Uncertainty-Guided Latent Diagnostic Trajectory Learning for Sequential Clinical Diagnosis

TL;DR

This paper introduces a novel framework for sequential clinical diagnosis that models diagnostic trajectories as latent paths, leveraging planning and diagnostic LLMs to improve accuracy and efficiency.

Contribution

It proposes a Latent Diagnostic Trajectory Learning framework that explicitly models evidence acquisition paths, enhancing diagnostic performance with fewer tests.

Findings

Outperforms baselines in diagnostic accuracy on MIMIC-CDM.

Requires fewer diagnostic tests than existing methods.

Trajectory-level posterior alignment is crucial for success.

Abstract

Clinical diagnosis requires sequential evidence acquisition under uncertainty. However, most Large Language Model (LLM) based diagnostic systems assume fully observed patient information and therefore do not explicitly model how clinical evidence should be sequentially acquired over time. Even when diagnosis is formulated as a sequential decision process, it is still challenging to learn effective diagnostic trajectories. This is because the space of possible evidence-acquisition paths is relatively large, while clinical datasets rarely provide explicit supervision information for desirable diagnostic paths. To this end, we formulate sequential diagnosis as a Latent Diagnostic Trajectory Learning (LDTL) framework based on a planning LLM agent and a diagnostic LLM agent. For the diagnostic LLM agent, diagnostic action sequences are treated as latent paths and we introduce a posterior distribution that prioritizes trajectories providing more diagnostic information. The planning LLM agent is then trained to follow this distribution, encouraging coherent diagnostic trajectories that progressively reduce uncertainty. Experiments on the MIMIC-CDM benchmark demonstrate that our proposed LDTL framework outperforms existing baselines in diagnostic accuracy under a sequential clinical diagnosis setting, while requiring fewer diagnostic tests. Furthermore, ablation studies highlight the critical role of trajectory-level posterior alignment in achieving these improvements.