Scaling Clinician-Grade Feature Generation from Clinical Notes with Multi-Agent Language Models

TL;DR

This paper introduces SNOW, a multi-agent LLM system that automates clinician-level feature extraction from clinical notes, achieving high prediction accuracy and scalability while reducing manual effort significantly.

Contribution

The study presents a novel multi-agent LLM framework, SNOW, that automates detailed clinical feature generation from notes, matching expert performance and enabling scalable, interpretable outcome prediction.

Findings

SNOW achieves comparable performance to manual feature extraction.

SNOW reduces human effort by approximately 48-fold.

SNOW generalizes well to external cohorts and different conditions.

Abstract

Developing accurate clinical prediction models is often bottlenecked by the difficulty of deriving meaningful structured features from unstructured EHR notes, a process that traditionally requires manual, unscalable clinical abstraction. In this study, we first established a rigorous patient-level Clinician Feature Generation (CFG) protocol, in which domain experts manually reviewed notes to define and extract nuanced features for a cohort of 147 patients with prostate cancer. As a high-fidelity ground truth, this labor-intensive process provided the blueprint for SNOW (Scalable Note-to-Outcome Workflow), a transparent multi-agent large language model (LLM) system designed to autonomously mimic the iterative reasoning and validation workflow of clinical experts. On 5-year cancer recurrence prediction, SNOW (AUC-ROC 0.767) achieved performance comparable to manual CFG (0.762) and outperformed structured baselines, clinician-guided LLM extraction, and six representational feature generation (RFG) approaches. Once configured, SNOW produced the full patient-level feature table in 12 hours with 5 hours of clinician oversight, reducing human expert effort by approximately 48-fold versus manual CFG. To test scalability where manual CFG is infeasible, we deployed SNOW on an external heart failure with preserved ejection fraction (HFpEF) cohort from MIMIC-IV (n=2,084); without task-specific tuning, SNOW generated prognostic features that outperformed baseline and RFG methods for 30-day (SNOW: 0.851) and 1-year (SNOW: 0.763) mortality prediction. These results demonstrate that a modular LLM agent-based system can scale expert-level feature generation from clinical notes, while enabling interpretable use of unstructured EHR text in outcome prediction and preserving generalizability across a variety of settings and conditions.