±0.002664 
## [31] train-logloss:0.332781±0.000603 test-logloss:0.344768±0.002678 
## [32] train-logloss:0.332006±0.000556 test-logloss:0.344213±0.002729 
## [33] train-logloss:0.331264±0.000477 test-logloss:0.343771±0.002804 
## [34] train-logloss:0.330328±0.000671 test-logloss:0.343179±0.002590 
## [35] train-logloss:0.329458±0.000900 test-logloss:0.342667±0.002506 
## [36] train-logloss:0.328791±0.000894 test-logloss:0.342289±0.002553 
## [37] train-logloss:0.328092±0.000902 test-logloss:0.341903±0.002469 
## [38] train-logloss:0.327393±0.000855 test-logloss:0.341551±0.002539 
## [39] train-logloss:0.326766±0.000842 test-logloss:0.341184±0.002577 
## [40] train-logloss:0.325928±0.000981 test-logloss:0.340666±0.002582 
## [41] train-logloss:0.325332±0.001000 test-logloss:0.340315±0.002594 
## [42] train-logloss:0.324696±0.000987 test-logloss:0.339933±0.002561 
## [43] train-logloss:0.324144±0.000948 test-logloss:0.339638±0.002526 
## [44] train-logloss:0.323527±0.000944 test-logloss:0.339257±0.002561 
## [45] train-logloss:0.322929±0.000939 test-logloss:0.338958±0.002491 
## [46] train-logloss:0.322271±0.000965 test-logloss:0.338626±0.002409 
## [47] train-logloss:0.321644±0.000941 test-logloss:0.338301±0.002471 
## [48] train-logloss:0.320952±0.000946 test-logloss:0.337977±0.002600 
## [49] train-logloss:0.320503±0.000965 test-logloss:0.337750±0.002608 
## [50] train-logloss:0.319906±0.000918 test-logloss:0.337422±0.002598 
## [51] train-logloss:0.319409±0.000916 test-logloss:0.337167±0.002605 
## [52] train-logloss:0.318881±0.000958 test-logloss:0.336895±0.002622 
## [53] train-logloss:0.318438±0.000998 test-logloss:0.336665±0.002538 
## [54] train-logloss:0.317960±0.000922 test-logloss:0.336434±0.002577 
## [55] train-logloss:0.317465±0.000828 test-logloss:0.336183±0.002553 
## [56] train-logloss:0.316912±0.000921 test-logloss:0.335971±0.002532 
## [57] train-logloss:0.316447±0.000807 test-logloss:0.335725±0.002516 
## [58] train-logloss:0.315997±0.000779 test-logloss:0.335460±0.002570 
## [59] train-logloss:0.315554±0.000701 test-logloss:0.335245±0.002617 
## [60] train-logloss:0.315119±0.000723 test-logloss:0.335050±0.002580 
## [61] train-logloss:0.314683±0.000732 test-logloss:0.334819±0.002571 
## [62] train-logloss:0.314234±0.000774 test-logloss:0.334622±0.002615 
## [63] train-logloss:0.313668±0.000915 test-logloss:0.334280±0.002679 
## [64] train-logloss:0.313258±0.000904 test-logloss:0.334119±0.002704 
## [65] train-logloss:0.312711±0.000795 test-logloss:0.333869±0.002760 
## [66] train-logloss:0.312351±0.000779 test-logloss:0.333705±0.002752 
## [67] train-logloss:0.311846±0.000811 test-logloss:0.333489±0.002631 
## [68] train-logloss:0.311443±0.000734 test-logloss:0.333279±0.002615 
## [69] train-logloss:0.311049±0.000702 test-logloss:0.333116±0.002670 
## [70] train-logloss:0.310648±0.000710 test-logloss:0.332953±0.002604 
## [71] train-logloss:0.310067±0.000665 test-logloss:0.332690±0.002528 
## [72] train-logloss:0.309643±0.000653 test-logloss:0.332502±0.002527 
## [73] train-logloss:0.309316±0.000646 test-logloss:0.332334±0.002540 
## [74] train-logloss:0.308913±0.000606 test-logloss:0.332172±0.002525 
## [75] train-logloss:0.308519±0.000595 test-logloss:0.332004±0.002532 
## [76] train-logloss:0.308018±0.000647 test-logloss:0.331735±0.002582 
## [77] train-logloss:0.307548±0.000702 test-logloss:0.331520±0.002509 
## [78] train-logloss:0.307161±0.000710 test-logloss:0.331370±0.002486 
## [79] train-logloss:0.306818±0.000668 test-logloss:0.331225±0.002452 
## [80] train-logloss:0.306478±0.000668 test-logloss:0.331115±0.002409 
## [81] train-logloss:0.306167±0.000657 test-logloss:0.331029±0.002434 
## [82] train-logloss:0.305782±0.000689 test-logloss:0.330864±0.002475 
## [83] train-logloss:0.305487±0.000613 test-logloss:0.330748±0.002469 
## [84] train-logloss:0.305148±0.000606 test-logloss:0.330630±0.002470 
## [85] train-logloss:0.304868±0.000592 test-logloss:0.330518±0.002447 
## [86] train-logloss:0.304524±0.000636 test-logloss:0.330396±0.002478 
## [87] train-logloss:0.304207±0.000682 test-logloss:0.330271±0.002446 
## [88] train-logloss:0.303878±0.000732 test-logloss:0.330192±0.002457 
## [89] train-logloss:0.303563±0.000741 test-logloss:0.330100±0.002493 
## [90] train-logloss:0.303200±0.000844 test-logloss:0.330000±0.002461 
## [91] train-logloss:0.302865±0.000875 test-logloss:0.329894±0.002460 
## [92] train-logloss:0.302570±0.000813 test-logloss:0.329758±0.002497 
## [93] train-logloss:0.302219±0.000730 test-logloss:0.329612±0.002520 
## [94] train-logloss:0.301852±0.000603 test-logloss:0.329497±0.002550 
## [95] train-logloss:0.301519±0.000611 test-logloss:0.329362±0.002519 
## [96] train-logloss:0.301179±0.000653 test-logloss:0.329280±0.002527 
## [97] train-logloss:0.300911±0.000684 test-logloss:0.329192±0.002528 
## [98] train-logloss:0.300577±0.000674 test-logloss:0.329061±0.002518 
## [99] train-logloss:0.300325±0.000697 test-logloss:0.328970±0.002496 
## [100]    train-logloss:0.300024±0.000736 test-logloss:0.328879±0.002492 
## [101]    train-logloss:0.299746±0.000746 test-logloss:0.328782±0.002506 
## [102]    train-logloss:0.299416±0.000713 test-logloss:0.328686±0.002578 
## [103]    train-logloss:0.299103±0.000648 test-logloss:0.328570±0.002616 
## [104]    train-logloss:0.298813±0.000616 test-logloss:0.328516±0.002575 
## [105]    train-logloss:0.298538±0.000633 test-logloss:0.328427±0.002572 
## [106]    train-logloss:0.298260±0.000641 test-logloss:0.328361±0.002592 
## [107]    train-logloss:0.297853±0.000669 test-logloss:0.328235±0.002483 
## [108]    train-logloss:0.297542±0.000739 test-logloss:0.328116±0.002431 
## [109]    train-logloss:0.297238±0.000707 test-logloss:0.328061±0.002477 
## [110]    train-logloss:0.297003±0.000700 test-logloss:0.328003±0.002512 
## [111]    train-logloss:0.296675±0.000705 test-logloss:0.327924±0.002431 
## [112]    train-logloss:0.296350±0.000707 test-logloss:0.327824±0.002462 
## [113]    train-logloss:0.296034±0.000784 test-logloss:0.327707±0.002452 
## [114]    train-logloss:0.295744±0.000725 test-logloss:0.327616±0.002406 
## [115]    train-logloss:0.295421±0.000665 test-logloss:0.327494±0.002431 
## [116]    train-logloss:0.295162±0.000676 test-logloss:0.327435±0.002488 
## [117]    train-logloss:0.294882±0.000675 test-logloss:0.327353±0.002497 
## [118]    train-logloss:0.294560±0.000761 test-logloss:0.327273±0.002510 
## [119]    train-logloss:0.294266±0.000825 test-logloss:0.327223±0.002513 
## [120]    train-logloss:0.294063±0.000810 test-logloss:0.327165±0.002521 
## [121]    train-logloss:0.293787±0.000775 test-logloss:0.327067±0.002616 
## [122]    train-logloss:0.293567±0.000789 test-logloss:0.327026±0.002590 
## [123]    train-logloss:0.293363±0.000800 test-logloss:0.326978±0.002627 
## [124]    train-logloss:0.293091±0.000796 test-logloss:0.326907±0.002600 
## [125]    train-logloss:0.292804±0.000814 test-logloss:0.326765±0.002444 
## [126]    train-logloss:0.292553±0.000840 test-logloss:0.326705±0.002385 
## [127]    train-logloss:0.292293±0.000819 test-logloss:0.326646±0.002375 
## [128]    train-logloss:0.292078±0.000846 test-logloss:0.326596±0.002364 
## [129]    train-logloss:0.291877±0.000823 test-logloss:0.326558±0.002355 
## [130]    train-logloss:0.291611±0.000854 test-logloss:0.326481±0.002365 
## [131]    train-logloss:0.291360±0.000844 test-logloss:0.326411±0.002378 
## [132]    train-logloss:0.291107±0.000944 test-logloss:0.326362±0.002344 
## [133]    train-logloss:0.290872±0.000911 test-logloss:0.326303±0.002365 
## [134]    train-logloss:0.290637±0.000903 test-logloss:0.326233±0.002350 
## [135]    train-logloss:0.290415±0.000926 test-logloss:0.326157±0.002395 
## [136]    train-logloss:0.290180±0.000929 test-logloss:0.326113±0.002386 
## [137]    train-logloss:0.289931±0.000947 test-logloss:0.326075±0.002402 
## [138]    train-logloss:0.289669±0.000942 test-logloss:0.326012±0.002408 
## [139]    train-logloss:0.289397±0.000933 test-logloss:0.325946±0.002416 
## [140]    train-logloss:0.289193±0.000931 test-logloss:0.325889±0.002409 
## [141]    train-logloss:0.288995±0.000885 test-logloss:0.325805±0.002485 
## [142]    train-logloss:0.288821±0.000859 test-logloss:0.325756±0.002511 
## [143]    train-logloss:0.288589±0.000828 test-logloss:0.325723±0.002473 
## [144]    train-logloss:0.288419±0.000848 test-logloss:0.325669±0.002445 
## [145]    train-logloss:0.288232±0.000829 test-logloss:0.325636±0.002410 
## [146]    train-logloss:0.288057±0.000806 test-logloss:0.325591±0.002374 
## [147]    train-logloss:0.287813±0.000808 test-logloss:0.325571±0.002359 
## [148]    train-logloss:0.287482±0.000717 test-logloss:0.325467±0.002374 
## [149]    train-logloss:0.287177±0.000710 test-logloss:0.325398±0.002301 
## [150]    train-logloss:0.287005±0.000707 test-logloss:0.325360±0.002297 
## [151]    train-logloss:0.286777±0.000687 test-logloss:0.325345±0.002336 
## [152]    train-logloss:0.286528±0.000637 test-logloss:0.325292±0.002314 
## [153]    train-logloss:0.286176±0.000742 test-logloss:0.325204±0.002283 
## [154]    train-logloss:0.285894±0.000845 test-logloss:0.325148±0.002259 
## [155]    train-logloss:0.285670±0.000835 test-logloss:0.325093±0.002249 
## [156]    train-logloss:0.285435±0.000835 test-logloss:0.325044±0.002197 
## [157]    train-logloss:0.285230±0.000854 test-logloss:0.325003±0.002164 
## [158]    train-logloss:0.285086±0.000854 test-logloss:0.324972±0.002160 
## [159]    train-logloss:0.284877±0.000860 test-logloss:0.324914±0.002156 
## [160]    train-logloss:0.284642±0.000852 test-logloss:0.324864±0.002146 
## [161]    train-logloss:0.284403±0.000864 test-logloss:0.324852±0.002148 
## [162]    train-logloss:0.284210±0.000883 test-logloss:0.324804±0.002141 
## [163]    train-logloss:0.284049±0.000887 test-logloss:0.324775±0.002128 
## [164]    train-logloss:0.283843±0.000868 test-logloss:0.324717±0.002130 
## [165]    train-logloss:0.283662±0.000899 test-logloss:0.324688±0.002138 
## [166]    train-logloss:0.283451±0.000917 test-logloss:0.324669±0.002123 
## [167]    train-logloss:0.283239±0.000948 test-logloss:0.324627±0.002100 
## [168]    train-logloss:0.283025±0.000932 test-logloss:0.324588±0.002123 
## [169]    train-logloss:0.282844±0.000942 test-logloss:0.324556±0.002112 
## [170]    train-logloss:0.282684±0.000968 test-logloss:0.324538±0.002117 
## [171]    train-logloss:0.282514±0.000999 test-logloss:0.324531±0.002130 
## [172]    train-logloss:0.282228±0.000989 test-logloss:0.324443±0.002136 
## [173]    train-logloss:0.281973±0.000966 test-logloss:0.324359±0.002160 
## [174]    train-logloss:0.281724±0.001059 test-logloss:0.324353±0.002173 
## [175]    train-logloss:0.281527±0.001111 test-logloss:0.324347±0.002197 
## [176]    train-logloss:0.281343±0.001103 test-logloss:0.324318±0.002204 
## [177]    train-logloss:0.281167±0.001086 test-logloss:0.324267±0.002233 
## [178]    train-logloss:0.281041±0.001064 test-logloss:0.324233±0.002241 
## [179]    train-logloss:0.280933±0.001059 test-logloss:0.324219±0.002246 
## [180]    train-logloss:0.280704±0.000968 test-logloss:0.324173±0.002283 
## [181]    train-logloss:0.280531±0.000923 test-logloss:0.324159±0.002294 
## [182]    train-logloss:0.280372±0.000917 test-logloss:0.324095±0.002298 
## [183]    train-logloss:0.280175±0.000903 test-logloss:0.324055±0.002293 
## [184]    train-logloss:0.279998±0.000927 test-logloss:0.324004±0.002299 
## [185]    train-logloss:0.279814±0.000989 test-logloss:0.323961±0.002278 
## [186]    train-logloss:0.279635±0.001043 test-logloss:0.323908±0.002265 
## [187]    train-logloss:0.279463±0.001097 test-logloss:0.323886±0.002259 
## [188]    train-logloss:0.279231±0.001192 test-logloss:0.323830±0.002231 
## [189]    train-logloss:0.279017±0.001214 test-logloss:0.323806±0.002226 
## [190]    train-logloss:0.278845±0.001284 test-logloss:0.323781±0.002221 
## [191]    train-logloss:0.278722±0.001293 test-logloss:0.323757±0.002229 
## [192]    train-logloss:0.278584±0.001284 test-logloss:0.323742±0.002253 
## [193]    train-logloss:0.278413±0.001275 test-logloss:0.323716±0.002245 
## [194]    train-logloss:0.278257±0.001291 test-logloss:0.323701±0.002239 
## [195]    train-logloss:0.278105±0.001291 test-logloss:0.323670±0.002265 
## [196]    train-logloss:0.277944±0.001290 test-logloss:0.323659±0.002246 
## [197]    train-logloss:0.277815±0.001285 test-logloss:0.323638±0.002256 
## [198]    train-logloss:0.277612±0.001281 test-logloss:0.323606±0.002277 
## [199]    train-logloss:0.277476±0.001317 test-logloss:0.323583±0.002266 
## [200]    train-logloss:0.277277±0.001267 test-logloss:0.323559±0.002287 
## [201]    train-logloss:0.277030±0.001218 test-logloss:0.323490±0.002277 
## [202]    train-logloss:0.276867±0.001195 test-logloss:0.323471±0.002287 
## [203]    train-logloss:0.276737±0.001178 test-logloss:0.323461±0.002309 
## [204]    train-logloss:0.276594±0.001174 test-logloss:0.323439±0.002318 
## [205]    train-logloss:0.276413±0.001201 test-logloss:0.323420±0.002313 
## [206]    train-logloss:0.276271±0.001182 test-logloss:0.323407±0.002305 
## [207]    train-logloss:0.276045±0.001115 test-logloss:0.323349±0.002343 
## [208]    train-logloss:0.275864±0.001062 test-logloss:0.323340±0.002345 
## [209]    train-logloss:0.275728±0.001098 test-logloss:0.323320±0.002335 
## [210]    train-logloss:0.275607±0.001144 test-logloss:0.323316±0.002340 
## [211]    train-logloss:0.275401±0.001204 test-logloss:0.323289±0.002323 
## [212]    train-logloss:0.275214±0.001240 test-logloss:0.323281±0.002322 
## [213]    train-logloss:0.275027±0.001256 test-logloss:0.323236±0.002312 
## [214]    train-logloss:0.274926±0.001278 test-logloss:0.323241±0.002330 
## [215]    train-logloss:0.274783±0.001239 test-logloss:0.323224±0.002342 
## [216]    train-logloss:0.274633±0.001224 test-logloss:0.323209±0.002340 
## [217]    train-logloss:0.274495±0.001214 test-logloss:0.323164±0.002329 
## [218]    train-logloss:0.274298±0.001202 test-logloss:0.323145±0.002321 
## [219]    train-logloss:0.274164±0.001218 test-logloss:0.323136±0.002324 
## [220]    train-logloss:0.274011±0.001222 test-logloss:0.323152±0.002303 
## [221]    train-logloss:0.273841±0.001177 test-logloss:0.323149±0.002340 
## [222]    train-logloss:0.273695±0.001150 test-logloss:0.323137±0.002364 
## [223]    train-logloss:0.273579±0.001162 test-logloss:0.323132±0.002367 
## [224]    train-logloss:0.273405±0.001128 test-logloss:0.323132±0.002340 
## [225]    train-logloss:0.273242±0.001075 test-logloss:0.323109±0.002334 
## [226]    train-logloss:0.272976±0.001105 test-logloss:0.323050±0.002270 
## [227]    train-logloss:0.272842±0.001097 test-logloss:0.323015±0.002277 
## [228]    train-logloss:0.272643±0.001048 test-logloss:0.322997±0.002331 
## [229]    train-logloss:0.272516±0.001025 test-logloss:0.322975±0.002328 
## [230]    train-logloss:0.272387±0.001031 test-logloss:0.322954±0.002299 
## [231]    train-logloss:0.272192±0.001058 test-logloss:0.322967±0.002285 
## [232]    train-logloss:0.272061±0.001069 test-logloss:0.322980±0.002277 
## [233]    train-logloss:0.271876±0.001126 test-logloss:0.322960±0.002306 
## [234]    train-logloss:0.271663±0.001187 test-logloss:0.322897±0.002285 
## [235]    train-logloss:0.271542±0.001183 test-logloss:0.322863±0.002281 
## [236]    train-logloss:0.271413±0.001193 test-logloss:0.322865±0.002283 
## [237]    train-logloss:0.271288±0.001193 test-logloss:0.322831±0.002284 
## [238]    train-logloss:0.271150±0.001175 test-logloss:0.322828±0.002305 
## [239]    train-logloss:0.270930±0.001232 test-logloss:0.322787±0.002283 
## [240]    train-logloss:0.270751±0.001250 test-logloss:0.322751±0.002269 
## [241]    train-logloss:0.270599±0.001224 test-logloss:0.322728±0.002269 
## [242]    train-logloss:0.270469±0.001216 test-logloss:0.322739±0.002273 
## [243]    train-logloss:0.270338±0.001226 test-logloss:0.322739±0.002278 
## [244]    train-logloss:0.270195±0.001269 test-logloss:0.322714±0.002220 
## [245]    train-logloss:0.270046±0.001290 test-logloss:0.322697±0.002209 
## [246]    train-logloss:0.269925±0.001290 test-logloss:0.322689±0.002223 
## [247]    train-logloss:0.269798±0.001296 test-logloss:0.322670±0.002232 
## [248]    train-logloss:0.269666±0.001320 test-logloss:0.322662±0.002214 
## [249]    train-logloss:0.269575±0.001335 test-logloss:0.322649±0.002230 
## [250]    train-logloss:0.269424±0.001344 test-logloss:0.322660±0.002220 
## [251]    train-logloss:0.269244±0.001265 test-logloss:0.322652±0.002206 
## [252]    train-logloss:0.269103±0.001195 test-logloss:0.322627±0.002212 
## [253]    train-logloss:0.268959±0.001166 test-logloss:0.322590±0.002230 
## [254]    train-logloss:0.268826±0.001175 test-logloss:0.322570±0.002210 
## [255]    train-logloss:0.268700±0.001224 test-logloss:0.322565±0.002212 
## [256]    train-logloss:0.268597±0.001252 test-logloss:0.322570±0.002212 
## [257]    train-logloss:0.268479±0.001227 test-logloss:0.322539±0.002211 
## [258]    train-logloss:0.268327±0.001155 test-logloss:0.322536±0.002211 
## [259]    train-logloss:0.268187±0.001149 test-logloss:0.322515±0.002208 
## [260]    train-logloss:0.268029±0.001164 test-logloss:0.322463±0.002213 
## [261]    train-logloss:0.267853±0.001136 test-logloss:0.322445±0.002198 
## [262]    train-logloss:0.267732±0.001177 test-logloss:0.322420±0.002212 
## [263]    train-logloss:0.267552±0.001245 test-logloss:0.322376±0.002192 
## [264]    train-logloss:0.267382±0.001221 test-logloss:0.322347±0.002173 
## [265]    train-logloss:0.267194±0.001120 test-logloss:0.322362±0.002148 
## [266]    train-logloss:0.267032±0.001103 test-logloss:0.322352±0.002123 
## [267]    train-logloss:0.266866±0.001063 test-logloss:0.322285±0.002158 
## [268]    train-logloss:0.266701±0.001045 test-logloss:0.322286±0.002168 
## [269]    train-logloss:0.266595±0.001085 test-logloss:0.322287±0.002165 
## [270]    train-logloss:0.266513±0.001084 test-logloss:0.322266±0.002164 
## [271]    train-logloss:0.266349±0.001089 test-logloss:0.322264±0.002157 
## [272]    train-logloss:0.266232±0.001133 test-logloss:0.322267±0.002177 
## [273]    train-logloss:0.266121±0.001107 test-logloss:0.322256±0.002198 
## [274]    train-logloss:0.265922±0.001066 test-logloss:0.322241±0.002195 
## [275]    train-logloss:0.265785±0.001137 test-logloss:0.322210±0.002179 
## [276]    train-logloss:0.265696±0.001136 test-logloss:0.322198±0.002196 
## [277]    train-logloss:0.265582±0.001129 test-logloss:0.322185±0.002181 
## [278]    train-logloss:0.265394±0.001144 test-logloss:0.322105±0.002187 
## [279]    train-logloss:0.265240±0.001144 test-logloss:0.322068±0.002192 
## [280]    train-logloss:0.265104±0.001146 test-logloss:0.322061±0.002196 
## [281]    train-logloss:0.264909±0.001130 test-logloss:0.322036±0.002212 
## [282]    train-logloss:0.264713±0.001171 test-logloss:0.322010±0.002189 
## [283]    train-logloss:0.264535±0.001162 test-logloss:0.322009±0.002168 
## [284]    train-logloss:0.264409±0.001190 test-logloss:0.321976±0.002122 
## [285]    train-logloss:0.264283±0.001180 test-logloss:0.321948±0.002114 
## [286]    train-logloss:0.264113±0.001199 test-logloss:0.321949±0.002095 
## [287]    train-logloss:0.263990±0.001226 test-logloss:0.321953±0.002111 
## [288]    train-logloss:0.263862±0.001260 test-logloss:0.321954±0.002071 
## [289]    train-logloss:0.263746±0.001292 test-logloss:0.321963±0.002093 
## [290]    train-logloss:0.263650±0.001313 test-logloss:0.321965±0.002100 
## [291]    train-logloss:0.263543±0.001353 test-logloss:0.321950±0.002110 
## [292]    train-logloss:0.263475±0.001340 test-logloss:0.321938±0.002100 
## [293]    train-logloss:0.263385±0.001359 test-logloss:0.321931±0.002102 
## [294]    train-logloss:0.263276±0.001369 test-logloss:0.321939±0.002108 
## [295]    train-logloss:0.263173±0.001352 test-logloss:0.321917±0.002115 
## [296]    train-logloss:0.263093±0.001353 test-logloss:0.321916±0.002107 
## [297]    train-logloss:0.262949±0.001366 test-logloss:0.321913±0.002108 
## [298]    train-logloss:0.262828±0.001323 test-logloss:0.321923±0.002115 
## [299]    train-logloss:0.262641±0.001331 test-logloss:0.321892±0.002162 
## [300]    train-logloss:0.262508±0.001238 test-logloss:0.321888±0.002162 
## [301]    train-logloss:0.262316±0.001153 test-logloss:0.321846±0.002224 
## [302]    train-logloss:0.262156±0.001140 test-logloss:0.321833±0.002235 
## [303]    train-logloss:0.262007±0.001052 test-logloss:0.321801±0.002230 
## [304]    train-logloss:0.261882±0.001066 test-logloss:0.321778±0.002225 
## [305]    train-logloss:0.261753±0.001022 test-logloss:0.321777±0.002237 
## [306]    train-logloss:0.261619±0.000944 test-logloss:0.321781±0.002251 
## [307]    train-logloss:0.261474±0.001012 test-logloss:0.321744±0.002242 
## [308]    train-logloss:0.261333±0.001020 test-logloss:0.321720±0.002241 
## [309]    train-logloss:0.261184±0.001037 test-logloss:0.321733±0.002255 
## [310]    train-logloss:0.261010±0.001131 test-logloss:0.321720±0.002237 
## [311]    train-logloss:0.260885±0.001200 test-logloss:0.321686±0.002251 
## [312]    train-logloss:0.260775±0.001222 test-logloss:0.321669±0.002244 
## [313]    train-logloss:0.260661±0.001279 test-logloss:0.321653±0.002226 
## [314]    train-logloss:0.260545±0.001327 test-logloss:0.321637±0.002242 
## [315]    train-logloss:0.260417±0.001388 test-logloss:0.321640±0.002269 
## [316]    train-logloss:0.260304±0.001418 test-logloss:0.321647±0.002274 
## [317]    train-logloss:0.260200±0.001428 test-logloss:0.321640±0.002269 
## [318]    train-logloss:0.260088±0.001427 test-logloss:0.321627±0.002272 
## [319]    train-logloss:0.259953±0.001439 test-logloss:0.321598±0.002286 
## [320]    train-logloss:0.259851±0.001432 test-logloss:0.321602±0.002310 
## [321]    train-logloss:0.259763±0.001424 test-logloss:0.321599±0.002308 
## [322]    train-logloss:0.259646±0.001394 test-logloss:0.321614±0.002266 
## [323]    train-logloss:0.259528±0.001351 test-logloss:0.321608±0.002281 
## [324]    train-logloss:0.259429±0.001357 test-logloss:0.321603±0.002269 
## [325]    train-logloss:0.259318±0.001349 test-logloss:0.321584±0.002253 
## [326]    train-logloss:0.259242±0.001351 test-logloss:0.321577±0.002256 
## [327]    train-logloss:0.259146±0.001333 test-logloss:0.321588±0.002261 
## [328]    train-logloss:0.259032±0.001324 test-logloss:0.321581±0.002289 
## [329]    train-logloss:0.258935±0.001342 test-logloss:0.321566±0.002274 
## [330]    train-logloss:0.258837±0.001322 test-logloss:0.321530±0.002269 
## [331]    train-logloss:0.258754±0.001304 test-logloss:0.321532±0.002271 
## [332]    train-logloss:0.258655±0.001298 test-logloss:0.321512±0.002282 
## [333]    train-logloss:0.258464±0.001387 test-logloss:0.321506±0.002270 
## [334]    train-logloss:0.258375±0.001375 test-logloss:0.321490±0.002263 
## [335]    train-logloss:0.258253±0.001437 test-logloss:0.321488±0.002266 
## [336]    train-logloss:0.258142±0.001425 test-logloss:0.321478±0.002230 
## [337]    train-logloss:0.258061±0.001419 test-logloss:0.321475±0.002253 
## [338]    train-logloss:0.257942±0.001431 test-logloss:0.321468±0.002268 
## [339]    train-logloss:0.257841±0.001402 test-logloss:0.321472±0.002266 
## [340]    train-logloss:0.257716±0.001358 test-logloss:0.321453±0.002259 
## [341]    train-logloss:0.257599±0.001297 test-logloss:0.321484±0.002241 
## [342]    train-logloss:0.257499±0.001313 test-logloss:0.321473±0.002219 
## [343]    train-logloss:0.257330±0.001322 test-logloss:0.321469±0.002265 
## [344]    train-logloss:0.257222±0.001354 test-logloss:0.321464±0.002256 
## [345]    train-logloss:0.257119±0.001397 test-logloss:0.321463±0.002262 
## [346]    train-logloss:0.256950±0.001522 test-logloss:0.321414±0.002263 
## [347]    train-logloss:0.256827±0.001535 test-logloss:0.321411±0.002256 
## [348]    train-logloss:0.256687±0.001565 test-logloss:0.321397±0.002260 
## [349]    train-logloss:0.256574±0.001558 test-logloss:0.321372±0.002275 
## [350]    train-logloss:0.256436±0.001544 test-logloss:0.321342±0.002257 
## [351]    train-logloss:0.256301±0.001566 test-logloss:0.321317±0.002184 
## [352]    train-logloss:0.256178±0.001550 test-logloss:0.321305±0.002193 
## [353]    train-logloss:0.256085±0.001553 test-logloss:0.321305±0.002191 
## [354]    train-logloss:0.256010±0.001544 test-logloss:0.321307±0.002181 
## [355]    train-logloss:0.255913±0.001537 test-logloss:0.321319±0.002199 
## [356]    train-logloss:0.255849±0.001560 test-logloss:0.321324±0.002194 
## [357]    train-logloss:0.255745±0.001568 test-logloss:0.321312±0.002194 
## [358]    train-logloss:0.255610±0.001594 test-logloss:0.321315±0.002229 
## [359]    train-logloss:0.255471±0.001597 test-logloss:0.321306±0.002264 
## [360]    train-logloss:0.255356±0.001591 test-logloss:0.321280±0.002288 
## [361]    train-logloss:0.255291±0.001577 test-logloss:0.321277±0.002304 
## [362]    train-logloss:0.255185±0.001600 test-logloss:0.321273±0.002317 
## [363]    train-logloss:0.255083±0.001606 test-logloss:0.321265±0.002309 
## [364]    train-logloss:0.255017±0.001613 test-logloss:0.321241±0.002316 
## [365]    train-logloss:0.254855±0.001632 test-logloss:0.321228±0.002291 
## [366]    train-logloss:0.254734±0.001560 test-logloss:0.321206±0.002314 
## [367]    train-logloss:0.254549±0.001461 test-logloss:0.321184±0.002354 
## [368]    train-logloss:0.254492±0.001443 test-logloss:0.321187±0.002338 
## [369]    train-logloss:0.254369±0.001380 test-logloss:0.321188±0.002342 
## [370]    train-logloss:0.254288±0.001359 test-logloss:0.321168±0.002339 
## [371]    train-logloss:0.254129±0.001323 test-logloss:0.321152±0.002362 
## [372]    train-logloss:0.254034±0.001309 test-logloss:0.321130±0.002362 
## [373]    train-logloss:0.253904±0.001261 test-logloss:0.321124±0.002368 
## [374]    train-logloss:0.253794±0.001236 test-logloss:0.321119±0.002377 
## [375]    train-logloss:0.253698±0.001247 test-logloss:0.321106±0.002385 
## [376]    train-logloss:0.253560±0.001227 test-logloss:0.321092±0.002370 
## [377]    train-logloss:0.253416±0.001215 test-logloss:0.321068±0.002373 
## [378]    train-logloss:0.253332±0.001225 test-logloss:0.321050±0.002360 
## [379]    train-logloss:0.253243±0.001220 test-logloss:0.321025±0.002390 
## [380]    train-logloss:0.253103±0.001307 test-logloss:0.321001±0.002376 
## [381]    train-logloss:0.253019±0.001320 test-logloss:0.320998±0.002361 
## [382]    train-logloss:0.252910±0.001310 test-logloss:0.321001±0.002333 
## [383]    train-logloss:0.252801±0.001320 test-logloss:0.320980±0.002324 
## [384]    train-logloss:0.252706±0.001317 test-logloss:0.320971±0.002320 
## [385]    train-logloss:0.252577±0.001338 test-logloss:0.320960±0.002309 
## [386]    train-logloss:0.252403±0.001296 test-logloss:0.320944±0.002302 
## [387]    train-logloss:0.252338±0.001292 test-logloss:0.320963±0.002300 
## [388]    train-logloss:0.252254±0.001344 test-logloss:0.320965±0.002305 
## [389]    train-logloss:0.252183±0.001347 test-logloss:0.320974±0.002306 
## [390]    train-logloss:0.252113±0.001360 test-logloss:0.320958±0.002307 
## [391]    train-logloss:0.251981±0.001322 test-logloss:0.320944±0.002307 
## [392]    train-logloss:0.251860±0.001307 test-logloss:0.320936±0.002300 
## [393]    train-logloss:0.251768±0.001322 test-logloss:0.320948±0.002305 
## [394]    train-logloss:0.251615±0.001242 test-logloss:0.320939±0.002337 
## [395]    train-logloss:0.251487±0.001247 test-logloss:0.320919±0.002331 
## [396]    train-logloss:0.251382±0.001240 test-logloss:0.320930±0.002347 
## [397]    train-logloss:0.251312±0.001237 test-logloss:0.320930±0.002365 
## [398]    train-logloss:0.251239±0.001253 test-logloss:0.320938±0.002353 
## [399]    train-logloss:0.251195±0.001243 test-logloss:0.320937±0.002356 
## [400]    train-logloss:0.251094±0.001219 test-logloss:0.320907±0.002349 
## [401]    train-logloss:0.250984±0.001240 test-logloss:0.320908±0.002360 
## [402]    train-logloss:0.250923±0.001248 test-logloss:0.320903±0.002357 
## [403]    train-logloss:0.250791±0.001223 test-logloss:0.320914±0.002353 
## [404]    train-logloss:0.250696±0.001218 test-logloss:0.320906±0.002353 
## [405]    train-logloss:0.250603±0.001211 test-logloss:0.320895±0.002341 
## [406]    train-logloss:0.250476±0.001182 test-logloss:0.320897±0.002398 
## [407]    train-logloss:0.250376±0.001155 test-logloss:0.320916±0.002396 
## [408]    train-logloss:0.250278±0.001157 test-logloss:0.320920±0.002371 
## [409]    train-logloss:0.250173±0.001139 test-logloss:0.320940±0.002385 
## [410]    train-logloss:0.250072±0.001105 test-logloss:0.320959±0.002401 
## [411]    train-logloss:0.250016±0.001102 test-logloss:0.320950±0.002396 
## [412]    train-logloss:0.249959±0.001093 test-logloss:0.320945±0.002388 
## [413]    train-logloss:0.249862±0.001115 test-logloss:0.320961±0.002374 
## [414]    train-logloss:0.249725±0.001171 test-logloss:0.320970±0.002373 
## Stopping. Best iteration:
## [415]    train-logloss:0.249650±0.001187 test-logloss:0.320976±0.002388
## 
## [415]    train-logloss:0.249650±0.001187 test-logloss:0.320976±0.002388

