Fig. 3: Subgroup analyses for the risk-stratification for CTRCD using AI-CTRCD scores. | Nature Communications

Fig. 3: Subgroup analyses for the risk-stratification for CTRCD using AI-CTRCD scores.

From: Artificial intelligence-enabled prediction of chemotherapy-induced cardiotoxicity from baseline electrocardiograms

Fig. 3

Kaplan–Meyer plots showing the cumulative incidence of CTRCD between high and low AI-CTRCD scores in those with hematologic malignancies (A), solid tumors (B), male sex (C), female sex (D), baseline LVEF > 60% (E), baseline LVEF ≤ 60% (F), and those treated with initial anthracycline dose of >40 mg/m2 (G) and ≤40 mg/m2 (H). The same cutoff as the primary analysis was used across all subgroups to define high and low AI-CTRCD score. Patients with high AI-CTRCD score were consistently at higher risk of CTRCD compared to patients with low AI-CTRCD score across subgroups. The transparent ribbons indicate 95%CI. Source data are provided as a Source Data file. CTRCD cancer therapy-related cardiac dysfunction, HR hazard ratio, AI artificial intelligence, LVEF left ventricular ejection fraction.

Back to article page