nirs.modules.MixedEffects

See also nirs.modules.MixedEffectsConnectivity

Untitled

nirs.modules.MixedEffects

This module runs second-level (or group) statistical models for task-based paradigm.
Second-level (e.g., group) models using fixed-effects, mixed-effects, or ANOVA are offered as modules in the toolbox. For all three of these processing modules, a formula is specified to define the statistical model used. This formula is defined in Wilkinson-Rogers notation [1], which is also used in Matlab (e.g., fitlme.m function) and R software packages. The statistical model formula is able to use any demographics variables stored in the first-level ChannelStats variables such as subject ID, group membership, age, etc. Both categorical and continuous variables are supported in these models. For more details, see Santosa et al., 2018 [2]. Examples of the Wilkinson–Roger’s notation is shown in this Table.
Screenshot 2024-10-02 at 9.55.39 AM.png
Example Usage (Continuous variable):
for i = 1:10
raw(i) = nirs.testing.simData([], @(t)nirs.testing.blockedStimDesign(t,15,25,2));
% task-based 15s duration, 25s between onset, and 2 conditions
end
 
% Adding demographic manually
% Let’s assume that the SubjectID is in the correct order
for i = 1:10
raw(i).demographics(‘Subject’) = [‘S’ num2str(i)];
end
 
%Let’s make up the table contain SubjectID, Group, AgeMonths, ADHD
for i = 1:10; Subject{i,1} = [‘S’ num2str(i)]; end
for i = 1:5; Group{i,1} = [‘S1’]; end
for i = 6:10; Group{i,1} = [‘S2’]; end
AgeMonths = [60; 65; 68; 60; 69; 72; 61; 68; 70; 66];
ADHD = [2; 6; 3; 9; 5; 7; 3; 7; 5; 4];
demo = table(Subject, Group, AgeMonths, ADHD);
 
job=nirs.modules.AddDemographics;
job.allowMissing = true;
job.demoTable=demo;
job.varToMatch=‘Subject’;
raw=job.run(raw);
 
nirs.createDemographicsTable(raw);
 
% Get hb data
job = nirs.modules.Resample();
job.Fs = 1; %For speed
job = nirs.modules.OpticalDensity(job);
job = nirs.modules.BeerLambertLaw(job);
hb = job.run(raw);
 
% First level Stats
job = nirs.modules.GLM();
Stats = job.run(hb);
…………….Finished 1 of 10. …………….Finished 2 of 10. …………….Finished 3 of 10. …………….Finished 4 of 10. …………….Finished 5 of 10. …………….Finished 6 of 10. …………….Finished 7 of 10. …………….Finished 8 of 10. …………….Finished 9 of 10. …………….Finished 10 of 10.
 
% Second-level or group-level
job = nirs.modules.MixedEffects();
job.formula = ‘beta ~ -1 + cond + cond:ADHD + (1|Subject)’;
job.robust = true; %Recommended to use the robust version
GroupStats1 = job.run(Stats);
 
GroupStats1.conditions
ans = 4×1 cell
‘A’
‘A:ADHD’
‘B’
‘B:ADHD’
%The GroupStats1 have 4 conditions: A, A:ADHD, B, B:ADHD
%A and B is the main terms of the group level obtained from 2 conditions
%A: ADHD and B:ADHD is the group level from ADHD as covariates
 
GroupStats1.draw(‘tstat’,[],‘p<0.05’)
ans =
1×8 Figure array: Figure Figure Figure Figure Figure Figure Figure Figure
Example Usage (Categorical variable):
% Let’s make the ADHD version 2 (low and high)
for i = 1:length(Stats)
if nirs.createDemographicsTable(Stats(i)).ADHD<6
Stats(i).demographics(‘ADHD2’) = ‘low’;
else
Stats(i).demographics(‘ADHD2’) = ‘high’;
end
end
 
% Second-level or group-level
job = nirs.modules.MixedEffects();
job.formula = ‘beta ~ -1 + cond:ADHD2 + (1|Subject)’;
job.robust = true; %Recommended to use the robust version
GroupStats2 = job.run(Stats);
 
GroupStats2.conditions
ans = 4×1 cell
‘A:ADHD2_high’
‘A:ADHD2_low’
‘B:ADHD2_high’
‘B:ADHD2_low’
%This GroupStats2 have 4 conditions:
%A:ADHD2_high, A:ADHD_low, B:ADHD2_high, and B:ADHD_low
 
GroupStats2.draw(‘tstat’,[],‘p<0.05’)
ans =
1×8 Figure array: Figure Figure Figure Figure Figure Figure Figure Figure
References
[1] G. N. Wilkinson, C. E. Rogers, Symbolic Description of Factorial Models for Analysis of Variance, Journal of the Royal Statistical Society Series C: Applied Statistics, Volume 22, Issue 3, November 1973, Pages 392–399, https://doi.org/10.2307/2346786
[2] Santosa, H.; Zhai, X.; Fishburn, F.; Huppert, T. The NIRS Brain AnalyzIR Toolbox. Algorithms 2018, 11, 73. https://doi.org/10.3390/a11050073