Code Like Humans: A Multi-Agent Solution for Medical Coding

Andreas Geert Motzfeldt; Joakim Edin; Casper L. Christensen; Christian Hardmeier; Lars Maaløe; Anna Rogers

doi:10.18653/v1/2025.findings-emnlp.1231

Code Like Humans: A Multi-Agent Solution for Medical Coding

Andreas Geert Motzfeldt
, Joakim Edin
, Casper L. Christensen
, Christian Hardmeier
, Lars Maaløe
, Anna Rogers

Corti.ai
University of Copenhagen

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

In medical coding, experts map unstructured clinical notes to alphanumeric codes for diagnoses and procedures. We introduce `Code Like Humans': a new agentic framework for medical coding with large language models. It implements official coding guidelines for human experts, and it is the first solution that can support the full ICD-10 coding system (+70K labels). It achieves the best performance to date on rare diagnosis codes. Fine-tuned discriminative classifiers retain an advantage for high-frequency codes, to which they are limited. Towards future work, we also contribute an analysis of system performance and identify its `blind spots' (codes that are systematically undercoded).

Original language	English
Title of host publication	Findings of the Association for Computational Linguistics: EMNLP 2025
Editors	Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, Violet Peng
Number of pages	16
Place of Publication	Suzhou, China
Publisher	Association for Computational Linguistics
Publication date	1 Nov 2025
Pages	22612-22627
ISBN (Print)	979-8-89176-335-7
DOIs	https://doi.org/10.18653/v1/2025.findings-emnlp.1231
Publication status	Published - 1 Nov 2025
Event	Empirical Methods in Natural Language Processing - Suzhou, China Duration: 4 Nov 2025 → 9 Nov 2025 Conference number: 4th https://2025.emnlp.org/

Conference

Conference	Empirical Methods in Natural Language Processing
Number	4th
Country/Territory	China
City	Suzhou
Period	04/11/2025 → 09/11/2025
Internet address	https://2025.emnlp.org/

Keywords

ICD-10 coding
Medical coding
Large language models
Clinical natural language processing
Automated coding

Access to Document

10.18653/v1/2025.findings-emnlp.1231

Reducing hallucinations in Medical QA
Rogers, A. (PI)
Innovation Fund Denmark
01/09/2024 → 29/05/2028
Project: Research

Cite this

Motzfeldt, A. G., Edin, J., Christensen, C. L., Hardmeier, C., Maaløe, L., & Rogers, A. (2025). Code Like Humans: A Multi-Agent Solution for Medical Coding. In C. Christodoulopoulos, T. Chakraborty, C. Rose, & V. Peng (Eds.), Findings of the Association for Computational Linguistics: EMNLP 2025 (pp. 22612-22627). Association for Computational Linguistics. https://doi.org/10.18653/v1/2025.findings-emnlp.1231

Motzfeldt, Andreas Geert ; Edin, Joakim ; Christensen, Casper L. et al. / Code Like Humans: A Multi-Agent Solution for Medical Coding. Findings of the Association for Computational Linguistics: EMNLP 2025. editor / Christos Christodoulopoulos ; Tanmoy Chakraborty ; Carolyn Rose ; Violet Peng. Suzhou, China : Association for Computational Linguistics, 2025. pp. 22612-22627

@inproceedings{1df8e96fd1ce4ce7a315e68e714fe881,

title = "Code Like Humans: A Multi-Agent Solution for Medical Coding",

abstract = "In medical coding, experts map unstructured clinical notes to alphanumeric codes for diagnoses and procedures. We introduce `Code Like Humans': a new agentic framework for medical coding with large language models. It implements official coding guidelines for human experts, and it is the first solution that can support the full ICD-10 coding system (+70K labels). It achieves the best performance to date on rare diagnosis codes. Fine-tuned discriminative classifiers retain an advantage for high-frequency codes, to which they are limited. Towards future work, we also contribute an analysis of system performance and identify its `blind spots' (codes that are systematically undercoded).",

keywords = "ICD-10 coding, Medical coding, Large language models, Clinical natural language processing, Automated coding",

author = "Motzfeldt, \{Andreas Geert\} and Joakim Edin and Christensen, \{Casper L.\} and Christian Hardmeier and Lars Maal{\o}e and Anna Rogers",

