HDPS codes data prep0301

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\id_list_0119.csv"
out=id_list  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;
data id_list1;
    set id_list; 
    array CHAR _character_ ; 
do over CHAR; 
    if CHAR="NA" then call missing(CHAR);
    else if missing(CHAR) then CHAR=" ";
end;
run ;

DATA id_list2;
	SET id_list1;
	array olddates $ M2_sputum_date M3_sputum_date M5_sputum_date Rx_date Rx_end_date_x
					OPT_min_date OPT_max_date min_disdate  max_disdate DILI_pre_date DILI_post_date
					ALT_max_date  ALT_abn_post_date ALT_2N_post_date ALT_5N_post_date ALT_10N_post_date
					ALP_max_date  ALP_abn_post_date ALP_2N_post_date ALP_5N_post_date ALP_10N_post_date
					Tbil_max_date Tbil_abn_post_date Tbil_2N_post_date Tbil_5N_post_date Tbil_10N_post_date
 					INR_max_date INR_abn_post_date INR_2N_post_date INR_3N_post_date INR_45N_post_date 
					PTA_min_date PTA_40_post_date
					Ascites_pre_dx_date 		Ascites_post_dx_date
					Hepaence_pre_dx_date 		Hepaence_post_dx_date
					Cirrhosis_pre_dx_date 		Cirrhosis_post_dx_date
					Fattyliver_pre_dx_date 		Fattyliver_post_dx_date
					HBV_pre_dx_date   			HBV_post_dx_date 
					HCV_pre_dx_date 			HCV_post_dx_date 
					AutoHepatitis_pre_dx_date 	AutoHepatitis_post_dx_date
					ALT_max_baseline_date	ALT_abn_pre_date
					ALP_max_baseline_date	ALP_abn_pre_date
					Tbil_max_baseline_date 	Tbil_abn_pre_date
					Death_date
					last_fol_up min_admdate max_admdate;
	array newdates dt1-dt57;
		do i = 1 to dim(olddates);
			newdates(i) = input(strip(olddates(i)),yymmdd10.);
		end;
		drop i;
		format dt1-dt57 yymmdd10.;
		DROP M2_sputum_date M3_sputum_date M5_sputum_date Rx_date Rx_end_date_x
					OPT_min_date OPT_max_date min_disdate  max_disdate DILI_pre_date DILI_post_date
					ALT_max_date  ALT_abn_post_date ALT_2N_post_date ALT_5N_post_date ALT_10N_post_date
					ALP_max_date  ALP_abn_post_date ALP_2N_post_date ALP_5N_post_date ALP_10N_post_date
					Tbil_max_date Tbil_abn_post_date Tbil_2N_post_date Tbil_5N_post_date Tbil_10N_post_date
 					INR_max_date INR_abn_post_date INR_2N_post_date INR_3N_post_date INR_45N_post_date 
					PTA_min_date PTA_40_post_date
					Ascites_pre_dx_date 		Ascites_post_dx_date
					Hepaence_pre_dx_date 		Hepaence_post_dx_date
					Cirrhosis_pre_dx_date 		Cirrhosis_post_dx_date
					Fattyliver_pre_dx_date 		Fattyliver_post_dx_date
					HBV_pre_dx_date   			HBV_post_dx_date 
					HCV_pre_dx_date 			HCV_post_dx_date 
					AutoHepatitis_pre_dx_date 	AutoHepatitis_post_dx_date
					ALT_max_baseline_date	ALT_abn_pre_date
					ALP_max_baseline_date	ALP_abn_pre_date
					Tbil_max_baseline_date 	Tbil_abn_pre_date
					Death_date
					last_fol_up min_admdate max_admdate;
RUN;
DATA id_list3;
	SET id_list2;
	RENAME dt1=M2_sputum_date;  RENAME dt2=M3_sputum_date;RENAME dt3=M5_sputum_date;RENAME dt4=Rx_date;RENAME dt5=Rx_end_date_x;
		  RENAME  dt6=OPT_min_date;  RENAME dt7=OPT_max_date;  RENAME dt8=min_disdate;  RENAME dt9=max_disdate;  RENAME dt10=DILI_pre_date;  RENAME dt11=DILI_post_date;
		  RENAME  dt12=ALT_max_date;  RENAME dt13=ALT_abn_post_date;  RENAME dt14=ALT_2N_post_date;  RENAME dt15=ALT_5N_post_date;  RENAME dt16=ALT_10N_post_date;
		  RENAME  dt17=ALP_max_date;  RENAME dt18=ALP_abn_post_date;  RENAME dt19=ALP_2N_post_date;  RENAME dt20=ALP_5N_post_date;  RENAME dt21=ALP_10N_post_date;
		  RENAME  dt22=Tbil_max_date;  RENAME dt23=Tbil_abn_post_date;  RENAME dt24=Tbil_2N_post_date;  RENAME dt25=Tbil_5N_post_date;  RENAME dt26=Tbil_10N_post_date;
 		  RENAME  dt27=INR_max_date;  RENAME dt28=INR_abn_post_date;  RENAME dt29=INR_2N_post_date;  RENAME dt30=INR_3N_post_date;  RENAME dt31=INR_45N_post_date;
		  RENAME  dt32=PTA_min_date;  RENAME dt33=PTA_40_post_date;  
			RENAME dt34=Ascites_pre_dx_date;  RENAME dt35=Ascites_post_dx_date
			;  RENAME dt36=Hepaence_pre_dx_date;  RENAME dt37=Hepaence_post_dx_date
			;  RENAME dt38=Cirrhosis_pre_dx_date;  RENAME dt39=Cirrhosis_post_dx_date
			;  RENAME dt40=Fattyliver_pre_dx_date;  RENAME dt41=Fattyliver_post_dx_date
			;  RENAME dt42=HBV_pre_dx_date;  RENAME dt43=HBV_post_dx_date 
			;  RENAME dt44=HCV_pre_dx_date;  RENAME dt45=HCV_post_dx_date 
			;  RENAME dt46=AutoHepatitis_pre_dx_date;  RENAME dt47=AutoHepatitis_post_dx_date
			;  RENAME dt48=ALT_max_baseline_date;  RENAME dt49=ALT_abn_pre_date
			;  RENAME dt50=ALP_max_baseline_date;  RENAME dt51=ALP_abn_pre_date
			;  RENAME dt52=Tbil_max_baseline_date;  RENAME dt53=Tbil_abn_pre_date
			;  RENAME dt54=Death_date
			;  RENAME dt55=last_fol_up;
			RENAME dt56=min_admdate;RENAME dt57= max_admdate;
RUN;

