MIMIC-SR-ICD11: A Dataset for Narrative-Based Diagnosis

TL;DR

This paper introduces MIMIC-SR-ICD11, a large dataset of clinical narratives aligned with ICD-11, and proposes LL-Rank, a likelihood-based re-ranking method that improves diagnostic label prediction from clinical reports.

Contribution

The paper provides a new narrative-based diagnostic dataset aligned with ICD-11 and introduces LL-Rank, a novel likelihood-based re-ranking framework that enhances label prediction accuracy.

Findings

LL-Rank outperforms baseline methods across seven models.

Ablation shows LL-Rank's PMI scoring isolates semantic compatibility.

Dataset enables improved diagnosis from clinical narratives.

Abstract

Disease diagnosis is a central pillar of modern healthcare, enabling early detection and timely intervention for acute conditions while guiding lifestyle adjustments and medication regimens to prevent or slow chronic disease. Self-reports preserve clinically salient signals that templated electronic health record (EHR) documentation often attenuates or omits, especially subtle but consequential details. To operationalize this shift, we introduce MIMIC-SR-ICD11, a large English diagnostic dataset built from EHR discharge notes and natively aligned to WHO ICD-11 terminology. We further present LL-Rank, a likelihood-based re-ranking framework that computes a length-normalized joint likelihood of each label given the clinical report context and subtracts the corresponding report-free prior likelihood for that label. Across seven model backbones, LL-Rank consistently outperforms a strong generation-plus-mapping baseline (GenMap). Ablation experiments show that LL-Rank's gains primarily stem from its PMI-based scoring, which isolates semantic compatibility from label frequency bias.