Skip to content
Snippets Groups Projects
Commit 22b1a583 authored by Vandewalle Gilles's avatar Vandewalle Gilles
Browse files

Upload New File

parent 6887ea40
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
Statistical_scripts/NBACK_Hypo_final.sas 0 → 100644
+ 320
− 0
View file @ 22b1a583
** NBACK 5 AREAS **;
Data NBack;
infile "C:\Users\gilles\Documents\EXPERIMENTS\gv19_HIGHLIGHT\results\Hypo 2023-2024\Hypo\Task_Mel_betas\Nback_hypo_Analysis_5areas_final.csv" dlm=',' firstobs=2;
input Subj Task Hypo Beta_NBack SDBeta_NBack Beta_EDI SDBeta_EDI;
run;
* data need to be actively sorted before they can used - even if they appear rightly sorted already;
Proc sort data=NBack; by task; run;
proc standard data=NBack out=NBack_Z mean=0 std=1; var Beta_NBack Beta_EDI; by task;
run;
data NBack_Z; set NBack_Z;
if abs(Beta_NBack) < 3 then Beta_NBack = '.';
if abs(Beta_EDI) < 3 then Beta_EDI = '.';
run;
Proc sort data=NBack_Z out=NBack_Z; by subj; run;
data NBack; set NBack;
if hypo = 1 then size = 49;
if hypo = 2 then size = 69;
if hypo = 3 then size = 283;
if hypo = 4 then size = 368;
if hypo = 5 then size = 284;
run;
Proc sort data = NBack; by subj; run;
Proc means data=NBack;
var age BMI;
output out = NBack_mean mean = age BMI stderr = SE_Age SE_BMI; run;
Proc univariate data = NBack ; var SDBeta_EDI;
histogram / midpoints= -0.02 to 0.02 by 0.001
lognormal
exponential
weibull
gamma
normal
beta
vaxis = axis1
name = 'MyHist';
inset n mean(5.3) std='Std Dev'(5.3) skewness(5.3)
/ pos = ne header = 'Summary Statistics';
axis1 label=(a=90 r=0);
run;
ODS GRAPHICS ON ;
PROC GLIMMIX data= NBack PLOTS=all;
CLASS Subj hypo task;
MODEL Beta_NBack = hypo task age sex bmi/ dist=normal DDFM = KR solution;
random intercept / subject=Subj solution;* you can comment this line if model does not work;
random hypo /type=ar(1) subject= subj*task;*to include repeated measures;
lsmeans hypo / adjust=tukey;
lsmeans hypo*emo /slicediff=emo adjust=tukey;
lsmeans emo / adjust=tukey;
lsmeans hypo*emo /slicediff=hypo adjust=tukey;
run;
ODS GRAPHICS OFF ;
ODS GRAPHICS ON ;
PROC GLIMMIX data= NBack PLOTS=all;
CLASS Subj hypo task;
MODEL Beta_EDI = hypo task age sex bmi/ dist=normal DDFM = KR solution;
random intercept / subject=Subj solution;* you can comment this line if model does not work;
random hypo /type=ar(1) subject= subj*task;*to include repeated measures;
lsmeans hypo / adjust=tukey;
lsmeans hypo*emo /slicediff=emo adjust=tukey;
lsmeans emo / adjust=tukey;
lsmeans hypo*emo /slicediff=hypo adjust=tukey;
run;
ODS GRAPHICS OFF ;
ODS GRAPHICS ON ; TITLE "GLM for cook's distance";* you can assess Cook's distance if data are normal;
PROC GLM DATA = Emo PLOTS=diagnostics(label);
CLASS Subj sex hypo emo;
MODEL Beta_EDI = hypo|emo age sex bmi/ solution;
RUN;
ODS GRAPHICS OFF;
proc gplot data=NBack; TITLE 'light x emo';
axis1 order=0 to 400 by 50; axis2 order=0 to 0.3 by 0.1;
plot SDBeta_EDI*size/ haxis=axis1 vaxis=axis2;
symbol1 i=std14jmt mode =include w=0.2 c=red v=none l=0.1;run;quit;run;
proc gplot data=NBack; TITLE 'light x emo';
axis1 order=0 to 6 by 1; axis2 order=-0.002 to 0.002 by 0.001;
plot Beta_EDI*hypo=task/ haxis=axis1 vaxis=axis2;
symbol1 i=std14jmt mode =include w=0.2 c=red v=none l=0.1;
symbol1 i=std14jmt mode =include w=0.2 c=blue v=none l=0.1;run;quit;run;
** NBACK 5 LIGHT ALL **;
Data NB_ALL;
infile "C:\Users\gilles\Documents\EXPERIMENTS\gv19_HIGHLIGHT\results\Islay 2023\Hypo\Light_level_betas\NBack_Hypo_5Con_analysis_ALL.csv" dlm=',' firstobs=2;
input Subj Age Sex BMI task hypo Beta_nback light;
if light = 0.2 then light = 0.16;
run;
Proc sort data=NB_ALL; by hypo task; run;
proc standard data=NB_ALL out=NB_ALL_Z mean=0 std=1; var Beta_nback; by hypo task; run;
data NB_ALL_Z; set NB_ALL_Z;
