<p>Estimation of LDA coefficients using the entire dataset.</p>

Let denote the resulting predictor matrix, and let and be the mean vectors of the predictors in the nonCR and CR groups, respectively. The within-class covariance matrices and were computed for each group, and a pooled within-class covariance matrix, , was then formed, where and are the numbers of nonCR and CR cases.

The discriminant weight vector was obtained as and prior class probabilities and were set equal to the empirical class proportions. The scalar intercept for the two-group discriminant function was calculated as For each subject with predictor vector , the linear discriminant score was defined as and classification was given by for CR and for nonCR. The elements of thus quantify the contribution of each standardized predictor to shifting a patient towards the CR or nonCR side of the discriminant boundary.</p> <p>(DOCX)</p>