d_eval <- d %>% 
  ungroup() %>%
  dplyr::select(id,
                lopnr,
                out_date,
                age,
                gender,
                region,
                muni,
                caredays,
                planned_contact,
                mdc,
                mvo_last,
                n_op,
                n_diag,
                disch_sabo,
                days_since_prev,
                count_yr,
                count_yr_unplanned,
                count_yr_ov_planned,
                count_yr_ov_unplanned,
                unplanreadmit30,
                mort30,
                days_to_next_unplanned,
                mort_days,
                sabo_diff,
                n_hc,
                n_htj) %>% #include months of HC before hospitalization 
  mutate(nh_propensity_xgb = propensity_xgb_cv$cv_predict$pred,
         nh_propensity_glm = predict(propensity_glm,newx = sm),
         nh_propensity = nh_propensity_xgb,
         mortorreadmit30 = pmax(unplanreadmit30,mort30),
         days_since_prev_fill = ifelse(is.na(days_since_prev),366,days_since_prev),
         prop_weight = disch_sabo/nh_propensity + (1-disch_sabo)/(1-nh_propensity),
         prop_strat = cut(nh_propensity,quantile(nh_propensity,probs = seq(0, 1, 0.1)),include.lowest = T),
         ts_days_to_next_unplanned = ifelse(days_to_next_unplanned>=0,days_to_next_unplanned,NA),
         ts_days_to_sabo = ifelse(sabo_diff>=0 & disch_sabo == 0,sabo_diff,NA),
         ts_days_to_mort = ifelse(mort_days>=0,mort_days,NA),
         ts_days_to_any = pmin(ts_days_to_next_unplanned,ts_days_to_mort,na.rm = T),
         
