COTCAgent: Preventive Consultation via Probabilistic Chain-of-Thought Completion

2026-05-14

AI Summary

This work addresses the challenge that current large language models struggle to model irregularly sampled time-series data and sparse labels in longitudinal electronic health record (EHR) reasoning, often generating clinically implausible trend hallucinations. To mitigate this, the authors propose COTCAgent—a hierarchical reasoning framework that decouples statistical computation, feature alignment, and language generation. The framework incorporates a probabilistic chain-of-thought completion mechanism, integrates executable code generation, leverages a symptom–trend–disease knowledge base with weighted scoring, and enforces structured evidence constraints. Evaluated on both an in-house dataset and HealthBench, COTCAgent achieves Top-1 accuracy of 90.47% and 70.41%, respectively, substantially outperforming existing medical agents and state-of-the-art large language models while enhancing both accuracy and interpretability in longitudinal EHR reasoning.

Abstract

As large language models empower healthcare, intelligent clinical decision support has developed rapidly. Longitudinal electronic health records (EHR) provide essential temporal evidence for accurate clinical diagnosis and analysis. However, current large language models have critical flaws in longitudinal EHR reasoning. First, lacking fine-grained statistical reasoning, they often hallucinate clinical trends and metrics when quantitative evidence is textually implied, biasing diagnostic inference. Second, non-uniform time series and scarce labels in longitudinal EHR hinder models from capturing long-range temporal dependencies, limiting reliable clinical reasoning. To address the above limitations, this work presents the Probabilistic Chain-of-Thought Completion Agent (COTCAgent), a hierarchical reasoning framework for longitudinal electronic health records. It consists of three core modules. The Temporal-Statistics Adapter (TSA) converts analytical plans into executable code for standardized trend output. The Chain-of-Thought Completion (COTC) layer leverages a symptom-trend-disease knowledge base with weighted scoring to evaluate disease risk, while the bounded completion module acquires structured evidence through standardized inquiries and iterative scoring constraints to ensure rigorous reasoning. By decoupling statistical computation, feature matching, and language generation, the framework eliminates reliance on complex multi-modal inputs and enables efficient longitudinal record analysis with lower computational overhead. Experimental results show that COTCAgent powered by Baichuan-M2 achieves 90.47% Top-1 accuracy on the self-built dataset and 70.41% on HealthBench, outperforming existing medical agents and mainstream large language models. The code is available at https://github.com/FrankDengAI/COTCAgent/.

Problem

• longitudinal electronic health records
• statistical reasoning
• temporal dependencies
• clinical decision support
• hallucination

Innovation

• Probabilistic Chain-of-Thought
• Longitudinal EHR Reasoning
• Temporal-Statistics Adapter
• Bounded Completion
• Clinical Decision Support