year = "2025",

month = nov,

day = "1",

doi = "10.18653/v1/2025.findings-emnlp.1231",

language = "English",

isbn = "979-8-89176-335-7",

pages = "22612--22627",

editor = "Christos Christodoulopoulos and Tanmoy Chakraborty and Carolyn Rose and Violet Peng",

booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",

publisher = "Association for Computational Linguistics",

address = "United States",

note = "Empirical Methods in Natural Language Processing ; Conference date: 04-11-2025 Through 09-11-2025",

url = "https://2025.emnlp.org/",

}

Motzfeldt, AG, Edin, J, Christensen, CL, Hardmeier, C, Maaløe, L & Rogers, A 2025, Code Like Humans: A Multi-Agent Solution for Medical Coding. in C Christodoulopoulos, T Chakraborty, C Rose & V Peng (eds), Findings of the Association for Computational Linguistics: EMNLP 2025. Association for Computational Linguistics, Suzhou, China, pp. 22612-22627, Empirical Methods in Natural Language Processing, Suzhou, China, 04/11/2025. https://doi.org/10.18653/v1/2025.findings-emnlp.1231

Code Like Humans: A Multi-Agent Solution for Medical Coding. / Motzfeldt, Andreas Geert; Edin, Joakim; Christensen, Casper L. et al.
Findings of the Association for Computational Linguistics: EMNLP 2025. ed. / Christos Christodoulopoulos; Tanmoy Chakraborty; Carolyn Rose; Violet Peng. Suzhou, China: Association for Computational Linguistics, 2025. p. 22612-22627.

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

TY - GEN

T1 - Code Like Humans: A Multi-Agent Solution for Medical Coding

AU - Motzfeldt, Andreas Geert

AU - Edin, Joakim

AU - Christensen, Casper L.

AU - Hardmeier, Christian

AU - Maaløe, Lars

AU - Rogers, Anna

N1 - Conference code: 4th

PY - 2025/11/1

Y1 - 2025/11/1

N2 - In medical coding, experts map unstructured clinical notes to alphanumeric codes for diagnoses and procedures. We introduce `Code Like Humans': a new agentic framework for medical coding with large language models. It implements official coding guidelines for human experts, and it is the first solution that can support the full ICD-10 coding system (+70K labels). It achieves the best performance to date on rare diagnosis codes. Fine-tuned discriminative classifiers retain an advantage for high-frequency codes, to which they are limited. Towards future work, we also contribute an analysis of system performance and identify its `blind spots' (codes that are systematically undercoded).

AB - In medical coding, experts map unstructured clinical notes to alphanumeric codes for diagnoses and procedures. We introduce `Code Like Humans': a new agentic framework for medical coding with large language models. It implements official coding guidelines for human experts, and it is the first solution that can support the full ICD-10 coding system (+70K labels). It achieves the best performance to date on rare diagnosis codes. Fine-tuned discriminative classifiers retain an advantage for high-frequency codes, to which they are limited. Towards future work, we also contribute an analysis of system performance and identify its `blind spots' (codes that are systematically undercoded).

KW - ICD-10 coding

KW - Medical coding

KW - Large language models

KW - Clinical natural language processing

KW - Automated coding

U2 - 10.18653/v1/2025.findings-emnlp.1231

DO - 10.18653/v1/2025.findings-emnlp.1231

M3 - Article in proceedings

SN - 979-8-89176-335-7

SP - 22612

EP - 22627

BT - Findings of the Association for Computational Linguistics: EMNLP 2025

A2 - Christodoulopoulos, Christos

A2 - Chakraborty, Tanmoy

A2 - Rose, Carolyn

A2 - Peng, Violet

PB - Association for Computational Linguistics

CY - Suzhou, China

T2 - Empirical Methods in Natural Language Processing

Y2 - 4 November 2025 through 9 November 2025

ER -

Motzfeldt AG, Edin J, Christensen CL, Hardmeier C, Maaløe L, Rogers A. Code Like Humans: A Multi-Agent Solution for Medical Coding. In Christodoulopoulos C, Chakraborty T, Rose C, Peng V, editors, Findings of the Association for Computational Linguistics: EMNLP 2025. Suzhou, China: Association for Computational Linguistics. 2025. p. 22612-22627 doi: 10.18653/v1/2025.findings-emnlp.1231

Code Like Humans: A Multi-Agent Solution for Medical Coding

Abstract

Conference

Keywords

Access to Document

Fingerprint

Projects

Reducing hallucinations in Medical QA

Cite this