         ts_mort_days_pp = pmin(ts_days_to_mort,
                             ts_days_to_sabo,
                             fu_days,na.rm = T),
         ts_mort_event_pp = ts_days_to_mort == ts_mort_days_pp & !is.na(ts_days_to_mort),
         
         
         ts_mort_days_itt = pmin(ts_days_to_mort,
                             fu_days,na.rm = T),
         ts_mort_event_itt = ts_days_to_mort == ts_mort_days_itt & !is.na(ts_days_to_mort),
         
         ts_any_days_itt = pmin(ts_days_to_any,
                             fu_days,na.rm = T),
         
         ts_any_event_itt = ts_days_to_any == ts_any_days_itt & !is.na(ts_days_to_any),
         
         ts_readmit_days_pp = pmin(ts_days_to_next_unplanned,
                             ts_days_to_mort,
                             ts_days_to_sabo,
                             fu_days,na.rm = T),
         ts_readmit_event_pp = as.factor(ifelse(ts_readmit_days_pp == ts_days_to_next_unplanned & 
                                 !is.na(ts_days_to_next_unplanned), "Readmission",
                               ifelse(ts_readmit_days_pp == ts_days_to_mort & 
                                 !is.na(ts_days_to_mort), "Death","Censored"))),
         
         ts_readmit_days_itt = pmin(ts_days_to_next_unplanned,
                             ts_days_to_mort,
                             fu_days,na.rm = T),
         
         ts_readmit_event_itt = factor(ifelse(ts_readmit_days_itt == ts_days_to_next_unplanned & 
                                         !is.na(ts_days_to_next_unplanned), "Readmission",
                                       ifelse(ts_readmit_days_itt == ts_days_to_mort & 
                                         !is.na(ts_days_to_mort), "Death","Censored")),
                                       levels = c("Censored","Readmission","Death")),
         
         ts_readmit_days_itt_se = pmin(ts_days_to_next_unplanned,
                             ts_days_to_mort,
                             fu_days,na.rm = T),
         ts_readmit_event_itt_se = ts_readmit_days_itt == ts_days_to_next_unplanned &
           !is.na(ts_days_to_next_unplanned),
         ts_mort_days_itt_7 = pmin(ts_mort_days_itt, 7),
         ts_mort_event_itt_7 = ifelse(ts_mort_days_itt <= 7 & ts_mort_event_itt, T, F),
         
         ts_mort_days_itt_30 = pmin(ts_mort_days_itt, 30),
         ts_mort_event_itt_30 = ifelse(ts_mort_days_itt <= 30 & ts_mort_event_itt, T, F),
         
         ts_readmit_days_itt_7 = pmin(ts_readmit_days_itt, 7),
         ts_readmit_event_itt_7 = factor(ifelse(ts_readmit_days_itt <= 7, 
                                                 as.character(ts_readmit_event_itt),
                                                 "Censored"),levels = c("Censored",
                                                                        "Readmission",
                                                                        "Death")),
         
         ts_readmit_days_itt_30 = pmin(ts_readmit_days_itt, 30),
         ts_readmit_event_itt_30 = factor(ifelse(ts_readmit_days_itt <= 30, 
                                                 as.character(ts_readmit_event_itt),
                                                 "Censored"),levels = c("Censored",
                                                                        "Readmission",
                                                                        "Death")),
         ts_any_days_itt_7 = pmin(ts_any_days_itt, 7),
         ts_any_event_itt_7 = ifelse(ts_any_days_itt <= 7 & ts_any_event_itt, T, F),
         
         ts_any_days_itt_30 = pmin(ts_any_days_itt, 30),
         ts_any_event_itt_30 = ifelse(ts_any_days_itt <= 30 & ts_any_event_itt, T, F),
         )

if(sens_analysis == "glm_propensity") {
  d_eval$nh_propensity <- d_eval$nh_propensity_glm
} 

Perform matching

if(file.exists("./m_exact.rda") & !reload){
  #load("./m_full.rda")
  #load("./m_caliper.rda")
  load("./m_exact.rda")
}else{
  
  pct_99 = quantile(d_eval$nh_propensity[d_eval$disch_sabo==0],probs = 0.99)
  
  # Force matchit to use all NH discharges by preventing it from selecting NH patients as controls
  m_d_eval <- d_eval %>% 
    group_by(lopnr) %>%
    filter(disch_sabo == 1 | !(any(disch_sabo == 1) & disch_sabo == 0),
           !(nh_propensity > pct_99 & disch_sabo == 1)) # remove NH discharges above 99th percentile of non-NH discharged propensities to ensure even post matching propensity score distribution
    
  
  # m_ind_full <- matchit(disch_sabo ~ nh_propensity, 
  #              data = m_d_eval, 
  #              method="nearest",
  #              ratio=1,
  #              unit.id = "lopnr" #this argument ensures that only one episode is used 
  #              )
  # 
  # m_ind_caliper <- matchit(disch_sabo ~ nh_propensity, 
  #              data = m_d_eval, 
  #              method="nearest",
  #              caliper = 0.05,
  #              ratio=1,
  #              unit.id = "lopnr"
  #              )
  
  m_ind_exact <- matchit(disch_sabo ~ nh_propensity, 
               data = m_d_eval, 
               method="nearest",
               caliper = 0.05,
               exact = ~ gender + mdc,
               ratio=1,
               unit.id = "lopnr" #this argument ensures that only one episode is used 
               )
  
  # save(m_ind_full,file = "./m_full.rda")
  # save(m_ind_caliper,file = "./m_caliper.rda")
  save(m_ind_exact,file = "./m_exact.rda")
  
}

#m_data_full <- match.data(m_ind_full)
#m_data_caliper <- match.data(m_ind_caliper)
load("./m_exact.rda")
m_data <- match.data(m_ind_exact)

