ALEDDDDD

/*1 : Import des données*/
proc contents data=source.demography short;
proc contents data=source.physical_exam short;
proc contents data=source.vital_signs short;
proc contents data=source.treatment_assignment short;
run;
/*------------------------------------------------------2.1----------------------------------------------------------*/

/* 2.1.1 : Age Today*/
DATA source.DM;
    TITLE "Demography w/ current age";
    SET source.demography;
    ATTRIB AGE_TODAY LABEL="Age today" LENGTH=3;
    AGE_TODAY=year(today()) - birthdtyy;
PROC PRINT DATA=source.dm LABEL NOOBS;
run;

/* 2.1.2 : Fusion DM et Treatment_assignment*/
DATA source.DM;
TITLE "Demography merged w/ treatment";
MERGE source.dm source.treatment_assignment;
BY USUBJID;
RUN;

PROC PRINT DATA=source.DM LABEL NOOBS;
run;

/* 2.1.3 : Age moyen max et min pour chaque groupe de traitement*/
PROC MEANS DATA=source.DM VAR MAX MIN NONOBS
MAXDEC=2;
VAR age;
CLASS trtcd;
;
title "Age statistics by treatment grp";
/*-----------------------------------------------2.2------------------------------------------------------------*/

/*2.2.1*/
DATA source.TAILLE;
SET SOURCE.PHYSICAL_EXAM;
WHERE HGT is not null;
KEEP USUBJID HGT;
RUN;
/*On récupere le poids et l'id de la visite correspondante*/
DATA source.POIDS;
SET SOURCE.PHYSICAL_EXAM;
WHERE WGT is not null;
KEEP USUBJID WGT WGTU VISID;
RUN;
/*Récupération des dates des visites*/
DATA source.DATE;
SET SOURCE.DATE_OF_VISIT;
KEEP USUBJID VISID VISDT;
RUN;
/*Tri par USUBJID pour la fusion des 3 tables*/
PROC SORT DATA=source.TAILLE;
BY USUBJID;
RUN;
PROC SORT DATA=source.POIDS;
BY USUBJID VISID;
RUN;
PROC SORT DATA=source.DATE;
BY USUBJID VISID;
RUN;
/*Fusion du poids et de la date de visite*/
DATA source.imc;
MERGE source.POIDS (IN=mark1) source.DATE (IN=mark2);
BY USUBJID VISID;
if mark1 and mark2 then output;
RUN;
/*Fusion de la taille et de la table imc*/
DATA source.imc;
MERGE source.TAILLE source.imc;
BY USUBJID;
RUN;
/*Tri pour avoir la dernière visite en premier*/
PROC SORT DATA=source.imc;
BY USUBJID DESCENDING VISDT;
RUN;
/*Supprime les doublons pour garder seulement la dernière visite de chaque individu*/
PROC SORT DATA=source.imc nodupkey;
BY USUBJID;
RUN;
/* creer un nouveau dataset ou il y a l'imc calcule*/
DATA source.CalculImc;
SET source.imc;
ATTRIB imc
        LABEL="IMC"
        LENGTH=3
        ;
        imc= (WGT/((HGT/100)*(HGT/100)));
RUN;

/*2.2.2: Merge PE */
DATA source.Physicalvisit;
MERGE source.physical_exam (IN=mark1) source.DATE (IN=mark2);
BY USUBJID VISID;
If mark1 and mark2 then output;
RUN;
/*fusion de physicalvisit avec dm pour avoir le dataset PE*/
DATA source.PE;
MERGE source.physicalvisit source.dm;
BY USUBJID;
RUN;
PROC PRINT DATA=source.PE;

/*-----------------------------------------------2.3------------------------------------------------------------*/

/*2.3.1 : Tension et 2.3.2 : Diagnostic*/
DATA source.Tension;
TITLE "Tension";
SET source.vital_signs;
ATTRIB tension label="Diagnostique tension" LENGTH=$40;
IF SYS > 180 OR DIA > 110 THEN
tension = "Hypertension sévère";
ELSE IF SYS > 159 OR DIA > 99 THEN
tension = "Hypertension modérée";
ELSE IF SYS > 139 OR DIA > 89 THEN
tension = "Hypertension légère";
ELSE IF SYS > 129 OR DIA > 84 THEN
tension = "Tension normale élevée";
ELSE IF SYS > 119 OR DIA > 79 THEN
tension = "Tension normale";
ELSE tension = "Tension optimale";
PROC PRINT DATA=source.Tension LABEL NOOBS;
run;

/*2.3.3 : Fusion avec Date_of_visit*/
DATA source.VS;
TITLE "Tension merged w/ Date_of_visit";
MERGE source.Tension source.Date_of_visit;
BY USUBJID;
IF SYS=. OR DIA =. THEN tension = "Non calculé";
run;
PROC PRINT DATA=source.Tension LABEL NOOBS;
run;

/*2.3.4 : Fusion VS et PE*/
PROC SORT data=source.PE;
BY USUBJID;
run;
PROC SORT data=source.VS;
BY USUBJID;
run;
DATA source.VS_PE;
TITLE "VS merged w/ PE";
MERGE source.VS source.PE;
BY USUBJID;
run;
PROC PRINT DATA=source.VS_PE LABEL NOOBS;
run;


/*2.3.5 : Fréquence paramètres*/
PROC SORT DATA=source.vs_pe OUT = source.vs_pe;
BY TRTCD;
run;
PROC FREQ DATA=source.vs_pe ORDER=freq;
TABLES SYS*TEMP / LIST NOCUM OUT=source.sys_avg_by_trtmntgrp;
BY TRTCD;
run;


/*2.3.6 : Égalité moyenne tension systolique */
PROC ANOVA DATA=source.vs_pe;
CLASS TRTCD;
MODEL SYS = TRTCD;
run;

PROC FREQ DATA=source.vs_pe;
TABLE SYS / chisq;
run;