# From Documents to Spans: Span-Centric Learning for LLM-based Explainable Medical Coding

arXiv: https://arxiv.org/abs/2603.15270 · HTML: https://mlsanity.com/paper/2603.15270/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 29 references that are in the map.
- First posted on arXiv 16 Mar 2026.

## Builds on (most cited first, 22 of 22)

- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models](https://mlsanity.com/paper/2403.13372/) (2024)
- [Explainable Prediction of Medical Codes from Clinical Text](https://mlsanity.com/paper/1802.05695/) (2018)
- [ACI-BENCH: a Novel Ambient Clinical Intelligence Dataset for Benchmarking Automatic Visit Note Generation](https://mlsanity.com/paper/2306.02022/) (2023)
- [Code Synonyms Do Matter: Multiple Synonyms Matching Network for Automatic ICD Coding](https://mlsanity.com/paper/2203.01515/) (2022)
- [PLM-ICD: Automatic ICD Coding with Pretrained Language Models](https://mlsanity.com/paper/2207.05289/) (2022)
- [Automated Medical Coding on MIMIC-III and MIMIC-IV: A Critical Review and Replicability Study](https://mlsanity.com/paper/2304.10909/) (2023)
- [MDACE: MIMIC Documents Annotated with Code Evidence](https://mlsanity.com/paper/2307.03859/) (2023)
- [CoRelation: Boosting Automatic ICD Coding Through Contextualized Code Relation Learning](https://mlsanity.com/paper/2402.15700/) (2024)
- [Automated clinical coding using off-the-shelf large language models](https://mlsanity.com/paper/2310.06552/) (2023)
- [An Unsupervised Approach to Achieve Supervised-Level Explainability in Healthcare Records](https://mlsanity.com/paper/2406.08958/) (2024)
- [MedCodER: A Generative AI Assistant for Medical Coding](https://mlsanity.com/paper/2409.15368/) (2024)
- [Code Like Humans: A Multi-Agent Solution for Medical Coding](https://mlsanity.com/paper/2509.05378/) (2025)
- [Accurate and Well-Calibrated ICD Code Assignment Through Attention Over Diverse Label Embeddings](https://mlsanity.com/paper/2402.03172/) (2024)
- [DKEC: Domain Knowledge Enhanced Multi-Label Classification for Diagnosis Prediction](https://mlsanity.com/paper/2310.07059/) (2023)
- [A General Knowledge Injection Framework for ICD Coding](https://mlsanity.com/paper/2505.18708/) (2025)
- [Beyond Label Attention: Transparency in Language Models for Automated Medical Coding via Dictionary Learning](https://mlsanity.com/paper/2411.00173/) (2024)
- [Aligning AI Research with the Needs of Clinical Coding Workflows: Eight Recommendations Based on US Data Analysis and Critical Review](https://mlsanity.com/paper/2412.18043/) (2024)
- [Toward Reliable Clinical Coding with Language Models: Verification and Lightweight Adaptation](https://mlsanity.com/paper/2510.07629/) (2025)
- [RuCCoD: Towards Automated ICD Coding in Russian](https://mlsanity.com/paper/2502.21263/) (2025)

## Abstract (arXiv)

Medical coding, which assigns standardized ICD codes to clinical documents, is a fundamental yet challenging task in modern healthcare. It requires extensive knowledge of a large code space and the ability to identify and interpret sparse supporting evidence in long, noisy clinical documents. Most existing methods rely on document-level supervision, which provides limited guidance on supporting evidence and insufficient training signals for rare codes. To address these limitations, we propose Span-Centric Learning (SCL), which separates coding knowledge acquisition from its application in full documents. SCL first learns coding knowledge from short spans, then learns to use this knowledge in full documents from a small set of annotated data. Experiments show that, our approach enables an 8B LLM to surpass state-of-the-art methods using only 248 complete documents for training. It also provides explicit supporting evidence for each predicted code, with strong alignment to human annotations. Further analysis shows that the model can learn knowledge of codes absent from documents through spans and apply that knowledge to full documents.