Propensity score validation

mean values by percentile

m_data <- m_data %>%
  ungroup() %>%
  mutate(prop_strat_match = cut(nh_propensity,
                                quantile(nh_propensity,
                                         probs = seq(0, 1, 0.1)),
                                include.lowest = T))

d_strat <- m_data %>%
  group_by(prop_strat_match,disch_sabo) %>%
  summarise(mean_age = mean(age),
            pct_female = mean(gender),
            mean_caredays = mean(caredays),
            mean_n_op = mean(n_op),
            mean_n_diag = mean(n_diag),
            mean_days_since_prev = mean(days_since_prev,na.rm=T),
            mean_count_ov_unplan = mean(count_yr_ov_unplanned),
            mean_count_yr = mean(count_yr),
            mean_count_yr_unplan = mean(count_yr_unplanned),
            pct_unplanreadmit30 = mean(unplanreadmit30),
            pct_mort30 = mean(mort30),
            mean_hc = mean(n_hc),
            mean_htj = mean(n_htj),
            n = n()) %>%
  pivot_longer(cols = -c(prop_strat_match,disch_sabo))

d_strat_grp <- m_data %>%
  group_by(prop_strat_match) %>%
  summarise(mean_age = mean(age),
            pct_female = mean(gender),
            mean_caredays = mean(caredays),
            mean_n_op = mean(n_op),
            mean_n_diag = mean(n_diag),
            mean_days_since_prev = mean(days_since_prev,na.rm=T),
            mean_count_ov_unplan = mean(count_yr_ov_unplanned),
            mean_count_yr = mean(count_yr),
            mean_count_yr_unplan = mean(count_yr_unplanned),
            mean_dischsabo = mean(disch_sabo),
            n_dischsabo = sum(disch_sabo),
            pct_unplanreadmit30 = mean(unplanreadmit30),
            pct_mort30 = mean(mort30),
            mean_hc = mean(n_hc),
            mean_htj = mean(n_htj),
            n = n()) %>%
  pivot_longer(cols = -c(prop_strat_match))

d_props <- m_data %>%
  group_by(disch_sabo) %>%
  summarise(mean_age = mean(age),
            pct_female = mean(gender),
            mean_caredays = mean(caredays),
            mean_n_op = mean(n_op),
            mean_n_diag = mean(n_diag),
            mean_days_since_prev = mean(days_since_prev,na.rm=T),
            mean_count_ov_unplan = mean(count_yr_ov_unplanned),
            mean_count_yr = mean(count_yr),
            mean_count_yr_unplan = mean(count_yr_unplanned),
            pct_unplanreadmit30 = mean(unplanreadmit30),
            pct_mort30 = mean(mort30),
            mean_hc = mean(n_hc),
            mean_htj = mean(n_htj),
            n = n()) %>% 
  pivot_longer(cols = -disch_sabo) %>%
  mutate(prop_strat = "Overall") %>%
  bind_rows(d_strat) %>%
  mutate(prop_strat = factor(prop_strat_match,levels = c("Overall",levels(d_strat$prop_strat_match))),
         prop_strat_n = as.numeric(prop_strat_match)-1,
         discharge = ifelse(disch_sabo == 1,"NH","Home"))


d_props %>%
  #filter(!name %in% c("pct_mort30","pct_unplanreadmit30")) %>% 
ggplot(aes(x = prop_strat_n,y=value,color=discharge))+
  geom_point() +
  scale_x_continuous(breaks = seq(0,10,1)) +
  facet_wrap(~name,scales = "free",ncol = 2) +
  theme(axis.text.x = element_text(angle = 90))

Pseudo r-squared

Pseudo.R2=function(object){
  stopifnot(object$family$family == "binomial")
  object0 = update(object, ~ 1)
  wt <- object$prior.weights # length(wt)
      y = object$y # weighted
  ones = round(y*wt)
  zeros = wt-ones
  fv <- object$fitted.values   # length(fv)
      if (is.null(object$na.action)) fv0 <- object0$fitted.values else
        fv0 <- object0$fitted.values[-object$na.action] # object may have missing values
  resp <- cbind(ones, zeros)
  Y <- apply(resp, 1, function(x) {c(rep(1, x[1]), rep(0, x[2]))} )
  if (is.list(Y)) Y <- unlist(Y) else Y <- c(Y)
  # length(Y); sum(Y)
  fv.exp <- c(apply(cbind(fv, wt), 1, function(x) rep(x[1], x[2])))
  if (is.list(fv.exp)) fv.exp <- unlist(fv.exp) else fv.exp <- c(fv.exp)
  # length(fv.exp)
  fv0.exp <- c(apply(cbind(fv0, wt), 1, function(x) rep(x[1], x[2])))
  if (is.list(fv0.exp)) fv0.exp <- unlist(fv0.exp) else fv0.exp <- c(fv0.exp)
  (ll = sum(log(dbinom(x=Y,size=1,prob=fv.exp))))
  (ll0 = sum(log(dbinom(x=Y,size=1,prob=fv0.exp))))

  n <- length(Y)
  G2 <- -2 * (ll0 - ll)
  McFadden.R2 <- 1 - ll/ll0
  CoxSnell.R2 <- 1 - exp((2 * (ll0 - ll))/n) # Cox & Snell / Maximum likelihood pseudo r-squared
  r2ML.max <- 1 - exp(ll0 * 2/n)
  Nagelkerke.R2 <- CoxSnell.R2/r2ML.max  # Nagelkerke / Cragg & Uhler's pseudo r-squared

  out <- c(llh = ll, llhNull = ll0, G2 = G2, McFadden = McFadden.R2,
           r2ML = CoxSnell.R2, r2CU = Nagelkerke.R2)
  out
}

  disch_test <- glm(disch_sabo ~ nh_propensity, family = "binomial",
                               data = d_eval)
  
  Pseudo.R2(disch_test)

##             llh         llhNull              G2        McFadden            r2ML 
##  -77400.2830848 -121160.5999627   87520.6337559       0.3611761       0.3155782 
##            r2CU 
##       0.4854975

Generate outcome predictions

m_d_risk <- m_data %>%
  left_join(select(d,-names(d)[names(d) %in% names(m_data)],id))


if(file.exists("./m_sm.rda") & !reload){
  load("./m_sm.rda")
}else{

  m_sm <- generate_sparsm(m_d_risk)
  
  save(m_sm,file = "./m_sm.rda")
  
}

if(file.exists("./outcome_mod.rda") & !reload){
  load("./outcome_mod.rda")
}else{
  
  outcome_mod <- list()
  
  params = list(objective = "binary:logistic")
  outcome_mod$mort7_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                            label = m_d_risk$ts_mort_event_itt_7),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
  outcome_mod$mort30_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                             label = m_d_risk$ts_mort_event_itt_30),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
  outcome_mod$mort90_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                             label = m_d_risk$ts_mort_event_itt),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
  outcome_mod$readmit7_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                            label = m_d_risk$ts_readmit_event_itt_7 == "Readmission"),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  outcome_mod$readmit30_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                             label = m_d_risk$ts_readmit_event_itt_30 == "Readmission"),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  outcome_mod$readmit90_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                             label = m_d_risk$ts_readmit_event_itt == "Readmission"),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
    outcome_mod$any7_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                            label = m_d_risk$ts_any_event_itt_7),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
  outcome_mod$any30_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                             label = m_d_risk$ts_any_event_itt_30),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
  outcome_mod$any90_xgb_cv <- xgb.cv(data = xgb.DMatrix(data  = m_sm,
                                             label = m_d_risk$ts_any_event_itt),
                         nrounds = 500,
                         nfold = 5,
                         prediction = T, 
                         early_stopping_rounds = 10,
                         params = params)
  