PROC CONTENTS DATA=id_list1 VARNUM;RUN;
PROC FREQ DATA=id_list3;
	TABLE Rx_outcome;
RUN;

DATA id_list4;
	SET id_list3;
	last_fol_up=max(Rx_end_date_x, OPT_max_date, max_admdate, max_disdate, Death_date);
	IF last_fol_up>Dx_date;
	IF age_cal>=18;
	IF Two_encounter_baseline=1;
	IF Rx_outcome="Dx_changed" THEN DELETE;
	Death_fol_up=min(last_fol_up, Death_date,22096,Dx_date+730)-Dx_date;
	Death_status=0;
	IF Death_fol_up=Death_date-Dx_date AND Death_date ^=. THEN Death_status=1;
	IF Rx_success=. THEN Rx_success=1;

	DILI_status=0;
	DILI_fol_up=min(DILI_post_date,last_fol_up, Rx_end_date_x, Death_date,22096,Dx_date+730)-Dx_date;
	IF DILI_post_date ^=. AND DILI_fol_up=DILI_post_date-Dx_date THEN DILI_status=1;
	IF Death_date ^=. AND DILI_post_date =. THEN DILI_status=2;
	
	DILI_final=0;
	IF DILI_status=1 THEN DILI_final=1;

	IF Co_Yinpian_ini_30=. THEN Co_Yinpian_ini_30=0;
	IF Co_TCM_ini_30=. THEN Co_TCM_ini_30=0;

	*IF baseline_hepato=1 THEN DELETE;
	*IF DILI_pre_type = " ";

	lab_abn_baseline=0;
	IF ALT_max_ratio_baseline ^=" " AND ALT_max_ratio_baseline>1 THEN lab_abn_baseline=1;
	IF ALP_max_ratio_baseline ^=" " AND ALP_max_ratio_baseline>1 THEN lab_abn_baseline=1;
	IF Tbil_max_ratio_baseline ^=" " AND Tbil_max_ratio_baseline>1 THEN lab_abn_baseline=1;

	IF Pregnancy_baseline=1 THEN DELETE;

	*IF lab_abn_baseline=0;

	*IF Rx_date ^=. ;
	*IF Rx_date-Dx_date<=7;
RUN;

PROC FREQ DATA=id_list4;
	TABLE DILI_status DILI_final;
RUN;

DATA id_list;
SET id_list4;
KEEP idcard Dx_date DILI_final baseline_hepato ini_hepato index_YEAR;
IF DILI_final^=1 THEN DILI_final=0;
RENAME DILI_final=outcome;
RENAME ini_hepato=exposure;
index_YEAR=YEAR(DX_date);
RUN;

