Can Post-Training Turn LLMs into Good Medical Coders? An Empirical Study of Generative ICD Coding

• URL: http://arxiv.org/abs/2606.13940v1
• Date: Thu, 11 Jun 2026 22:04:50 GMT
• Title: Can Post-Training Turn LLMs into Good Medical Coders? An Empirical Study of Generative ICD Coding
• Abstract summary: Generative large language models (LLMs) are often reported as weak medical coders.<n>We present a controlled empirical study of post-training for generative ICD coding.
• Score: 40.63304781183767
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Automated International Classification of Diseases (ICD) coding is a core medical-coding task for billing, epidemiology, and clinical decision support. Generative large language models (LLMs) are often reported as weak medical coders, but this finding mainly comes from inference-time settings such as prompting, retrieval, reranking, or tool use, leaving the role of task-specific post-training underexplored. We present a controlled empirical study of post-training for generative ICD coding, comparing discriminative baselines with LLM coders across prompting, supervised fine-tuning, and reinforcement learning under a common protocol and metric set. To our knowledge, this is the first study to evaluate RL-based post-training for generative LLM coders in ICD coding. We further introduce PHI, a diagnostic curriculum that extends GRPO to refine missed-code cases. Our results show that prompting-only evaluation substantially underestimates the potential of LLMs for ICD coding. SFT provides the main capability jump, GRPO further improves code-set prediction beyond SFT, and PHI provides targeted gains on macro-level performance. These findings suggest that the main bottleneck is not the generative formulation alone, but how the model is adapted and optimized for full-taxonomy recall. We release our code, data splits, and checkpoints at https://github.com/AlexandreWANG915/LLM4ICD.

Related papers

• From Documents to Spans: Code-Centric Learning for LLM-based ICD Coding [29.729356191729888]
  We propose Code-Centric Learning, a training framework that shifts supervision from full clinical documents to scalable, short evidence spans.<n>Our proposed framework consists of a mixed training strategy and code-centric data expansion, which substantially reduces training cost, improves accuracy on unseen ICD codes and preserves interpretability.
  arXiv  Detail & Related papers  (2026-03-16T13:37:01Z)
• From Code Foundation Models to Agents and Applications: A Practical Guide to Code Intelligence [150.3696990310269]
  Large language models (LLMs) have transformed automated software development by enabling direct translation of natural language descriptions into functional code.<n>We provide a comprehensive synthesis and practical guide (a series of analytic and probing experiments) about code LLMs.<n>We analyze the code capability of the general LLMs (GPT-4, Claude, LLaMA) and code-specialized LLMs (StarCoder, Code LLaMA, DeepSeek-Coder, and QwenCoder)
  arXiv  Detail & Related papers  (2025-11-23T17:09:34Z)
• Is Compression Really Linear with Code Intelligence? [60.123628177110206]
  textitFormat Annealing is a lightweight, transparent training methodology designed to assess the intrinsic capabilities of pre-trained models equitably.<n>Our empirical results reveal a fundamental logarithmic relationship between measured code intelligence and bits-per-character (BPC)<n>Our work provides a more nuanced understanding of compression's role in developing code intelligence and contributes a robust evaluation framework in the code domain.
  arXiv  Detail & Related papers  (2025-05-16T16:59:14Z)
• SnipGen: A Mining Repository Framework for Evaluating LLMs for Code [51.07471575337676]
  Language Models (LLMs) are trained on extensive datasets that include code repositories.<n> evaluating their effectiveness poses significant challenges due to the potential overlap between the datasets used for training and those employed for evaluation.<n>We introduce SnipGen, a comprehensive repository mining framework designed to leverage prompt engineering across various downstream tasks for code generation.
  arXiv  Detail & Related papers  (2025-02-10T21:28:15Z)
• OpenCoder: The Open Cookbook for Top-Tier Code Large Language Models [76.59316249991657]
  Large language models (LLMs) for code have become indispensable in various domains, including code generation, reasoning tasks and agent systems.<n>While open-access code LLMs are increasingly approaching the performance levels of proprietary models, high-quality code LLMs remain limited.<n>We introduce OpenCoder, a top-tier code LLM that not only achieves performance comparable to leading models but also serves as an "open cookbook" for the research community.
  arXiv  Detail & Related papers  (2024-11-07T17:47:25Z)
• Exploring LLM Multi-Agents for ICD Coding [15.730751450511333]
  The proposed multi-agent method for ICD coding effectively mimics the real-world coding process and improves performance on both common and rare codes. Our method achieves comparable results to state-of-the-art ICD coding methods that require extensive pre-training or fine-tuning, and outperforms them in rare code accuracy, and explainability.
  arXiv  Detail & Related papers  (2024-04-01T15:17:39Z)
• Automated clinical coding using off-the-shelf large language models [10.365958121087305]
  The task of assigning diagnostic ICD codes to patient hospital admissions is typically performed by expert human coders. Efforts towards automated ICD coding are dominated by supervised deep learning models. In this work, we leverage off-the-shelf pre-trained generative large language models to develop a practical solution.
  arXiv  Detail & Related papers  (2023-10-10T11:56:48Z)
• Read, Attend, and Code: Pushing the Limits of Medical Codes Prediction from Clinical Notes by Machines [0.42641920138420947]
  We present our Read, Attend, and Code (RAC) model for learning the medical code assignment mappings. RAC establishes a new state of the art (SOTA) considerably outperforming the current best Macro-F1 by 18.7%. This new milestone marks a meaningful step toward fully autonomous medical coding (AMC) in machines.
  arXiv  Detail & Related papers  (2021-07-10T06:01:58Z)
• TransICD: Transformer Based Code-wise Attention Model for Explainable ICD Coding [5.273190477622007]
  International Classification of Disease (ICD) coding procedure has been shown to be effective and crucial to the billing system in medical sector. Currently, ICD codes are assigned to a clinical note manually which is likely to cause many errors. In this project, we apply a transformer-based architecture to capture the interdependence among the tokens of a document and then use a code-wise attention mechanism to learn code-specific representations of the entire document.
  arXiv  Detail & Related papers  (2021-03-28T05:34:32Z)
• A Meta-embedding-based Ensemble Approach for ICD Coding Prediction [64.42386426730695]
  International Classification of Diseases (ICD) are the de facto codes used globally for clinical coding. These codes enable healthcare providers to claim reimbursement and facilitate efficient storage and retrieval of diagnostic information. Our proposed approach enhances the performance of neural models by effectively training word vectors using routine medical data as well as external knowledge from scientific articles.
  arXiv  Detail & Related papers  (2021-02-26T17:49:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.