  
  save(outcome_mod,file = "./outcome_mod.rda")
}

## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.048491±0.001336 test-logloss:0.049513±0.005160 
## [2]  train-logloss:0.046794±0.001297 test-logloss:0.049224±0.005269 
## [3]  train-logloss:0.045576±0.001385 test-logloss:0.049043±0.005115 
## [4]  train-logloss:0.044687±0.001405 test-logloss:0.048804±0.005116 
## [5]  train-logloss:0.043784±0.001311 test-logloss:0.048629±0.005099 
## [6]  train-logloss:0.042914±0.001384 test-logloss:0.048523±0.005191 
## [7]  train-logloss:0.042421±0.001345 test-logloss:0.048441±0.005184 
## [8]  train-logloss:0.041865±0.001383 test-logloss:0.048408±0.005210 
## [9]  train-logloss:0.041354±0.001494 test-logloss:0.048355±0.005203 
## [10] train-logloss:0.040871±0.001562 test-logloss:0.048349±0.005188 
## [11] train-logloss:0.040272±0.001546 test-logloss:0.048270±0.005199 
## [12] train-logloss:0.039936±0.001435 test-logloss:0.048291±0.005199 
## [13] train-logloss:0.039281±0.001560 test-logloss:0.048414±0.005256 
## [14] train-logloss:0.038902±0.001682 test-logloss:0.048356±0.005230 
## [15] train-logloss:0.038558±0.001649 test-logloss:0.048370±0.005276 
## [16] train-logloss:0.038216±0.001685 test-logloss:0.048403±0.005248 
## [17] train-logloss:0.037851±0.001669 test-logloss:0.048438±0.005229 
## [18] train-logloss:0.037642±0.001674 test-logloss:0.048481±0.005285 
## [19] train-logloss:0.037358±0.001660 test-logloss:0.048442±0.005271 
## [20] train-logloss:0.037089±0.001615 test-logloss:0.048453±0.005219 
## Stopping. Best iteration:
## [21] train-logloss:0.036785±0.001566 test-logloss:0.048451±0.005230
## 
## [21] train-logloss:0.036785±0.001566 test-logloss:0.048451±0.005230 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.197837±0.001618 test-logloss:0.199992±0.005848 
## [2]  train-logloss:0.192601±0.001726 test-logloss:0.197199±0.006076 
## [3]  train-logloss:0.189120±0.001739 test-logloss:0.195413±0.006085 
## [4]  train-logloss:0.186429±0.001848 test-logloss:0.194309±0.006125 
## [5]  train-logloss:0.183834±0.001938 test-logloss:0.193796±0.006403 
## [6]  train-logloss:0.181727±0.002248 test-logloss:0.193110±0.006576 
## [7]  train-logloss:0.180009±0.002272 test-logloss:0.192642±0.006620 
## [8]  train-logloss:0.178203±0.001871 test-logloss:0.192247±0.006732 
## [9]  train-logloss:0.176777±0.001650 test-logloss:0.191800±0.006723 
## [10] train-logloss:0.175674±0.001503 test-logloss:0.191384±0.006709 
## [11] train-logloss:0.174421±0.001716 test-logloss:0.191133±0.006608 
## [12] train-logloss:0.173231±0.001686 test-logloss:0.190942±0.006452 
## [13] train-logloss:0.172140±0.001423 test-logloss:0.190743±0.006350 
## [14] train-logloss:0.171180±0.001475 test-logloss:0.190584±0.006210 
## [15] train-logloss:0.170419±0.001488 test-logloss:0.190459±0.006295 
## [16] train-logloss:0.169408±0.001403 test-logloss:0.190378±0.006356 
## [17] train-logloss:0.168530±0.001134 test-logloss:0.190217±0.006389 
## [18] train-logloss:0.167620±0.001207 test-logloss:0.190101±0.006347 
## [19] train-logloss:0.167030±0.001089 test-logloss:0.189964±0.006259 
## [20] train-logloss:0.166289±0.000999 test-logloss:0.189852±0.006227 
## [21] train-logloss:0.165574±0.001263 test-logloss:0.189722±0.006190 
## [22] train-logloss:0.165048±0.001329 test-logloss:0.189678±0.006328 
## [23] train-logloss:0.164447±0.001284 test-logloss:0.189574±0.006350 
## [24] train-logloss:0.163805±0.001552 test-logloss:0.189633±0.006284 
## [25] train-logloss:0.163100±0.001719 test-logloss:0.189647±0.006254 
## [26] train-logloss:0.162527±0.001891 test-logloss:0.189647±0.006253 
## [27] train-logloss:0.161860±0.002098 test-logloss:0.189577±0.006283 
## [28] train-logloss:0.161326±0.002031 test-logloss:0.189456±0.006287 
## [29] train-logloss:0.160747±0.002102 test-logloss:0.189415±0.006392 
## [30] train-logloss:0.160106±0.002366 test-logloss:0.189348±0.006291 
## [31] train-logloss:0.159597±0.002341 test-logloss:0.189351±0.006386 
## [32] train-logloss:0.159122±0.002382 test-logloss:0.189316±0.006455 
## [33] train-logloss:0.158632±0.002317 test-logloss:0.189312±0.006488 
## [34] train-logloss:0.158324±0.002329 test-logloss:0.189213±0.006507 
## [35] train-logloss:0.157799±0.002260 test-logloss:0.189224±0.006444 
## [36] train-logloss:0.157159±0.002380 test-logloss:0.189236±0.006430 
## [37] train-logloss:0.156540±0.002196 test-logloss:0.189258±0.006482 
## [38] train-logloss:0.156230±0.002140 test-logloss:0.189221±0.006463 
## [39] train-logloss:0.155854±0.002048 test-logloss:0.189217±0.006474 
## [40] train-logloss:0.155504±0.001975 test-logloss:0.189188±0.006441 
## [41] train-logloss:0.155127±0.002027 test-logloss:0.189117±0.006434 
## [42] train-logloss:0.154675±0.002088 test-logloss:0.189039±0.006329 
## [43] train-logloss:0.154279±0.002175 test-logloss:0.188971±0.006216 
## [44] train-logloss:0.153919±0.002195 test-logloss:0.188903±0.006208 
## [45] train-logloss:0.153401±0.002025 test-logloss:0.188945±0.006111 
## [46] train-logloss:0.153015±0.002076 test-logloss:0.188889±0.006135 
## [47] train-logloss:0.152618±0.001967 test-logloss:0.188850±0.006094 
## [48] train-logloss:0.152277±0.001899 test-logloss:0.188846±0.006099 
## [49] train-logloss:0.151931±0.001870 test-logloss:0.188914±0.006058 
## [50] train-logloss:0.151667±0.001861 test-logloss:0.188893±0.006079 
## [51] train-logloss:0.151407±0.001873 test-logloss:0.188909±0.006105 
## [52] train-logloss:0.151074±0.001952 test-logloss:0.189025±0.006089 
## [53] train-logloss:0.150640±0.001925 test-logloss:0.188981±0.006110 
## [54] train-logloss:0.150160±0.002025 test-logloss:0.189026±0.006141 
## [55] train-logloss:0.149865±0.001931 test-logloss:0.189093±0.006185 
## [56] train-logloss:0.149598±0.001866 test-logloss:0.189120±0.006246 
## [57] train-logloss:0.149261±0.001917 test-logloss:0.189140±0.006202 
## Stopping. Best iteration:
## [58] train-logloss:0.148926±0.001841 test-logloss:0.189174±0.006224
## 
## [58] train-logloss:0.148926±0.001841 test-logloss:0.189174±0.006224 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.377527±0.000879 test-logloss:0.379048±0.003239 
## [2]  train-logloss:0.366862±0.000685 test-logloss:0.370523±0.003057 
## [3]  train-logloss:0.359134±0.000647 test-logloss:0.364995±0.002749 
## [4]  train-logloss:0.354145±0.000547 test-logloss:0.361509±0.002815 
## [5]  train-logloss:0.349714±0.000835 test-logloss:0.358497±0.002544 
## [6]  train-logloss:0.346297±0.000844 test-logloss:0.356277±0.002462 
## [7]  train-logloss:0.343020±0.001189 test-logloss:0.354770±0.002658 
## [8]  train-logloss:0.340356±0.000771 test-logloss:0.353311±0.002646 
## [9]  train-logloss:0.337904±0.001194 test-logloss:0.352084±0.002770 
## [10] train-logloss:0.335869±0.001433 test-logloss:0.351033±0.002518 
## [11] train-logloss:0.333849±0.001322 test-logloss:0.350219±0.002668 
## [12] train-logloss:0.332091±0.001181 test-logloss:0.349372±0.002849 
## [13] train-logloss:0.330452±0.001120 test-logloss:0.348489±0.002656 
## [14] train-logloss:0.328813±0.001108 test-logloss:0.348041±0.002627 
## [15] train-logloss:0.327673±0.001179 test-logloss:0.347453±0.002721 
## [16] train-logloss:0.326418±0.001000 test-logloss:0.346993±0.002707 
## [17] train-logloss:0.325353±0.001008 test-logloss:0.346763±0.002759 
## [18] train-logloss:0.324121±0.001099 test-logloss:0.346314±0.002781 
## [19] train-logloss:0.323160±0.001173 test-logloss:0.345741±0.002755 
## [20] train-logloss:0.322109±0.001143 test-logloss:0.345324±0.002761 
## [21] train-logloss:0.321025±0.001057 test-logloss:0.344915±0.002877 
## [22] train-logloss:0.319889±0.000837 test-logloss:0.344580±0.002855 
## [23] train-logloss:0.318958±0.001027 test-logloss:0.344288±0.002818 
## [24] train-logloss:0.318244±0.000972 test-logloss:0.344071±0.002882 
## [25] train-logloss:0.317402±0.001033 test-logloss:0.343771±0.002816 
## [26] train-logloss:0.316360±0.001141 test-logloss:0.343735±0.002736 
## [27] train-logloss:0.315535±0.001273 test-logloss:0.343430±0.002647 
## [28] train-logloss:0.314744±0.001132 test-logloss:0.343064±0.002710 
## [29] train-logloss:0.314077±0.001140 test-logloss:0.342848±0.002664 
## [30] train-logloss:0.313304±0.001212 test-logloss:0.342657±0.002745 
## [31] train-logloss:0.312329±0.001312 test-logloss:0.342595±0.002803 
## [32] train-logloss:0.311587±0.001624 test-logloss:0.342394±0.002906 
## [33] train-logloss:0.311024±0.001650 test-logloss:0.342353±0.002939 
## [34] train-logloss:0.310377±0.001601 test-logloss:0.342117±0.003077 
## [35] train-logloss:0.309589±0.001529 test-logloss:0.341906±0.003036 
## [36] train-logloss:0.308978±0.001638 test-logloss:0.341798±0.003065 
## [37] train-logloss:0.308371±0.001631 test-logloss:0.341673±0.003105 
## [38] train-logloss:0.307750±0.001708 test-logloss:0.341627±0.003138 
## [39] train-logloss:0.307194±0.001694 test-logloss:0.341462±0.003165 
## [40] train-logloss:0.306614±0.001663 test-logloss:0.341402±0.003122 
## [41] train-logloss:0.305914±0.001522 test-logloss:0.341336±0.003046 
## [42] train-logloss:0.305241±0.001505 test-logloss:0.341318±0.003112 
## [43] train-logloss:0.304446±0.001412 test-logloss:0.341323±0.002986 
## [44] train-logloss:0.303784±0.001490 test-logloss:0.341221±0.003034 
## [45] train-logloss:0.303174±0.001517 test-logloss:0.341289±0.003077 
## [46] train-logloss:0.302706±0.001516 test-logloss:0.341246±0.003118 
## [47] train-logloss:0.302153±0.001664 test-logloss:0.341154±0.003044 
## [48] train-logloss:0.301570±0.001524 test-logloss:0.341109±0.003028 
## [49] train-logloss:0.300892±0.001581 test-logloss:0.341020±0.003031 
## [50] train-logloss:0.300368±0.001526 test-logloss:0.341016±0.003038 
## [51] train-logloss:0.299928±0.001449 test-logloss:0.340941±0.003049 
## [52] train-logloss:0.299385±0.001334 test-logloss:0.340878±0.003060 
## [53] train-logloss:0.298886±0.001443 test-logloss:0.340772±0.003010 
## [54] train-logloss:0.298346±0.001717 test-logloss:0.340832±0.003148 
## [55] train-logloss:0.297816±0.001623 test-logloss:0.340753±0.003162 
## [56] train-logloss:0.297304±0.001665 test-logloss:0.340727±0.003153 
## [57] train-logloss:0.296848±0.001666 test-logloss:0.340739±0.003127 
## [58] train-logloss:0.296330±0.001581 test-logloss:0.340813±0.003080 
## [59] train-logloss:0.295796±0.001488 test-logloss:0.340769±0.003097 
## [60] train-logloss:0.295357±0.001490 test-logloss:0.340720±0.003155 
## [61] train-logloss:0.294896±0.001426 test-logloss:0.340731±0.003225 
## [62] train-logloss:0.294573±0.001444 test-logloss:0.340656±0.003280 
## [63] train-logloss:0.294183±0.001304 test-logloss:0.340607±0.003184 
## [64] train-logloss:0.293694±0.001287 test-logloss:0.340508±0.003245 
## [65] train-logloss:0.293268±0.001273 test-logloss:0.340503±0.003264 
## [66] train-logloss:0.292838±0.001188 test-logloss:0.340565±0.003163 
## [67] train-logloss:0.292469±0.001233 test-logloss:0.340540±0.003235 
## [68] train-logloss:0.292138±0.001229 test-logloss:0.340475±0.003238 
## [69] train-logloss:0.291853±0.001297 test-logloss:0.340499±0.003285 
## [70] train-logloss:0.291462±0.001257 test-logloss:0.340406±0.003274 
## [71] train-logloss:0.290987±0.001443 test-logloss:0.340430±0.003152 
## [72] train-logloss:0.290570±0.001369 test-logloss:0.340407±0.003140 
## [73] train-logloss:0.290095±0.001424 test-logloss:0.340401±0.003095 
## [74] train-logloss:0.289666±0.001541 test-logloss:0.340369±0.003215 
## [75] train-logloss:0.289151±0.001624 test-logloss:0.340391±0.003136 
## [76] train-logloss:0.288813±0.001730 test-logloss:0.340440±0.003137 
## [77] train-logloss:0.288400±0.001804 test-logloss:0.340423±0.003151 
## [78] train-logloss:0.288064±0.001771 test-logloss:0.340387±0.003122 
## [79] train-logloss:0.287590±0.001736 test-logloss:0.340339±0.003222 
## [80] train-logloss:0.287214±0.001586 test-logloss:0.340363±0.003045 
## [81] train-logloss:0.286906±0.001602 test-logloss:0.340272±0.003120 
## [82] train-logloss:0.286590±0.001611 test-logloss:0.340326±0.003056 
## [83] train-logloss:0.286231±0.001549 test-logloss:0.340420±0.003015 
## [84] train-logloss:0.285885±0.001596 test-logloss:0.340380±0.003081 
## [85] train-logloss:0.285482±0.001632 test-logloss:0.340399±0.003016 
## [86] train-logloss:0.285129±0.001639 test-logloss:0.340423±0.002956 
## [87] train-logloss:0.284777±0.001417 test-logloss:0.340405±0.002936 
## [88] train-logloss:0.284521±0.001400 test-logloss:0.340388±0.002974 
## [89] train-logloss:0.284247±0.001421 test-logloss:0.340363±0.002972 
## [90] train-logloss:0.283941±0.001152 test-logloss:0.340384±0.003001 
## Stopping. Best iteration:
## [91] train-logloss:0.283673±0.001199 test-logloss:0.340344±0.003024
## 
## [91] train-logloss:0.283673±0.001199 test-logloss:0.340344±0.003024 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.221253±0.002654 test-logloss:0.223925±0.010479 
## [2]  train-logloss:0.218223±0.002402 test-logloss:0.223469±0.010663 
## [3]  train-logloss:0.215949±0.002359 test-logloss:0.223191±0.010331 
## [4]  train-logloss:0.213953±0.002338 test-logloss:0.222895±0.010237 
## [5]  train-logloss:0.212507±0.002261 test-logloss:0.222840±0.010411 
## [6]  train-logloss:0.210967±0.002145 test-logloss:0.222910±0.010364 
## [7]  train-logloss:0.209605±0.002191 test-logloss:0.222890±0.010425 
## [8]  train-logloss:0.208456±0.002041 test-logloss:0.222787±0.010328 
## [9]  train-logloss:0.207570±0.001994 test-logloss:0.222805±0.010245 
## [10] train-logloss:0.206763±0.001941 test-logloss:0.222831±0.010188 
## [11] train-logloss:0.205776±0.002007 test-logloss:0.222847±0.010141 
## [12] train-logloss:0.204926±0.001844 test-logloss:0.222825±0.010162 
## [13] train-logloss:0.204377±0.001845 test-logloss:0.222762±0.010129 
## [14] train-logloss:0.203729±0.001843 test-logloss:0.222706±0.010193 
## [15] train-logloss:0.203011±0.001649 test-logloss:0.222691±0.010181 
## [16] train-logloss:0.202405±0.001738 test-logloss:0.222755±0.010203 
## [17] train-logloss:0.201740±0.001718 test-logloss:0.222799±0.010248 
## [18] train-logloss:0.201291±0.001879 test-logloss:0.222857±0.010249 
## [19] train-logloss:0.200794±0.001772 test-logloss:0.222993±0.010243 
## [20] train-logloss:0.200438±0.001871 test-logloss:0.223050±0.010181 
## [21] train-logloss:0.199967±0.001775 test-logloss:0.223064±0.010248 
## [22] train-logloss:0.199474±0.001785 test-logloss:0.223056±0.010219 
## [23] train-logloss:0.199010±0.001917 test-logloss:0.223149±0.010283 
## [24] train-logloss:0.198638±0.001978 test-logloss:0.223194±0.010255 
## Stopping. Best iteration:
## [25] train-logloss:0.198270±0.001980 test-logloss:0.223259±0.010229
## 
## [25] train-logloss:0.198270±0.001980 test-logloss:0.223259±0.010229 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.440578±0.000897 test-logloss:0.443593±0.003555 
