Accurate clinical coding is essential for healthcare documentation, billing, and decision-making. While prior work shows that off-the-shelf LLMs struggle with this task, evaluations based on exact match metrics often overlook errors where predicted codes are hierarchically close but incorrect. Our analysis reveals that such hierarchical misalignments account for a substantial portion of LLM failures. We show that lightweight interventions, including prompt engineering and small-scale fine-tuning, can improve accuracy without the computational overhead of search-based methods. To address hierarchically near-miss errors, we introduce clinical code verification as both a standalone task and a pipeline component. To mitigate the limitations in existing datasets, such as incomplete evidence and inpatient bias in MIMIC, we release an expert double-annotated benchmark of outpatient clinical notes with ICD-10 codes. Our results highlight verification as an effective and reliable step toward improving LLM-based medical coding.

This study presents a novel two-stage Retrieve-Rank system for automated ICD-10-CM medical coding, comparing its performance against a Vanilla Large Language Model (LLM) approach. Evaluating both systems on a dataset of 100 single-term medical conditions, the Retrieve-Rank system achieved 100% accuracy in predicting correct ICD-10-CM codes, significantly outperforming the Vanilla LLM (GPT-3.5-turbo), which achieved only 6% accuracy. Our analysis demonstrates the Retrieve-Rank system's superior precision in handling various medical terms across different specialties. While these results are promising, we acknowledge the limitations of using simplified inputs and the need for further testing on more complex, realistic medical cases. This research contributes to the ongoing effort to improve the efficiency and accuracy of medical coding, highlighting the importance of retrieval-based approaches.

The increasing use of tools and solutions based on Large Language Models (LLMs) for various tasks in the medical domain has become a prominent trend. Their use in this highly critical and sensitive domain has thus raised important questions about their robustness, especially in response to variations in input, and the reliability of the generated outputs. This study addresses these questions by constructing a textual dataset based on the ICD-10-CM code descriptions, widely used in US hospitals and containing many clinical terms, and their easily reproducible rephrasing. We then benchmarked existing embedding models, either generalist or specialized in the clinical domain, in a semantic search task where the goal was to correctly match the rephrased text to the original description. Our results showed that generalist models performed better than clinical models, suggesting that existing clinical specialized models are more sensitive to small changes in input that confuse them. The highlighted problem of specialized models may be due to the fact that they have not been trained on sufficient data, and in particular on datasets that are not diverse enough to have a reliable global language understanding, which is still necessary for accurate handling of medical documents.

Identifying suicidality including suicidal ideation, attempts, and risk factors in electronic health record data in clinical notes is difficult. A major difficulty is the lack of training samples given the small number of true positive instances among the increasingly large number of patients being screened. This paper describes a novel methodology that identifies suicidality in clinical notes by addressing this data sparsity issue through zero-shot learning. U.S. Veterans Affairs clinical notes served as data. The training dataset label was determined using diagnostic codes of suicide attempt and self-harm. A base string associated with the target label of suicidality was used to provide auxiliary information by narrowing the positive training cases to those containing the base string. A deep neural network was trained by mapping the training documents contents to a semantic space. For comparison, we trained another deep neural network using the identical training dataset labels and bag-of-words features. The zero shot learning model outperformed the baseline model in terms of AUC, sensitivity, specificity, and positive predictive value at multiple probability thresholds. In applying a 0.90 probability threshold, the methodology identified notes not associated with a relevant ICD 10 CM code that documented suicidality, with 94 percent accuracy. This new method can effectively identify suicidality without requiring manual annotation.