Cardiology

Systematic review: AI with decision support improves workflow, engagement, and reduces mortality

Systematic review of 32 randomized trials (27 meta-analyzed) found AI interventions reduced diagnostic/workflow time (SMD -0.71). Decision-support implementations were associated with lower all-cause mortality (RR 0.84, 95% CI 0.75–0.94; NNT = 32). Behavioral nudges improved medication adherence (RR 1.59; NNT = 12) and time savings ranged about 30–120 s per task. Limitations include restricted blinding and few sham-AI controls, which may bias effect estimates. 1

Large multinational validation supports PREVENT and SCORE2 for CVD risk stratification

Validation across 44 cohorts and 18 trials included 6,422,714 individuals with mean follow-up 5.1 years. Both PREVENT and SCORE2 demonstrated generally good discrimination and calibration across regions and trial settings. Analyses produced scaling factors to enable PREVENT risk prediction over 1–9 years for research and trial enrollment. These results support adopting PREVENT or SCORE2 for clinical risk estimation, with attention to outcome-definition differences. 2

Transitional-care intervention did not increase high guideline-directed therapy adherence at 6 months

Multicenter randomized trial of 982 acute HF patients found no difference in high guideline-adherence at 6 months (49.6% vs 44.6%; OR 1.12, P=0.37). No significant changes occurred in quality-of-life scores or clinical outcomes in the intention-to-treat analysis. Per-protocol analysis showed patients receiving all intervention components had higher adherence (adjusted OR 1.56). Authors conclude implementation fidelity, not intervention concept, likely determines benefit for transitional HF care. 3