## [2]  train-logloss:0.435144±0.001001 test-logloss:0.440802±0.003422 
## [3]  train-logloss:0.431241±0.000919 test-logloss:0.439203±0.003756 
## [4]  train-logloss:0.428442±0.000867 test-logloss:0.438310±0.003558 
## [5]  train-logloss:0.426070±0.000756 test-logloss:0.437851±0.003449 
## [6]  train-logloss:0.423770±0.000731 test-logloss:0.437431±0.003533 
## [7]  train-logloss:0.421810±0.000976 test-logloss:0.437099±0.003618 
## [8]  train-logloss:0.420117±0.001138 test-logloss:0.436997±0.003707 
## [9]  train-logloss:0.418757±0.001397 test-logloss:0.436828±0.003711 
## [10] train-logloss:0.417607±0.001462 test-logloss:0.436611±0.003679 
## [11] train-logloss:0.416606±0.001432 test-logloss:0.436356±0.003584 
## [12] train-logloss:0.415391±0.001341 test-logloss:0.436380±0.003566 
## [13] train-logloss:0.414798±0.001314 test-logloss:0.436341±0.003515 
## [14] train-logloss:0.413878±0.001323 test-logloss:0.436232±0.003528 
## [15] train-logloss:0.412784±0.001457 test-logloss:0.436118±0.003573 
## [16] train-logloss:0.411988±0.001499 test-logloss:0.436054±0.003617 
## [17] train-logloss:0.411255±0.001517 test-logloss:0.435974±0.003610 
## [18] train-logloss:0.410633±0.001534 test-logloss:0.435973±0.003674 
## [19] train-logloss:0.409878±0.001604 test-logloss:0.435972±0.003710 
## [20] train-logloss:0.409101±0.001456 test-logloss:0.435944±0.003677 
## [21] train-logloss:0.408555±0.001596 test-logloss:0.435871±0.003625 
## [22] train-logloss:0.407927±0.001770 test-logloss:0.435918±0.003601 
## [23] train-logloss:0.407200±0.001814 test-logloss:0.435925±0.003553 
## [24] train-logloss:0.406641±0.001680 test-logloss:0.435952±0.003611 
## [25] train-logloss:0.405916±0.001696 test-logloss:0.435911±0.003689 
## [26] train-logloss:0.405381±0.001814 test-logloss:0.435852±0.003700 
## [27] train-logloss:0.404707±0.001858 test-logloss:0.436021±0.003648 
## [28] train-logloss:0.404008±0.001675 test-logloss:0.436105±0.003678 
## [29] train-logloss:0.403506±0.001819 test-logloss:0.436169±0.003648 
## [30] train-logloss:0.402884±0.001649 test-logloss:0.436262±0.003640 
## [31] train-logloss:0.402362±0.001680 test-logloss:0.436231±0.003705 
## [32] train-logloss:0.401867±0.001504 test-logloss:0.436296±0.003712 
## [33] train-logloss:0.401399±0.001570 test-logloss:0.436307±0.003765 
## [34] train-logloss:0.400920±0.001741 test-logloss:0.436276±0.003815 
## [35] train-logloss:0.400403±0.001795 test-logloss:0.436359±0.003811 
## Stopping. Best iteration:
## [36] train-logloss:0.399962±0.001946 test-logloss:0.436380±0.003865
## 
## [36] train-logloss:0.399962±0.001946 test-logloss:0.436380±0.003865 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.590367±0.001237 test-logloss:0.592859±0.004344 
## [2]  train-logloss:0.583628±0.001283 test-logloss:0.588553±0.004198 
## [3]  train-logloss:0.578911±0.001288 test-logloss:0.585998±0.004280 
## [4]  train-logloss:0.575479±0.001479 test-logloss:0.584529±0.004112 
## [5]  train-logloss:0.572919±0.001447 test-logloss:0.583656±0.004098 
## [6]  train-logloss:0.570425±0.001415 test-logloss:0.582861±0.003982 
## [7]  train-logloss:0.568372±0.001554 test-logloss:0.582327±0.003921 
## [8]  train-logloss:0.566581±0.001513 test-logloss:0.582019±0.003761 
## [9]  train-logloss:0.564982±0.001427 test-logloss:0.581698±0.003769 
## [10] train-logloss:0.563768±0.001355 test-logloss:0.581463±0.003849 
## [11] train-logloss:0.562690±0.001281 test-logloss:0.581153±0.003743 
## [12] train-logloss:0.561593±0.001130 test-logloss:0.580939±0.003782 
## [13] train-logloss:0.560631±0.001142 test-logloss:0.580865±0.003756 
## [14] train-logloss:0.559618±0.001212 test-logloss:0.580815±0.003683 
## [15] train-logloss:0.558605±0.001209 test-logloss:0.580708±0.003838 
## [16] train-logloss:0.557663±0.001214 test-logloss:0.580629±0.003773 
## [17] train-logloss:0.556723±0.001302 test-logloss:0.580596±0.003721 
## [18] train-logloss:0.556036±0.001284 test-logloss:0.580629±0.003797 
## [19] train-logloss:0.555235±0.001386 test-logloss:0.580659±0.003901 
## [20] train-logloss:0.554475±0.001570 test-logloss:0.580558±0.003942 
## [21] train-logloss:0.553927±0.001605 test-logloss:0.580513±0.003944 
## [22] train-logloss:0.553337±0.001533 test-logloss:0.580477±0.003947 
## [23] train-logloss:0.552574±0.001620 test-logloss:0.580588±0.004084 
## [24] train-logloss:0.551605±0.001489 test-logloss:0.580620±0.004168 
## [25] train-logloss:0.550787±0.001393 test-logloss:0.580548±0.004181 
## [26] train-logloss:0.550119±0.001284 test-logloss:0.580495±0.004177 
## [27] train-logloss:0.549515±0.001579 test-logloss:0.580548±0.004166 
## [28] train-logloss:0.548936±0.001499 test-logloss:0.580603±0.004082 
## [29] train-logloss:0.548394±0.001470 test-logloss:0.580620±0.004030 
## [30] train-logloss:0.547691±0.001744 test-logloss:0.580591±0.004051 
## [31] train-logloss:0.547205±0.001927 test-logloss:0.580569±0.004053 
## Stopping. Best iteration:
## [32] train-logloss:0.546728±0.001945 test-logloss:0.580630±0.004043
## 
## [32] train-logloss:0.546728±0.001945 test-logloss:0.580630±0.004043 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.241168±0.002269 test-logloss:0.243651±0.008761 
## [2]  train-logloss:0.237847±0.002355 test-logloss:0.242800±0.009050 
## [3]  train-logloss:0.235088±0.002315 test-logloss:0.242142±0.009215 
## [4]  train-logloss:0.233075±0.002624 test-logloss:0.241883±0.009385 
## [5]  train-logloss:0.231541±0.002711 test-logloss:0.241836±0.009523 
## [6]  train-logloss:0.230127±0.002706 test-logloss:0.241707±0.009632 
## [7]  train-logloss:0.229214±0.002690 test-logloss:0.241538±0.009568 
## [8]  train-logloss:0.228150±0.002651 test-logloss:0.241564±0.009576 
## [9]  train-logloss:0.227208±0.002640 test-logloss:0.241567±0.009467 
## [10] train-logloss:0.226261±0.002554 test-logloss:0.241545±0.009631 
## [11] train-logloss:0.225330±0.002497 test-logloss:0.241522±0.009664 
## [12] train-logloss:0.224393±0.002579 test-logloss:0.241623±0.009622 
## [13] train-logloss:0.223541±0.002484 test-logloss:0.241590±0.009760 
## [14] train-logloss:0.223005±0.002371 test-logloss:0.241611±0.009721 
## [15] train-logloss:0.222265±0.002455 test-logloss:0.241600±0.009779 
## [16] train-logloss:0.221758±0.002352 test-logloss:0.241483±0.009779 
## [17] train-logloss:0.221208±0.002373 test-logloss:0.241414±0.009885 
## [18] train-logloss:0.220500±0.002634 test-logloss:0.241469±0.009994 
## [19] train-logloss:0.219853±0.002487 test-logloss:0.241508±0.009930 
## [20] train-logloss:0.219345±0.002406 test-logloss:0.241501±0.009891 
## [21] train-logloss:0.218884±0.002477 test-logloss:0.241552±0.009906 
## [22] train-logloss:0.218468±0.002483 test-logloss:0.241607±0.009909 
## [23] train-logloss:0.217938±0.002468 test-logloss:0.241567±0.009931 
## [24] train-logloss:0.217476±0.002391 test-logloss:0.241479±0.009947 
## [25] train-logloss:0.216966±0.002596 test-logloss:0.241560±0.009937 
## [26] train-logloss:0.216372±0.002648 test-logloss:0.241650±0.009994 
## Stopping. Best iteration:
## [27] train-logloss:0.216014±0.002708 test-logloss:0.241649±0.010046
## 
## [27] train-logloss:0.216014±0.002708 test-logloss:0.241649±0.010046 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.490601±0.001773 test-logloss:0.493428±0.006694 
## [2]  train-logloss:0.483422±0.001850 test-logloss:0.488278±0.006731 
## [3]  train-logloss:0.478417±0.001961 test-logloss:0.485446±0.006618 
## [4]  train-logloss:0.474476±0.002019 test-logloss:0.483687±0.006737 
## [5]  train-logloss:0.471511±0.002260 test-logloss:0.482420±0.006730 
## [6]  train-logloss:0.469067±0.002144 test-logloss:0.481781±0.006793 
## [7]  train-logloss:0.466733±0.002069 test-logloss:0.480980±0.007109 
## [8]  train-logloss:0.464716±0.002339 test-logloss:0.480413±0.007075 
## [9]  train-logloss:0.463008±0.002521 test-logloss:0.479903±0.007156 
## [10] train-logloss:0.461403±0.002154 test-logloss:0.479619±0.007114 
## [11] train-logloss:0.460053±0.002105 test-logloss:0.479329±0.007039 
## [12] train-logloss:0.458918±0.002178 test-logloss:0.479083±0.007031 
## [13] train-logloss:0.457851±0.002145 test-logloss:0.478980±0.007038 
## [14] train-logloss:0.456737±0.002046 test-logloss:0.478826±0.007234 
## [15] train-logloss:0.455844±0.002157 test-logloss:0.478879±0.007221 
## [16] train-logloss:0.454795±0.002246 test-logloss:0.478870±0.007322 
## [17] train-logloss:0.453754±0.002281 test-logloss:0.478863±0.007416 
## [18] train-logloss:0.452970±0.002230 test-logloss:0.478867±0.007335 
## [19] train-logloss:0.452107±0.002299 test-logloss:0.478835±0.007429 
## [20] train-logloss:0.451359±0.002144 test-logloss:0.478767±0.007441 
## [21] train-logloss:0.450464±0.002412 test-logloss:0.478817±0.007599 
## [22] train-logloss:0.449557±0.002414 test-logloss:0.478733±0.007725 
## [23] train-logloss:0.448785±0.002321 test-logloss:0.478781±0.007751 
## [24] train-logloss:0.448070±0.002187 test-logloss:0.478837±0.007938 
## [25] train-logloss:0.447466±0.002254 test-logloss:0.478801±0.007779 
## [26] train-logloss:0.446778±0.002283 test-logloss:0.478680±0.007773 
## [27] train-logloss:0.446148±0.002300 test-logloss:0.478721±0.007883 
## [28] train-logloss:0.445634±0.002305 test-logloss:0.478655±0.007932 
## [29] train-logloss:0.444805±0.002578 test-logloss:0.478593±0.007956 
## [30] train-logloss:0.444103±0.002519 test-logloss:0.478585±0.007934 
## [31] train-logloss:0.443398±0.002561 test-logloss:0.478493±0.007978 
## [32] train-logloss:0.442915±0.002689 test-logloss:0.478543±0.007969 
## [33] train-logloss:0.442202±0.002838 test-logloss:0.478530±0.008114 
## [34] train-logloss:0.441473±0.002864 test-logloss:0.478515±0.008143 
## [35] train-logloss:0.440940±0.002904 test-logloss:0.478592±0.008243 
## [36] train-logloss:0.440402±0.003042 test-logloss:0.478588±0.008232 
## [37] train-logloss:0.439735±0.002997 test-logloss:0.478589±0.008257 
## [38] train-logloss:0.439102±0.003256 test-logloss:0.478657±0.008217 
## [39] train-logloss:0.438637±0.003207 test-logloss:0.478633±0.008192 
## [40] train-logloss:0.438232±0.003159 test-logloss:0.478680±0.008196 
## Stopping. Best iteration:
## [41] train-logloss:0.437712±0.003179 test-logloss:0.478599±0.008170
## 
## [41] train-logloss:0.437712±0.003179 test-logloss:0.478599±0.008170 
## Multiple eval metrics are present. Will use test_logloss for early stopping.
## Will train until test_logloss hasn't improved in 10 rounds.
## 
## [1]  train-logloss:0.635158±0.000446 test-logloss:0.637296±0.001479 
## [2]  train-logloss:0.624816±0.000496 test-logloss:0.629195±0.001668 
## [3]  train-logloss:0.617538±0.000597 test-logloss:0.624125±0.002038 
## [4]  train-logloss:0.612463±0.000674 test-logloss:0.620654±0.001898 
## [5]  train-logloss:0.608682±0.000669 test-logloss:0.618121±0.002076 
## [6]  train-logloss:0.605588±0.000733 test-logloss:0.616264±0.002232 
## [7]  train-logloss:0.602992±0.000864 test-logloss:0.614827±0.002138 
## [8]  train-logloss:0.600629±0.001200 test-logloss:0.613864±0.002244 
## [9]  train-logloss:0.598469±0.001150 test-logloss:0.613088±0.002429 
## [10] train-logloss:0.596699±0.000878 test-logloss:0.612129±0.002417 
## [11] train-logloss:0.595001±0.001027 test-logloss:0.611596±0.002514 
## [12] train-logloss:0.593241±0.001126 test-logloss:0.611064±0.002361 
## [13] train-logloss:0.591842±0.000835 test-logloss:0.610522±0.002331 
## [14] train-logloss:0.590385±0.000849 test-logloss:0.610181±0.002271 
## [15] train-logloss:0.588907±0.001066 test-logloss:0.609981±0.002087 
## [16] train-logloss:0.587796±0.001056 test-logloss:0.609735±0.002202 
## [17] train-logloss:0.586587±0.000728 test-logloss:0.609372±0.002178 
## [18] train-logloss:0.585527±0.000909 test-logloss:0.609001±0.002236 
## [19] train-logloss:0.584461±0.000776 test-logloss:0.608722±0.002181 
## [20] train-logloss:0.583714±0.000700 test-logloss:0.608351±0.002314 
## [21] train-logloss:0.582538±0.000672 test-logloss:0.608175±0.002184 
## [22] train-logloss:0.581477±0.000671 test-logloss:0.607994±0.002319 
## [23] train-logloss:0.580561±0.000739 test-logloss:0.607937±0.002323 
## [24] train-logloss:0.579695±0.000673 test-logloss:0.607841±0.002417 
## [25] train-logloss:0.578728±0.000663 test-logloss:0.607731±0.002546 
## [26] train-logloss:0.577475±0.000491 test-logloss:0.607606±0.002536 
## [27] train-logloss:0.576717±0.000476 test-logloss:0.607508±0.002434 
## [28] train-logloss:0.575858±0.000467 test-logloss:0.607408±0.002401 
## [29] train-logloss:0.575231±0.000423 test-logloss:0.607223±0.002508 
## [30] train-logloss:0.574453±0.000423 test-logloss:0.607124±0.002603 
## [31] train-logloss:0.573686±0.000522 test-logloss:0.607003±0.002509 
## [32] train-logloss:0.573121±0.000462 test-logloss:0.606869±0.002467 
## [33] train-logloss:0.572495±0.000425 test-logloss:0.606761±0.002324 
## [34] train-logloss:0.571896±0.000402 test-logloss:0.606703±0.002231 
## [35] train-logloss:0.571235±0.000460 test-logloss:0.606654±0.002260 
## [36] train-logloss:0.570449±0.000578 test-logloss:0.606578±0.002255 
## [37] train-logloss:0.569778±0.000721 test-logloss:0.606552±0.002293 
## [38] train-logloss:0.569110±0.000706 test-logloss:0.606621±0.002399 
## [39] train-logloss:0.568515±0.000643 test-logloss:0.606585±0.002339 
## [40] train-logloss:0.568030±0.000709 test-logloss:0.606523±0.002297 
## [41] train-logloss:0.567303±0.000635 test-logloss:0.606490±0.002258 
## [42] train-logloss:0.566527±0.000734 test-logloss:0.606489±0.002267 
## [43] train-logloss:0.566102±0.000705 test-logloss:0.606468±0.002212 
## [44] train-logloss:0.565675±0.000811 test-logloss:0.606482±0.002214 
## [45] train-logloss:0.565088±0.000771 test-logloss:0.606425±0.002157 
## [46] train-logloss:0.564524±0.000812 test-logloss:0.606459±0.002231 
## [47] train-logloss:0.563859±0.000881 test-logloss:0.606403±0.002180 
## [48] train-logloss:0.563380±0.000932 test-logloss:0.606380±0.002186 
## [49] train-logloss:0.562831±0.000848 test-logloss:0.606321±0.002216 
## [50] train-logloss:0.562199±0.001024 test-logloss:0.606374±0.002338 
## [51] train-logloss:0.561784±0.001117 test-logloss:0.606382±0.002379 
## [52] train-logloss:0.561289±0.001193 test-logloss:0.606486±0.002350 
## [53] train-logloss:0.560729±0.001313 test-logloss:0.606432±0.002340 
## [54] train-logloss:0.560221±0.001220 test-logloss:0.606443±0.002288 
## [55] train-logloss:0.559788±0.001295 test-logloss:0.606489±0.002256 
## [56] train-logloss:0.559239±0.001249 test-logloss:0.606490±0.002247 
## [57] train-logloss:0.558845±0.001299 test-logloss:0.606426±0.002271 
## [58] train-logloss:0.558302±0.001266 test-logloss:0.606403±0.002270 
## Stopping. Best iteration:
## [59] train-logloss:0.557808±0.001185 test-logloss:0.606402±0.002391
## 
## [59] train-logloss:0.557808±0.001185 test-logloss:0.606402±0.002391

