A statistical multivariate model using principal component analysis of Raman spectra from stem cell-derived CMs and non-CMs achieved 97% specificity and 96% sensitivity [64].
The variables with p value less than 0.5 in the bivariate analysis were entered to the multivariate model using the enter regression method.
After describing the relationships between explanatory variables, we built a multivariate model using logistic Generalized Linear Models (GLMs) and estimated parameters with a Generalized Estimating Equations (GEE) procedure.
No significant association between making an index visit to the ED at times of high in-hospital bed occupancy and the probability for making an unplanned 72-h revisit was revealed in the multivariate model using occupancy <85% for reference.
The multivariate model using DNA methylation, eight histone modifications, and two regulatory protein complexes (CTCF and cohesin) as predictors has better performance (accuracy = 95.1%) than any univariate models of single predictors.
In the multivariate model using modified logistic regression, history of suicide attempt (aOR = 1.36, 95%CI 1.03 1.78), two or more medical conditions (aOR = 1.87, 95%CI 1.30 2.67) and use of crystal methamphetamine in the week prior to their current sentence (aOR = 1.52, 95%CI 1.05 2.22) were statistically significantly associated with current poor psychiatric well-being.
We developed the multivariate model using a backward stepwise regression analysis [23].
Clinical and laboratory features with P<0.15 were then entered into a multivariate model using forward and backward likelihood ratio logistic regression.
We also repeated the third multivariate model using this subsample.
Subsequently, we included these characteristics in the multivariate model using backward selection.
A multivariate model using a stepwise regression analysis was performed to correct for confounders.
