AI Predicts Peak VO2 From Cardiac Ultrasound
Researchers from Weill Cornell, Cornell Tech, Columbia, and NewYork-Presbyterian published March 3 in npj Digital Medicine that a multimodal AI model predicts peak VO2 from routine cardiac ultrasound videos and electronic health records. Trained on 1,000 patients and validated on 127 across three campuses, the model achieved roughly 85% discrimination. The approach could expand screening for advanced heart failure and reduce reliance on specialized cardiopulmonary exercise testing, pending clinical trials.
Key Points
- 1Predicts peak VO2 with ~85% accuracy using echocardiogram videos and EHR data
- 2Reduces reliance on cardiopulmonary exercise testing, expanding diagnostic reach beyond specialized centers
- 3Enables scalable screening to identify advanced heart failure patients suitable for specialist referral
Scoring Rationale
Peer-reviewed, multicampus results and clear clinical relevance, but limited external sample size and pending prospective clinical validation.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