  preds <- data.frame(mort7_pred = unlist(outcome_mod$mort7_xgb_cv$cv_predict),
                      mort30_pred = unlist(outcome_mod$mort30_xgb_cv$cv_predict),
                      mort90_pred = unlist(outcome_mod$mort90_xgb_cv$cv_predict),
                      readmit7_pred = unlist(outcome_mod$readmit7_xgb_cv$cv_predict),
                      readmit30_pred = unlist(outcome_mod$readmit30_xgb_cv$cv_predict),
                      readmit90_pred = unlist(outcome_mod$readmit90_xgb_cv$cv_predict),
                      any7_pred = unlist(outcome_mod$any7_xgb_cv$cv_predict),
                      any30_pred = unlist(outcome_mod$any30_xgb_cv$cv_predict),
                      any90_pred = unlist(outcome_mod$any90_xgb_cv$cv_predict))
  
  
  m_data <- m_data %>%
    as_tibble() %>%
    ungroup() %>%
    bind_cols(preds)

Matching table

meanfun <- function(data, i){
  d <- data[i]
  return(mean(d,na.rm=T))   
}

boot_cis <- function(d, s = meanfun, r=100,rd = 2,paste = T){
  bo <- boot(d, statistic=s, R=r)
  bci <- boot.ci(bo, conf=0.95,type = "perc")
  
  if(paste){
    o <- paste0(round(bci$t0,rd)," (",
              round(bci$percent[4],rd),"-",
              round(bci$percent[5],rd),")")
  } else{
    o <- list(bci$t0,
              bci$percent[4],
              bci$percent[5])
  }
  
  return(o)
}

m_strat <- m_data %>%
  group_by(disch_sabo) %>%
  summarise(n = as.character(n()),
            mean_age = boot_cis(age),
            pct_female = boot_cis(gender),
            mean_caredays = boot_cis(caredays),
            mean_n_op = boot_cis(n_op),
            mean_n_diag = boot_cis(n_diag),
            mean_days_since_prev = boot_cis(days_since_prev),
            mean_count_yr = boot_cis(count_yr),
            mean_count_yr_unplan = boot_cis(count_yr_unplanned),
            mean_hc_prev = boot_cis(n_hc), 
            mean_htj_prev = boot_cis(n_htj), 
            mean_amb_unplanned = boot_cis(count_yr_ov_unplanned),
            mean_amb_planned = boot_cis(count_yr_ov_planned),
            pct_unplanreadmit30 = boot_cis(unplanreadmit30),
            pct_mort30 = boot_cis(mort30)) %>%
  pivot_longer(cols = -c(disch_sabo)) %>%
  mutate(discharge = ifelse(disch_sabo == 1,"NH","Home"))

d_strat <- d %>%
  group_by(disch_sabo) %>%
  summarise(n = as.character(n()),
            mean_age = boot_cis(age),
            pct_female = boot_cis(gender),
            mean_caredays = boot_cis(caredays),
            mean_n_op = boot_cis(n_op),
            mean_n_diag = boot_cis(n_diag),
            mean_days_since_prev = boot_cis(days_since_prev),
            mean_count_yr = boot_cis(count_yr),
            mean_count_yr_unplan = boot_cis(count_yr_unplanned),
            mean_hc_prev = boot_cis(n_hc),
            mean_htj_prev = boot_cis(n_htj),
            mean_amb_unplanned = boot_cis(count_yr_ov_unplanned),
            mean_amb_planned = boot_cis(count_yr_ov_planned),
            pct_unplanreadmit30 = boot_cis(unplanreadmit30),
            pct_mort30 = boot_cis(mort30)) %>%
  pivot_longer(cols = -c(disch_sabo)) %>%
  mutate(discharge = ifelse(disch_sabo == 1,"NH","Home"))

smd <- m_data %>%
  as_tibble() %>%
  select(disch_sabo,
         mean_age = age,
         mean_caredays = caredays,
         mean_n_op = n_op,
         mean_n_diag = n_diag,
         mean_days_since_prev = days_since_prev,
         mean_count_yr = count_yr,
         mean_count_yr_unplan = count_yr_unplanned,
         mean_hc_prev = n_hc,
         mean_htj_prev = n_htj,
         mean_amb_unplanned = count_yr_ov_unplanned,
         mean_amb_planned = count_yr_ov_planned) %>%
  pivot_longer(-disch_sabo, names_to = "variable", values_to = "value") %>%
  group_by(variable) %>%
  summarise(effectsize::cohens_d(value ~ disch_sabo, ci = 0.95)) %>%
  mutate(pasted = paste0(round(Cohens_d,3)," (",round(CI_low,3),"-",round(CI_high,3),")"))

#Balance in original data
d_strat_wide <- d_strat %>%
  select(-disch_sabo) %>%
  pivot_wider(names_from = discharge,
              values_from = value,
              names_prefix = "raw ")

#Balance in matched data
m_strat_wide <- m_strat %>%
  select(-disch_sabo) %>%
  pivot_wider(names_from = discharge,
              values_from = value,
              names_prefix = "matched ")

match_table <- bind_cols(d_strat_wide,select(m_strat_wide,-name)) %>%
  left_join(select(smd,name = variable,smd_ci = pasted))

kable(match_table)

name	raw Home	raw NH	matched Home	matched NH	smd_ci
n	180396	50419	31494	31494	NA
mean_age	81.99 (81.94-82.02)	83.75 (83.67-83.81)	82.77 (82.66-82.86)	82.83 (82.76-82.95)	-0.008 (-0.023-0.008)
pct_female	0.6 (0.6-0.6)	0.59 (0.59-0.6)	0.59 (0.58-0.59)	0.59 (0.58-0.59)	NA
mean_caredays	10.97 (10.93-11.01)	16 (15.85-16.13)	14.59 (14.48-14.72)	14.8 (14.66-14.9)	-0.019 (-0.035–0.003)
mean_n_op	2.74 (2.73-2.76)	3.11 (3.09-3.14)	3.09 (3.05-3.12)	3.06 (3.03-3.1)	0.009 (-0.007-0.024)
mean_n_diag	5.45 (5.44-5.46)	5.79 (5.77-5.81)	5.53 (5.5-5.56)	5.62 (5.58-5.65)	-0.03 (-0.045–0.014)
mean_days_since_prev	122.29 (121.52-123.07)	86.96 (85.44-88.62)	96.6 (94.55-98.64)	97.76 (95.73-99.76)	-0.011 (-0.042-0.02)
mean_count_yr	0.73 (0.72-0.73)	0.59 (0.58-0.6)	0.46 (0.45-0.47)	0.49 (0.47-0.5)	-0.023 (-0.039–0.008)
mean_count_yr_unplan	0.58 (0.58-0.59)	0.51 (0.5-0.52)	0.38 (0.37-0.39)	0.4 (0.39-0.41)	-0.027 (-0.043–0.011)
mean_hc_prev	2.07 (2.04-2.1)	0.16 (0.14-0.17)	0.27 (0.24-0.29)	0.25 (0.23-0.28)	0.009 (-0.007-0.025)
mean_htj_prev	3.99 (3.94-4.03)	0.26 (0.24-0.29)	0.41 (0.38-0.44)	0.4 (0.37-0.44)	0 (-0.015-0.016)
mean_amb_unplanned	1.29 (1.28-1.3)	0.98 (0.96-1)	0.76 (0.74-0.78)	0.81 (0.79-0.84)	-0.032 (-0.048–0.017)
mean_amb_planned	1.94 (1.92-1.98)	1.08 (1.04-1.13)	1.05 (0.99-1.1)	1.1 (1.03-1.15)	-0.011 (-0.027-0.004)
pct_unplanreadmit30	0.21 (0.2-0.21)	0.14 (0.14-0.14)	0.19 (0.18-0.19)	0.14 (0.14-0.15)	NA
pct_mort30	0.03 (0.03-0.03)	0.08 (0.07-0.08)	0.04 (0.03-0.04)	0.07 (0.07-0.08)	NA

Categorical predictor distributions

Major Diagnostic Group

d_eval %>%
  left_join(mdc_df) %>%
  group_by(disch_sabo,mdc_name) %>%
  summarise(n = n()) %>%
  group_by(discharge = ifelse(disch_sabo == 1,"NH","Home")) %>%
  mutate(pct = n/sum(n)) %>%
  ggplot(aes(y=pct,x= mdc_name,fill = discharge)) +
  geom_bar(stat="identity", position = position_dodge()) +
  labs(y= "% of patients with MDC  (raw)") +
  theme(axis.text.x = element_text(angle = 90,hjust = 1,vjust = 0.3))

m_data %>%
  left_join(mdc_df) %>%
  group_by(disch_sabo,mdc_name) %>%
  summarise(n = n()) %>%
  group_by(discharge = ifelse(disch_sabo == 1,"NH","Home")) %>%
  mutate(pct = n/sum(n)) %>%
  ggplot(aes(y=pct,x= mdc_name,fill = discharge)) +
  geom_bar(stat="identity", position = position_dodge()) +
  labs(y= "% of patients with MDC  (raw)") +
  theme(axis.text.x = element_text(angle = 90,hjust = 1,vjust = 0.3))

Hospital Ward

d_eval %>%
  group_by(disch_sabo,mvo_last) %>%
  summarise(n = n()) %>%
  group_by(discharge = ifelse(disch_sabo == 1,"NH","Home")) %>%
  mutate(pct = n/sum(n)) %>%
  ggplot(aes(y=pct,x= as.factor(mvo_last),fill = discharge)) +
  geom_bar(stat="identity", position = position_dodge()) +
  labs(y= "Percent of patients with ward code (raw)") +
  theme(axis.text.x = element_text(angle = 90,hjust = 1,vjust = 0.3))

  labs(x = "Hospital ward code")

## <ggplot2::labels> List of 1
##  $ x: chr "Hospital ward code"

m_data %>%
  group_by(disch_sabo,mvo_last) %>%
  summarise(n = n()) %>%
  group_by(discharge = ifelse(disch_sabo == 1,"NH","Home")) %>%
  mutate(pct = n/sum(n)) %>%
  ggplot(aes(y=pct,x= as.factor(mvo_last),fill = discharge)) +
  geom_bar(stat="identity", position = position_dodge()) +
  labs(y= "Percent of patients with ward code (raw)") +
  theme(axis.text.x = element_text(angle = 90,hjust = 1,vjust = 0.3))

  labs(x = "Hospital ward code")

## <ggplot2::labels> List of 1
##  $ x: chr "Hospital ward code"

These codes are a bit difficult to translate and likely have limited interpretability outside of Sweden anyway… Ward code lists may be found with the National Board of Health and Welfare: https://www.socialstyrelsen.se/globalassets/sharepoint-dokument/dokument-webb/klassifikationer-och-koder/sjukhuskoder-kodlista-verksamhetsomraden-2006.pdf

Region

d_eval %>%
  group_by(disch_sabo,region) %>%
  summarise(n = n()) %>%
  group_by(discharge = ifelse(disch_sabo == 1,"NH","Home")) %>%
  mutate(pct = n/sum(n)) %>%
  ggplot(aes(y=pct,x= as.factor(region),fill = discharge)) +
  geom_bar(stat="identity", position = position_dodge()) +
  labs(y= "Percent of patients in region (raw)") +
  theme(axis.text.x = element_text(angle = 90,hjust = 1,vjust = 0.3)) +
  labs(x = "Region")

m_data %>%
  group_by(disch_sabo,region) %>%
  summarise(n = n()) %>%
  group_by(discharge = ifelse(disch_sabo == 1,"NH","Home")) %>%
  mutate(pct = n/sum(n)) %>%
  ggplot(aes(y=pct,x= as.factor(region),fill = discharge)) +
  geom_bar(stat="identity", position = position_dodge()) +
  labs(y= "Percent of patients in region (raw)") +
  theme(axis.text.x = element_text(angle = 90,hjust = 1,vjust = 0.3)) +
  labs(x = "Region")

Propensity score distribution in raw and matched samples

ggplot(d_eval, aes(x = nh_propensity, 
              fill = as.factor(disch_sabo))) +
  geom_density(alpha = 0.5, colour = "grey50") +
  #geom_point(aes(y= 0,x=nh_propensity,)) +
  geom_rug(aes(color = prop_strat)) +
  scale_color_brewer(palette = "Spectral")

d_eval %>%
  group_by(prop_strat,disch_sabo) %>%
  summarise(n = n()) %>%
  pivot_wider(names_from = disch_sabo,
              values_from = n)

ggplot(m_data, aes(x = nh_propensity, 
              fill = as.factor(disch_sabo))) +
  geom_density(alpha = 0.5, colour = "grey50") +
  #geom_point(aes(y= 0,x=nh_propensity,)) +
  geom_rug(aes(color = prop_strat)) +
  scale_color_brewer(palette = "Spectral")

m_data %>%
  group_by(prop_strat,disch_sabo) %>%
  summarise(n = n()) %>%
  pivot_wider(names_from = disch_sabo,
              values_from = n)

Propensity score variable description

description <- c("caredays" = "Duration of hospital stay",  
                 "diagsec" = "Secondary diagnosis", 
                 "age" = "Patient age", 
                 "region" = "Region", 
                 "muni" = "Municipality", 
                 "mvo" = "Hospital ward", 
                 "date" = "Date of discharge", 
                 "prevSince" = "Days since previous hospital stay", 
                 "countYrUnplanned" = "Number of unplanned hospital stays in the last year", 
                 "countHomecareMonths" = "Number of months of home health care prior to discharge",
                 "countHomeserviceMonths" = "Number of months of home services prior to discharge",
                 "countAmbPlanned" = "Number of planned contacts with ambulatory care in the previous year",
                 "countAmbUnplanned" = "Number of unplanned contacts with ambulatory care in the previous year",
                 "countDiags" = "Number of diagnosis codes recorded during hospital stay",
                 "countInterventions" = "Number of intervention codes recorded during hospital stay",
                 "prevCaredays" = "Duration of previous hostial stay", 
                 "week" = "Week of discharge", 
                 "hosp" = "Discharging hospital", 
                 "civil" = "Patient civil status", 
                 "diagprim" = "Patient primary diagnosis", 
                 "op" = "Operation (KVÅ) code", 
                 "countYr" = "Number of hospital stays in the last year", 
                 "female" = "Sex of patient", 
                 "weekday" = "Weekday of discharge", 
                 "born" = "Patient region of birth", 
                 "prevdiag" = "Primary diagnosis during previous hospital stay")

importance_matrix <- xgb.importance(propensity_xgb) %>%
  mutate(type = str_split_i(Feature,"_",1)) %>%
  left_join(as_tibble(list(type = names(description),
                      description = description)))

t <- xgb.plot.shap.summary(data = sm, model = propensity_xgb,top_n = 100)

mean_shap <- t$data %>%
  mutate(valshap = feature_value*shap_value) %>%
  group_by(Feature = feature) %>%
  summarise(mean_shap = mean(valshap))

importance_matrix <- importance_matrix %>%
  left_join(mean_shap) %>%
  select(description,Feature,Gain,Cover,mean_shap)

vars <- colnames(m_sm)[colnames(m_sm) %in% importance_matrix$Feature]

smd_matched <- m_sm[,vars] %>%
  as.matrix() %>%
  as_tibble() %>%
  bind_cols(select(m_data,disch_sabo)) %>%
  pivot_longer(-disch_sabo, names_to = "variable", values_to = "value") %>%
  group_by(variable) %>%
  summarise(effectsize::cohens_d(value ~ disch_sabo)) %>%
  left_join(importance_matrix,by = c("variable"="Feature")) %>%
  mutate(smd_95ci = paste0(round(Cohens_d,3)," (",round(CI_low,3),":",round(CI_high,3),")")) %>%
  transmute(variable,
            description,
            smd_95ci,
            Gain = round(Gain,5),
            Cover = round(Cover,5),
            mean_shap = round(mean_shap,5),
            Cohens_d) %>%
  arrange(desc(Gain))

