Posted 2/17/2024, 3:08:00 AM
Study Tests AI Tool's Ability to Predict Risk of Hospital-Acquired Kidney Injury, But Finds Limitations
- Commercial machine learning tool tested for predicting hospital-acquired acute kidney injury (HA-AKI) risk
- Model better at ruling out low-risk patients than identifying high-risk patients for HA-AKI
- Performance lower than developer's internal validation, highlighting need to validate AI models
- Could support clinical decisions but further study needed before implementation
- More reliable predicting Stage 1 HA-AKI versus more severe cases