This work proposes Knowledge-Guided Reasoning over Clinical Evidence with LLMs (KREL), a framework that leverages LLMs for clinical text understanding and reasoning while integrating external ICD coding guidelines as structured knowledge, and enables tight coupling between domain knowledge and LLM reasoning.
Abstract
Automatic Medical Coding (AMC), which assigns standardized International Classification of Diseases (ICD) codes to clinical notes, is essential for medical reimbursement, quality reporting, and clinical research. Existing pre-trained language model (PLM)-based methods typically formulate AMC as an extreme multi-label classification problem over a predefined code set, while recent large language model (LLM)-based approaches instead frame it as generation or multi-step reasoning. However, key challenges remain, including the extreme length of clinical notes that hinders effective interpretation, the vast ICD label space, and complex coding rules that are not explicitly captured by LLMs. In this work, we propose Knowledge-Guided Reasoning over Clinical Evidence with LLMs (KREL), a framework that leverages LLMs for clinical text understanding and reasoning while integrating external ICD coding guidelines as structured knowledge. This design enables tight coupling between domain knowledge and LLM reasoning, reducing hallucinations and improving compliance with coding standards. Experiments on benchmark datasets show that KREL consistently outperforms strong PLM-based and state-of-the-art LLM-based baselines.
Background ICD-10 coding is an essential process in healthcare systems that supports clinical management, reimbursement, and health data analytics. However, the complexity of its hierarchical structure and the large number of available codes make manual coding limited in terms of time, cost, and consistency. Despite growing research in this area, evidence remains fragmented, particularly regarding real-world implementation readiness. Objective To review and synthesize existing knowledge on algorithms, datasets, evaluation methods, and real-world implementation readiness of automatic ICD-10 coding systems. Methods Eligible studies were original research articles, preprints, or conference papers published in English between January 1, 2020 and December 31, 2025, and retrieved from seven academic databases: Scopus, PubMed, Web of Science, IEEE Xplore, ACM Digital Library, arXiv, and Google Scholar. Studies were included if they investigated automatic ICD-10 coding from clinical text using machine learning, deep learning, transformer-based, or large language model (LLM) approaches. Methodological quality was assessed using a research-question-driven appraisal framework. This systematic review followed PRISMA 2020 guidance and was preregistered in the Open Science Framework (OSF) at https://osf.io/cegqk. Results A total of 257 records were identified, of which 24 studies met the inclusion criteria and contributed 296 experimental evaluations overall. Study quality was high in 7 studies, moderate in 8, and limited by technical or methodological concerns in 9. Hybrid deep learning (Hybrid DL) was most often used as the main automated coding approach, while machine learning (ML) and rule-based approaches were mainly used as baselines. F1-macro was consistently lower than F1-micro among studies reporting both metrics. Hybrid DL showed the most stable performance under all-code or full-code evaluation, while AI model performance varied by the documents-per-label (D/L) ratio. Discussion The evidence indicates continued technical progress, particularly through Hybrid DL and transformer-based approaches, while LLM-based methods remain emerging and less consistently effective for structured multi-label coding. The observed D/L–performance relationship suggested that AI model selection should consider dataset structure and label support, in addition to algorithmic complexity. Conclusion AI-based automatic ICD-10 coding is a promising approach for clinical coding support. Future research should prioritize rare-label imbalance, reproducibility, explainability, and validation across diverse clinical settings. Systematic Review Registration https://osf.io/cegqk.
K. Tangudomkit, Sawrawit Chairat, Sitthichok Chaichulee· Frontiers in Digital Health· 0 citations
International Classification of Diseases (ICD) codes enable correct billing, insurance reimbursement, and healthcare analytics. However, manual coding is time-consuming, expensive, and error-prone, creating bottlenecks in clinical workflow and limiting scalability. Artificial intelligence (AI) has emerged as a promising solution for automated ICD code assignment from unstructured clinical text. This systematic review explores the current state of automated ICD coding research, examining models applied to diverse clinical documents including discharge summaries, electronic health records, nursing notes, and pathology reports. Following PRISMA guidelines, we searched six databases for studies published between 2019 and 2024, selecting 54 relevant studies from 4,280 initial citations. Our analysis reveals the use of diverse datasets, preprocessing techniques, and feature extraction methods, alongside a clear evolution from traditional machine learning to deep learning approaches, with substantial architectural diversity across convolutional, recurrent, transformer, and hybrid models. Performance varies considerably across dataset configurations, with models achieving higher accuracy on frequent code subsets compared to full label spaces. However, critical gaps persist: overreliance on single-language, single-institution datasets limits generalizability; difficulties in predicting rare codes remain unresolved; lack of model interpretability undermines clinical trust; and inconsistent evaluation protocols hinder meaningful comparison. To address these challenges, we propose a 5P evidence-grounded research agenda: Population Diversity, Performance Robustness, Prediction of Rare Codes, Provenance Transparency, and Practical Integration. These findings underscore AI’s potential to transform ICD coding while highlighting the need for standardized benchmarks, rigorous external validation, multilingual datasets, and explainable architectures to enable equitable and effective deployment in real-world healthcare systems.
Abdul Rehman Khalid, Haider Ali, Kounen Fathima et al.· Journal of medical systems· 0 citations
Automatic medical coding assigns ICD codes to clinical notes, but it remains challenging due to long documents, imbalanced label distributions, and diverse terms. These challenges are especially severe for rare codes, which have limited training instances and are easily confused with semantically similar labels. We introduce CoLa-ICD, a knowledge-enhanced framework for long-tail prediction. CoLa-ICD enriches ICD labels with external terms, models dependencies among related codes, and learns stronger alignment between label semantics and clinical evidence for long-tail prediction. Experiments show that CoLa-ICD improves long-tail prediction with larger gains in larger and sparser label spaces and achieves state-of-the-art performance in AUC, F1, and P@k. Our code is available at https://github.com/youwillbethebest/Cola-ICD.
Yihan Cheng, Veronica Liesaputra, A. Trotman· 0 citations
A dual-view approach that connects clinical practice with computational methods is presented, establishing a five-level competency scheme following Miller’s Pyramid and linking deductive, inductive, and abductive reasoning patterns to common medical goals and tasks.
Current LLMs do not achieve inter-rater reliability levels comparable to medical professionals in clinical information extraction from ENT documentation, suggesting they are best suited for initial extraction with human verification rather than autonomous operation.
L. Barrett, N. Joshi, A. S. North et al.· medRxiv· 0 citations