Towards symbolic regression for interpretable clinical decision scores

TL;DR

This paper introduces Brush, a symbolic regression algorithm that integrates rule-based logic for creating interpretable clinical decision scores with high predictive accuracy.

Contribution

Brush combines decision-tree-like splitting with non-linear optimization, enabling data-driven, interpretable clinical risk scores that outperform or match existing models.

Findings

Brush achieves Pareto-optimal performance on SRBench.

It successfully recapitulates two clinical scoring systems with high accuracy.

Brush produces simpler models with comparable or better predictive performance.

Abstract

Medical decision-making makes frequent use of algorithms that combine risk equations with rules, providing clear and standardized treatment pathways. Symbolic regression (SR) traditionally limits its search space to continuous function forms and their parameters, making it difficult to model this decision-making. However, due to its ability to derive data-driven, interpretable models, SR holds promise for developing data-driven clinical risk scores. To that end we introduce Brush, an SR algorithm that combines decision-tree-like splitting algorithms with non-linear constant optimization, allowing for seamless integration of rule-based logic into symbolic regression and classification models. Brush achieves Pareto-optimal performance on SRBench, and was applied to recapitulate two widely used clinical scoring systems, achieving high accuracy and interpretable models. Compared to decision trees, random forests, and other SR methods, Brush achieves comparable or superior predictive performance while producing simpler models.