LIBNAME workingD "E:\Ningbo_Projects\TB\TB project\Workingd\CSV files";
proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\IPT_dx_all_0131.csv"
out=IPT_dx_all  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\OPT_dx_all_0131.csv"
out=OPT_dx_all  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

DATA ipt_dx;
	SET IPT_dx_all;
	IF adm_time-365<=Dx_date<adm_time-30;
	IF Dx_code in ("@DIAGNOSIS_ICD@"," ","-","1","--") THEN DELETE;
	KEEP idcard Dx_code;
DATA opt_dx;
	SET OPT_dx_all;
	IF clinic_time-365<=Dx_date<clinic_time-30;
	IF Dx_code in ("@DIAGNOSIS_ICD@"," ","-","1","--","健康查体") THEN DELETE;
	KEEP idcard Dx_code;
RUN;

PROC sql;
	CREATE table ipt_dx as
	SELECT a.idcard,
		a.Dx_code
	FROM ipt_dx a
	INNER JOIN id_list4 b
	ON a.idcard = b.idcard;
QUIT;

PROC sql;
	CREATE table OPT_dx_all as
	SELECT a.idcard,
		a.Dx_code
	FROM OPT_dx_all a
	INNER JOIN id_list4 b
	ON a.idcard = b.idcard;
QUIT;


PROC FREQ DATA=id_list4 ORDER=FREQ;
	TABLE DILI_final;
RUN;

*Inclusion of medication domains;
LIBNAME workingD "E:\Ningbo_Projects\TB\TB project\Workingd\CSV files";
proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\IPT_pharm_Tx_0131.csv"
out=IPT_pharm_all  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\OPT_pharm_Tx_0131.csv"
out=OPT_pharm_all  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

PROC sql;
	CREATE table IPT_pharm as
	SELECT a.idcard,
		a.item_name
	FROM IPT_pharm_all a
	INNER JOIN id_list4 b
	ON a.idcard = b.idcard AND a.intake_start_time<b.Dx_date-30 AND a.intake_start_time>=b.Dx_date-365;
QUIT;

PROC sql;
	CREATE table OPT_pharm as
	SELECT a.idcard,
		a.item_name
	FROM OPT_pharm_all a
	INNER JOIN id_list4 b
	ON a.idcard = b.idcard AND a.clinic_time<b.Dx_date-30 AND a.clinic_time>=b.Dx_date-365;
QUIT;

PROC FREQ DATA=id_list4 ORDER=FREQ;
	TABLE DILI_final ini_hepato;
RUN;


%include "E:/Ningbo_Projects/TB/toolbox/sas/hdmacros.sas";
Title1 'High-dimensional propensity score adjustment';
Title2 '(study description)';
%InitHDMacros (
var_patient_id = idcard,
var_exposure = ini_hepato,
var_outcome = DILI_final,
vars_demographic = Age_cal Sex_x,
vars_force_categorical = index_YEAR,
vars_ignore = none,
top_n = 200,
k = 500,
path_temp_dir = %QUOTE(E:/Ningbo_Projects/TB/),
/* remember, no spaces in path_jar_file! */
path_jar_file = %QUOTE(E:/Ningbo_Projects/TB/toolbox/java/pharmacoepi.jar),
outcome_model_deciles = 0,
outcome_model_matched = 1,
input_cohort = Id_list4,
input_dim1 = opt_dx Dx_code,
input_dim2 = ipt_dx Dx_code,
input_dim3 = IPT_pharm item_name,
input_dim4 = OPT_pharm item_name,

output_cohort = scored_cohort,
result_estimates = estimates,
result_diagnostic = result_diagnostic
);
%DoHDVariableSelection;
%EstimateHDPS(dataset_with_ps);
%RunOutcomeModels(dataset_with_ps, ps);

PROC univariate DATA=Result_diagnostic_all_vars;
	CLASS selected_for_ps;
	VAR outcome_assoc_ranking_var;
	histogram;
RUN;

PROC FREQ DATA=Result_diagnostic_all_vars;
	TABLE selected_for_ps;
RUN;

PROC SORT DATA=Result_diagnostic_all_vars;
	BY outcome_assoc_ranking_var;
RUN;

DATA Dataset_with_ps;
	SET Dataset_with_ps;
	IF Dx_date=. THEN DELETE;
RUN;

DATA workingD.Dataset_with_ps_wodx;
	SET Dataset_with_ps;
DATA workingD.Result_diagnostic_all_vars_wodx;
	SET Result_diagnostic_all_vars;
DATA workingD.Result_diagnostic_dim_codes_wodx;
	SET Result_diagnostic_dim_codes;
RUN;

DATA temp;
	SET Result_diagnostic_all_vars;
	IF selected_for_ps= "true";
RUN;