if abs(Beta_nback) < 3 then Beta_nback = '.';
run;
Proc sort data=NB_ALL_Z; by hypo subj task light; run;
Proc univariate data = NB_ALL ; var Beta_nback;
histogram / midpoints= -1 to 1 by 0.1
lognormal
exponential
weibull
gamma
normal
beta
vaxis = axis1
name = 'MyHist';
inset n mean(5.3) std='Std Dev'(5.3) skewness(5.3)
/ pos = ne header = 'Summary Statistics';
axis1 label=(a=90 r=0);
run;
Proc sort data=NB_ALL; by hypo task; run;
proc standard data=NB_ALL out=NB_ALL_Z mean=0 std=1; var Beta_nback; by hypo task; run;
data NB_ALL_Z; set NB_ALL_Z;
if abs(Beta_nback) < 3 then Beta_nback = '.';
run;
Proc sort data=NB_ALL; by hypo subj task light; run;
data NB_ALL; set NB_ALL;
if subj = 24 then delete;
if subj = 76 then delete;
if subj > 84 then delete;
if Beta_nback < -2.1 then Beta_nback = '.';
run;
Proc sort data=NB_ALL; by subj; run;
ODS GRAPHICS ON ;
PROC GLIMMIX data= NB_ALL PLOTS=all;
CLASS subj sex task light hypo;
MODEL Beta_nback = task light|hypo age sex bmi/ dist=normal DDFM = KR solution;
random intercept /type=un subject=Subj solution;* you can comment this line if model does not work;
random light /type=cs subject=subj*hypo*task;*to include repeated measures;
*lsmeans light / adjust=tukey;
*lsmeans hypo / adjust=tukey;
*lsmeans light /slicediff=hypo adjust=tukey;
*lsmeans task / adjust=tukey;
*lsmeans light*hypo /slicediff=hypo adjust=tukey;
*lsmeans hypo*light /slicediff=light adjust=tukey;
*estimate 'light, hypo a' light 1 1 1 1 1 hypo 1 0 0 0 0 light*hypo 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0;
run;
ODS GRAPHICS OFF ;
ODS GRAPHICS ON ;
PROC GLIMMIX data= NB_ALL PLOTS=all;
CLASS subj sex task light;
MODEL Beta_nback = task light age sex bmi/ dist=normal DDFM = KR solution;
random intercept / type=un subject=Subj solution;* you can comment this line if model does not work;
random _residual_ /type=ar(1) subject= light(task);*to include repeated measures;
*lsmeans light / adjust=tukey;
*lsmeans light /slicediff=hypo adjust=tukey;
*lsmeans task / adjust=tukey;
*lsmeans light*hypo /slicediff=hypo adjust=tukey;
*lsmeans hypo*light /slicediff=light adjust=tukey;
run;
ODS GRAPHICS OFF ;
proc sort data=NB_ALL; by hypo light emo;run;
proc gplot data=NB_ALL; TITLE 'light x emo';
axis1 order=-10 to 200 by 30; axis2 order=-0.25 to 0.25 by 0.05;
plot Beta_nback*light=hypo/ haxis=axis1 vaxis=axis2;
symbol1 i=std14jmt mode =include w=0.2 c=blue v=none l=0.1;
symbol2 i=std14jmt mode =include w=0.2 c=red v=none l=0.1;
symbol3 i=std14jmt mode =include w=0.2 c=green v=none l=0.1;
symbol4 i=std14jmt mode =include w=0.2 c=cyan v=none l=0.1;
symbol5 i=std14jmt mode =include w=0.2 c=black v=none l=0.1;run;quit;run;
proc sort data=NB_ALL; by light emo;run;
proc gplot data=NB_ALL; TITLE 'light x emo';
axis1 order=-10 to 200 by 30; axis2 order=-1 to 1.5 by 0.5;
plot Beta_nback*light/ haxis=axis1 vaxis=axis2;
symbol1 i=std14jmt mode =include w=0.2 c=red v=none l=0.1;run;quit;run;
ODS GRAPHICS ON ; TITLE "test";
PROC GLM DATA = NB_ALL PLOTS=all ;
CLASS subj sex task light hypo;
MODEL Beta_nback = task light|hypo age sex bmi;
random subj;
repeated light(task);
OUTPUT OUT = stat P=pred R=Residual RSTUDENT=r1 DFFITS=dffits COOKD=cookd H=hatvalue PRESS=res_del; RUN;
ODS GRAPHICS OFF ;
PROC CORR pearson spearman data=NB_ALL plots=scatter (alpha=.20 .30); var theta_REM diff_cube_odd; run; *you can put as many variables as you want;
ODS GRAPHICS OFF ;
** including performace;
***********************;
Data Perf_NB;
infile "C:\Users\gilles\Documents\EXPERIMENTS\gv19_HIGHLIGHT\results\Islay 2023\Hypo\Performance\Hypo_5Con_Analysis_NBACK_average_38_RTs.csv" dlm=',' firstobs=2;
input Subj Age Sex BMI task Light Betas_Nback Accuracy RT_Hits RTs;
keep Subj task Light Accuracy RT_Hits RTs;
if subj = 59 then delete;
if subj = 64 then delete;
