[2604.04241] Learning An Interpretable Risk Scoring System for Maximizing Decision Net Benefit

Computer Science > Machine Learning arXiv:2604.04241 (cs) [Submitted on 5 Apr 2026] Title:Learning An Interpretable Risk Scoring System for Maximizing Decision Net Benefit Authors:Wenhao Chi, Ş. İlker Birbil View a PDF of the paper titled Learning An Interpretable Risk Scoring System for Maximizing Decision Net Benefit, by Wenhao Chi and \c{S}. \.Ilker Birbil View PDF HTML (experimental) Abstract:Risk scoring systems are widely used in high-stakes domains to assist decision-making. However, existing approaches often focus on optimizing predictive accuracy or likelihood-based criteria, which may not align with the main goal of maximizing utility. In this paper, we propose a novel risk scoring system that directly optimizes net benefit over a range of decision thresholds. The model is formulated as a sparse integer linear programming problem which enables the construction of a transparent scoring system with integer coefficients, and hence, facilitates interpretation and practical application. We also establish fundamental relationships among net benefit, discrimination, and calibration. Our analysis proves that optimizing net benefit also guarantees conventional performance measures. We thoroughly evaluated our method on multiple public datasets as well as on a real-world clinical dataset. This computational study demonstrated that our interpretable method can effectively achieve high net benefit while maintaining competitive discrimination and calibration performance. Comm...