Learning Interpretable Point-Based Clinical Risk Scores via Direct Optimization
Researchers have developed new machine learning algorithms to directly optimize interpretable clinical risk scores. These algorithms use a flexible greedy optimization strategy to learn additive scoring rules with non-negative integer points. The method was applied to a large electronic health record cohort to create a comorbidity score for predicting post-discharge mortality. AI
IMPACT Introduces a novel machine learning approach for creating more accurate and interpretable clinical risk scores, potentially improving patient care and outcomes.