if accuracy < 0.6 then delete;
run;
Proc sort data=Perf_NB; by task light ; run;
proc standard data=Perf_NB out=Perf_NB_Z mean=0 std=1; var Accuracy RT_Hits RTs; by task; run;
data Perf_NB_Z; set Perf_NB_Z;
if abs(Accuracy) < 3 then Accuracy = '.';
if abs(RT_Hits) < 3 then RT_Hits = '.';
if abs(RTs) < 3 then RTs = '.';
run;
Proc sort data=Perf_NB_Z; by subj; run;
Proc sort data=NB_ALL; by Subj light task; run;
Proc sort data=Perf_NB; by Subj light task; run;
data NB_ALL_perf; merge NB_ALL Perf_NB; by Subj light task; run;
Proc sort data=NB_ALL_Z; by Subj light task; run;
Proc sort data=Perf_NB_Z; by Subj light task; run;
data NB_ALL_perf_Z; merge NB_ALL_Z Perf_NB_Z; by Subj light task; run;
data hypo; set NB_ALL_perf;
if hypo NE 5 then delete;
if task = 0 then delete;
if accuracy < 0.6 then delete; * subj 30 is low at times;
if subj = 65 & light = 0.16 & task = 1 then accuracy = '.';
run;
data hypo_Z; set NB_ALL_perf_Z;
if hypo NE 1 then delete;
if task = 1 then delete;
if subj = 59 then delete;
run;
Proc univariate data = hypo ; var RT_Hits;
histogram / midpoints= 800 to 1600 by 50
lognormal
exponential
weibull
gamma
normal
beta
vaxis = axis1
name = 'MyHist';
inset n mean(5.3) std='Std Dev'(5.3) skewness(5.3)
/ pos = ne header = 'Summary Statistics';
axis1 label=(a=90 r=0);
run;
Proc univariate data = hypo ; var accuracy;
histogram / midpoints= 0 to 1 by 0.1
lognormal
exponential
weibull
gamma
normal
beta
vaxis = axis1
name = 'MyHist';
inset n mean(5.3) std='Std Dev'(5.3) skewness(5.3)
/ pos = ne header = 'Summary Statistics';
axis1 label=(a=90 r=0);
run;
ODS GRAPHICS ON ;
PROC GLIMMIX data= hypo PLOTS=all;
CLASS subj sex light;
MODEL accuracy = beta_nback light age sex bmi/ dist=normal DDFM = KR solution;
random intercept / type=un subject=Subj solution;* you can comment this line if model does not work;
random light /type=ar(1) subject= subj*light;*to include repeated measures;
*random light /type=ar(1) subject= subj*light;*to include repeated measures;
*lsmeans light / adjust=tukey;
*lsmeans light /slicediff=hypo adjust=tukey;
*lsmeans emo / adjust=tukey;
*lsmeans light*hypo /slicediff=hypo adjust=tukey;
*lsmeans light*hypo /slicediff=light adjust=tukey;
run;
ODS GRAPHICS OFF ;
ODS GRAPHICS ON ;
PROC GLIMMIX data= hypo PLOTS=all;
CLASS subj sex light;
MODEL RT_hits = beta_nback light age sex bmi/ dist=normal DDFM = KR solution;
random intercept / type=un subject=Subj solution;* you can comment this line if model does not work;
*random light /type=ar(1) subject= subj*light;*to include repeated measures;
random _residual_ /type=ar(1) subject= light;*to include repeated measures;
*lsmeans light / adjust=tukey;
*lsmeans light /slicediff=hypo adjust=tukey;
*lsmeans emo / adjust=tukey;
*lsmeans light*hypo /slicediff=hypo adjust=tukey;
*lsmeans light*hypo /slicediff=light adjust=tukey;
run;
ODS GRAPHICS OFF ;
proc sort data=hypo; by task light subj;run;
proc gplot data=hypo; TITLE 'light x emo';
axis1 order=-10 to 200 by 30; axis2 order=0.6 to 1 by 0.1;
plot accuracy*light=task/ haxis=axis1 vaxis=axis2;
symbol1 i=std14jmt mode =include w=0.2 c=blue v=none l=0.1;
symbol2 i=std14jmt mode =include w=0.2 c=red v=none l=0.1;
run;quit;run;
proc sort data=hypo; by task light subj;run;
proc gplot data=hypo; TITLE 'light x emo';
axis1 order=-10 to 200 by 30; axis2 order=600 to 1500 by 100;
plot RT_hits*light=task/ haxis=axis1 vaxis=axis2;
symbol1 i=std14jmt mode =include w=0.2 c=blue v=none l=0.1;
symbol2 i=std14jmt mode =include w=0.2 c=red v=none l=0.1;
run;quit;run;
PROC CORR pearson spearman data=hypo plots=scatter (alpha=.20 .30) out = PearsonR; var Accuracy light; by subj; run; *you can put as many variables as you want;
ODS GRAPHICS OFF ;
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment