For the secondary outcome of decline in discharge status (n=41 events), two separate multivariate logistic regression models were used to evaluate delirium as a binary predictor (ever vs never) and as an ordinal variable (0=reference, 1, 2, 3 days; figure 2).
Co-medications were evaluated as binary covariates (use/no-use).
To facilitate clinical interpretation of the resulting clusters, sustained remission and sustained relapse were evaluated as binary outcomes for each cluster.
HIV RNA at follow-up under the lowest threshold level (under 2.5 copies/ml), primary HIV infection at the start of therapy, stranger origin, male gender, risk categories (heterosexual, men who have sex with men, drug addiction) and tropism based on the presence of R5 or X4 variants in PBMCs at T0, T1 and T2 were evaluated as binary variables.
In both COPDGene and the FHS, the percentage of HAAs was evaluated as both a continuous and as a binary variable selected on the basis of a prior presentation [ 11] (e.g. ≥10 % or <10%and≥6.4%.4 % or <6.4%[≥6.4%4 % corresponds to the ≥ upper 95th percentile] of the lung occupied by HAAs in COPDGene).
Reproducibility between duplicate runs was evaluated using a binary comparison map.
The immunohistochemical parameters were evaluated on either a binary scale ('no reaction' or 'reaction', coded as '−' or '+') or an ordinal scale 'no reaction', 'weak', 'moderate' or 'strong reaction' (coded as '−', '+', '++' or '+++').
The bottleneck prediction algorithm is evaluated by treating it as a binary classifier problem and adapted the appropriate evaluation metrics.
We need at most n=97 (=1.96 × 0.25/0.1) patients to satisfy a 0.1 precision in evaluating a binary mutation marker (in formula (2), let p=0.5, as it gives the highest possible value for the right hand side).