Variable summary

importance_matrix %>%
  group_by(description) %>%
  summarise(n_values = n(),
            sum_gain = sum(Gain)) %>%
  arrange(desc(sum_gain)) %>%
  kable()

description	n_values	sum_gain
Number of months of home services prior to discharge	1	0.2063592
Secondary diagnosis	605	0.1898181
Duration of hospital stay	1	0.1226844
Region	21	0.0535806
Number of months of home health care prior to discharge	1	0.0526059
Municipality	201	0.0499560
Operation (KVÅ) code	314	0.0446787
Patient age	1	0.0444714
Hospital ward	35	0.0369663
Duration of previous hostial stay	1	0.0281324
Number of unplanned contacts with ambulatory care in the previous year	1	0.0278809
Date of discharge	1	0.0235106
Discharging hospital	70	0.0224705
Days since previous hospital stay	1	0.0162427
Patient primary diagnosis	195	0.0136386
Week of discharge	1	0.0125138
Number of intervention codes recorded during hospital stay	1	0.0101896
Number of planned contacts with ambulatory care in the previous year	1	0.0075637
Number of diagnosis codes recorded during hospital stay	1	0.0073374
Weekday of discharge	7	0.0066980
Patient civil status	4	0.0066910
Number of unplanned hospital stays in the last year	1	0.0058417
Primary diagnosis during previous hospital stay	89	0.0036634
Sex of patient	1	0.0029272
Number of hospital stays in the last year	1	0.0018558
Patient region of birth	4	0.0017219

Detailed variable table

Note: Higher SHAP values indicate a marginal, linear association with a higher likelihood of discharge to NH. Note that many ICD codes were masked by the providers of the registry data to maintain patient anonymity.

datatable(select(smd_matched,-Cohens_d))

Partial dependence plot

xgb.plot.shap(sm, model = propensity_xgb,top_n = 5)

SMD for all predictors

smd_matched %>%
  ggplot(aes(x=Cohens_d,y=reorder(variable,Gain))) +
  geom_point() +
  theme(
    axis.text.y = element_blank(),
    axis.ticks.y = element_blank() )+
  
  labs(y = "Parameter in propensity score model",
       x = "Standardized Mean Difference")

Main analysis

Estimate models

if(file.exists("./cif.rda") & !reload){
  load("./cif.rda")
  load("./hr.rda")
}else{
  
  cif <- list()
  hr <- list()
  
  cif$mort <- tidycmprsk::cuminc(Surv(ts_mort_days_itt, 
                                      as.factor(ts_mort_event_itt)) ~ disch_sabo, 
                     cluster = lopnr, 
                     d_eval)
  
  cif$mort_prop <- tidycmprsk::cuminc(Surv(ts_mort_days_itt, 
                                           as.factor(ts_mort_event_itt)) ~ disch_sabo,
                     cluster = subclass, 
                     m_data)
  
  hr$mort7 <- coxph(Surv(ts_mort_days_itt_7,
                         ts_mort_event_itt_7) ~ disch_sabo,
                    cluster = lopnr,
                    data = d_eval)
  
  hr$mort30 <- coxph(Surv(ts_mort_days_itt_30,
                         ts_mort_event_itt_30) ~ disch_sabo,
                    cluster = lopnr,
                    data = d_eval)
  
  hr$mort90 <- coxph(Surv(ts_mort_days_itt,
                         ts_mort_event_itt) ~ disch_sabo,
                    cluster = lopnr,
                    data = d_eval)
  
  hr$mort7_prop <- coxph(Surv(ts_mort_days_itt_7,
                         ts_mort_event_itt_7) ~ disch_sabo,
                    cluster = subclass,
                    data = m_data)
  
  hr$mort30_prop <- coxph(Surv(ts_mort_days_itt_30,
                         ts_mort_event_itt_30) ~ disch_sabo,
                    cluster = subclass,
                    data = m_data)
  hr$mort90_prop <- coxph(Surv(ts_mort_days_itt,
                         ts_mort_event_itt) ~ disch_sabo,
                    cluster = subclass,
                    data = m_data)
  
  hr$mort7_dr <- coxph(Surv(ts_mort_days_itt_7,
                         ts_mort_event_itt_7) ~ disch_sabo + 
                         mort7_pred,
                    cluster = subclass,
                    data = m_data)
  
  hr$mort30_dr <- coxph(Surv(ts_mort_days_itt_30,
                         ts_mort_event_itt_30) ~ disch_sabo + mort30_pred,
                    cluster = subclass,
                    data = m_data)
  hr$mort90_dr <- coxph(Surv(ts_mort_days_itt,
                         ts_mort_event_itt) ~ disch_sabo + mort90_pred,
                    cluster = subclass,
                    data = m_data)
  
  ## Readmission
  
  
  cif$readmit <- tidycmprsk::cuminc(Surv(ts_readmit_days_itt, 
                                                     ts_readmit_event_itt) ~ disch_sabo,
                     cluster = lopnr, 
                     d_eval)
  
  cif$readmit_prop <- tidycmprsk::cuminc(Surv(ts_readmit_days_itt, 
                                              ts_readmit_event_itt) ~ disch_sabo,
                     cluster = subclass, 
                     m_data)
  
  # cif$readmit_prop_dr <- tidycmprsk::cuminc(Surv(ts_readmit_days_itt, 
  #                                                         as.numeric(ts_readmit_event_itt)) ~ disch_sabo + readmit30_pred,
  #                    cluster = subclass, 
  #                    m_data)
  
  library(fastcmprsk)
  
  hr$readmit7 <- fastCrr(Crisk(ts_readmit_days_itt_7,ts_readmit_event_itt_7,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo,
                    data = d_eval)
  hr$readmit30 <- fastCrr(Crisk(ts_readmit_days_itt_30,ts_readmit_event_itt_30,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo,
                    data = d_eval)
  hr$readmit90 <- fastCrr(Crisk(ts_readmit_days_itt,ts_readmit_event_itt,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo,
                    data = d_eval)
  
  hr$readmit7_prop <- fastCrr(Crisk(ts_readmit_days_itt_7,ts_readmit_event_itt_7,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo,
                    data = m_data)
  hr$readmit30_prop <- fastCrr(Crisk(ts_readmit_days_itt_30,ts_readmit_event_itt_30,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo,
                    data = m_data)
  hr$readmit90_prop <- fastCrr(Crisk(ts_readmit_days_itt,ts_readmit_event_itt,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo,
                    data = m_data)
  
  hr$readmit7_dr <- fastCrr(Crisk(ts_readmit_days_itt_7,ts_readmit_event_itt_7,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo + readmit7_pred,
                    data = m_data)
  hr$readmit30_dr <- fastCrr(Crisk(ts_readmit_days_itt_30,ts_readmit_event_itt_30,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo + readmit30_pred,
                    data = m_data)
  hr$readmit90_dr <- fastCrr(Crisk(ts_readmit_days_itt,ts_readmit_event_itt,
                         failcode = "Readmission",
                         cencode = "Censored") ~ disch_sabo + readmit90_pred,
                    data = m_data)
  
  
  # Composite outcome
  
    cif$any <- tidycmprsk::cuminc(Surv(ts_any_days_itt, 
                                      as.factor(ts_any_event_itt)) ~ disch_sabo, 
                     cluster = lopnr, 
                     d_eval)
  
  cif$any_prop <- tidycmprsk::cuminc(Surv(ts_any_days_itt, 
                                           as.factor(ts_any_event_itt)) ~ disch_sabo,
                     cluster = subclass, 
                     m_data)
  
  hr$any7 <- coxph(Surv(ts_any_days_itt_7,
                         ts_any_event_itt_7) ~ disch_sabo,
                    cluster = lopnr,
                    data = d_eval)
  
  hr$any30 <- coxph(Surv(ts_any_days_itt_30,
                         ts_any_event_itt_30) ~ disch_sabo,
                    cluster = lopnr,
                    data = d_eval)
  
  hr$any90 <- coxph(Surv(ts_any_days_itt,
                         ts_any_event_itt) ~ disch_sabo,
                    cluster = lopnr,
                    data = d_eval)
  
  hr$any7_prop <- coxph(Surv(ts_any_days_itt_7,
                         ts_any_event_itt_7) ~ disch_sabo,
                    cluster = subclass,
                    data = m_data)
  
  hr$any30_prop <- coxph(Surv(ts_any_days_itt_30,
                         ts_any_event_itt_30) ~ disch_sabo,
                    cluster = subclass,
                    data = m_data)
  hr$any90_prop <- coxph(Surv(ts_any_days_itt,
                         ts_any_event_itt) ~ disch_sabo,
                    cluster = subclass,
                    data = m_data)
  
  hr$any7_dr <- coxph(Surv(ts_any_days_itt_7,
                         ts_any_event_itt_7) ~ disch_sabo + 
                         any7_pred,
                    cluster = subclass,
                    data = m_data)
  
  hr$any30_dr <- coxph(Surv(ts_any_days_itt_30,
                         ts_any_event_itt_30) ~ disch_sabo + any30_pred,
                    cluster = subclass,
                    data = m_data)
  hr$any90_dr <- coxph(Surv(ts_any_days_itt,
                         ts_any_event_itt) ~ disch_sabo + any90_pred,
                    cluster = subclass,
                    data = m_data)
  
  save(cif,file = "./cif.rda")
  save(hr,file = "./hr.rda")
}

Plot cumulative incidence curves

cif$mort %>%
  ggcuminc(outcome = "TRUE") +
  add_confidence_interval() +
  labs(y = "Mortality cumulative incidence (raw)",
       x = "Days since discharge") + 
  scale_color_manual(values = c("red", "blue")) +
  scale_fill_manual(values = c("red", "blue")) +
  geom_vline(xintercept = c(7,30,90))

cif$mort_prop %>%
  ggcuminc(outcome = "TRUE") +
  add_confidence_interval() +
  labs(y = "Mortality cumulative incidence (matched)",
       x = "Days since discharge") + 
  scale_color_manual(values = c("red", "blue")) +
  scale_fill_manual(values = c("red", "blue")) +
  geom_vline(xintercept = c(7,30,90))

cif$readmit %>%
  ggcuminc(outcome = "Readmission") +
  add_confidence_interval() +
  labs(y = "Readmission cumulative incidence (raw)",
       x = "Days since discharge") + 
  scale_color_manual(values = c("red", "blue")) +
  scale_fill_manual(values = c("red", "blue")) +
  geom_vline(xintercept = c(7,30,90))

cif$readmit_prop %>%
  ggcuminc(outcome = "Readmission") +
  add_confidence_interval() +
  labs(y = "Readmission cumulative incidence (matched)",
       x = "Days since discharge") + 
  scale_color_manual(values = c("red", "blue")) +
  scale_fill_manual(values = c("red", "blue")) +
  geom_vline(xintercept = c(7,30,90))

cif$any %>%
  ggcuminc(outcome = "TRUE") +
  add_confidence_interval() +
  labs(y = "Composite cumulative incidence (raw)",
       x = "Days since discharge") + 
  scale_color_manual(values = c("red", "blue")) +
  scale_fill_manual(values = c("red", "blue")) +
  geom_vline(xintercept = c(7,30,90))

cif$any_prop %>%
  ggcuminc(outcome = "TRUE") +
  add_confidence_interval() +
  labs(y = "Composite cumulative incidence (matched)",
       x = "Days since discharge") + 
  scale_color_manual(values = c("red", "blue")) +
  scale_fill_manual(values = c("red", "blue")) +
  geom_vline(xintercept = c(7,30,90))

Hazard ratio table

paste_cox_ci <- function(mod,r=3){
  paste0(round(exp(coef(mod)["disch_sabo"]),r), " (",
        round(exp(confint(mod)["disch_sabo",1]),r),"-",
        round(exp(confint(mod)["disch_sabo",2]),r),")")
}

paste_crr_ci <- function(mod,r=3){
  
  ci <- confint(mod)
  if(first(class(ci)) == "matrix"){
    ci <- ci[1,]
  }

  paste0(round(exp(coef(mod)[1]),r), " (",
        round(exp(ci[1]),r),"-",
        round(exp(ci[2]),r),")")
}

if(file.exists("./outcome_table.rda") & !reload){
  load("./outcome_table.rda")
}else{

raw <- data.frame("raw" = c(paste_cox_ci(hr$mort7),
                            paste_cox_ci(hr$mort30),
                            paste_cox_ci(hr$mort90),
                            paste_crr_ci(hr$readmit7),
                            paste_crr_ci(hr$readmit30),
                            paste_crr_ci(hr$readmit90),
                            paste_crr_ci(hr$any7),
                            paste_crr_ci(hr$any30),
                            paste_crr_ci(hr$any90)))

matched <- data.frame("matched" = c(paste_cox_ci(hr$mort7_prop),
                            paste_cox_ci(hr$mort30_prop),
                            paste_cox_ci(hr$mort90_prop),
                            paste_crr_ci(hr$readmit7_prop),
                            paste_crr_ci(hr$readmit30_prop),
                            paste_crr_ci(hr$readmit90_prop),
                            paste_crr_ci(hr$any7_prop),
                            paste_crr_ci(hr$any30_prop),
                            paste_crr_ci(hr$any90_prop)))

doublerobust <- data.frame("doublerobust" = c(paste_cox_ci(hr$mort7_dr),
                            paste_cox_ci(hr$mort30_dr),
                            paste_cox_ci(hr$mort90_dr),
                            paste_crr_ci(hr$readmit7_dr),
                            paste_crr_ci(hr$readmit30_dr),
                            paste_crr_ci(hr$readmit90_dr),
                            paste_crr_ci(hr$any7_dr),
                            paste_crr_ci(hr$any30_dr),
                            paste_crr_ci(hr$any90_dr)))

outcome_table <- data.frame("outcome" = c(rep("Mortality",3),
                                          rep("Readmission",3),
                                          rep("Composite",3)),
                            "Time" = c(rep(c("7","30","90"),3))) %>%
  bind_cols(raw) %>%
  bind_cols(matched) %>%
  bind_cols(doublerobust)

  save(outcome_table,file = "./outcome_table.rda")

}

kable(outcome_table)

outcome	Time	raw	matched	doublerobust
Mortality	7	2.311 (2.092-2.553)	1.84 (1.549-2.185)	1.786 (1.502-2.124)
Mortality	30	2.384 (2.29-2.483)	2.007 (1.871-2.153)	1.979 (1.842-2.126)
Mortality	90	2.043 (1.992-2.094)	1.899 (1.819-1.982)	1.936 (1.851-2.023)
Readmission	7	0.574 (0.548-0.601)	0.671 (0.631-0.714)	0.672 (0.632-0.716)
Readmission	30	0.653 (0.635-0.672)	0.742 (0.717-0.768)	0.732 (0.707-0.757)
Readmission	90	0.708 (0.694-0.722)	0.81 (0.786-0.835)	0.794 (0.77-0.818)
Composite	7	0.68 (0.653-0.707)	0.77 (0.725-0.818)	0.772 (0.727-0.82)
Composite	30	0.869 (0.85-0.888)	0.944 (0.912-0.977)	0.928 (0.897-0.961)
Composite	90	0.976 (0.961-0.991)	1.069 (1.042-1.096)	1.051 (1.024-1.079)