Confidence intervals, logistic regression

Deal all,

I am using logistic regression to fit my data. In the end I want a prevision with confidence intervals.

I am using the code that follows:

 CVO = cvpartition(y2,'k',10);
C=zeros(96,3);
for i = 1:CVO.NumTestSets
trIdx = CVO.training(i);
teIdx = CVO.test(i);
mdl = fitglm(x1(trIdx, :),y2(trIdx, :),'linear','Distribution','binomial','Link','logit')
[score, yci] = predict(mdl,x1(teIdx, :));
C(teIdx)=score;
C(teIdx, 2)=yci(:,1);
C(teIdx, 3)=yci(:,2);

For each observation I obtain the probability and confidence intervals (95%) that are not simetric around the mean prevision. How are these confidence intervals determined by MatLabR2014a?

Thanks in advance.

Best,

Elizabeth Vieira