Fig. 3: Primary outcome results - IRIS. | npj Digital Medicine

Fig. 3: Primary outcome results - IRIS.

From: An explainable machine learning-based phenomapping strategy for adaptive predictive enrichment in randomized clinical trials

Fig. 3

(a) Cumulative patient enrollment, (b) cumulative primary outcome events, (c) Cox regression-derived estimates (Hazard Ratios) for the effects of pioglitazone versus placebo on the primary outcome, and (d) corresponding p values. For each timepoint, three bars are presented; the first one on the left (blue bar) denotes the phenomapping-enriched adaptive runs, the middle (dark red) denotes an adaptive simulation after random reshuffling of baseline covariates while preserving the same average treatment effect (negative control), and the one on the right (light red) refers to the original trial. The error bars denote the standard error of mean (SEM) across n = 10 adaptive simulations. Between-group comparisons and related p values are derived from one-sample t-test (using the original “including all patients” arm as reference). HTE heterogeneous treatment effect, IRIS Insulin Resistance Intervention after Stroke.

Back to article page