AgentScore: Autoformulation of Deployable Clinical Scoring Systems

TL;DR

AgentScore is a novel method that uses large language models and a verification loop to automatically generate interpretable, deployable clinical scoring systems that outperform existing methods across multiple tasks.

Contribution

It introduces a semantically guided optimization approach leveraging LLMs for creating clinically deployable scoring systems with strong predictive performance.

Findings

Outperforms existing score-generation methods on eight tasks.

Achieves AUC comparable to flexible interpretable models.

Outperforms established guideline scores on external tasks.

Abstract

Modern clinical practice relies on evidence-based guidelines implemented as compact scoring systems composed of a small number of interpretable decision rules. While machine-learning models achieve strong performance, many fail to translate into routine clinical use due to misalignment with workflow constraints such as memorability, auditability, and bedside execution. We argue that this gap arises not from insufficient predictive power, but from optimizing over model classes that are incompatible with guideline deployment. Deployable guidelines often take the form of unit-weighted clinical checklists, formed by thresholding the sum of binary rules, but learning such scores requires searching an exponentially large discrete space of possible rule sets. We introduce AgentScore, which performs semantically guided optimization in this space by using LLMs to propose candidate rules and a deterministic, data-grounded verification-and-selection loop to enforce statistical validity and deployability constraints. Across eight clinical prediction tasks, AgentScore outperforms existing score-generation methods and achieves AUC comparable to more flexible interpretable models despite operating under stronger structural constraints. On two additional externally validated tasks, AgentScore achieves higher discrimination than established guideline-based scores.