Background: Medical coding is the process that converts clinical documentation into standard medical codes. Codes are used for several key purposes in a hospital (eg, insurance reimbursement and performance analysis); therefore, their optimization is crucial. With the rapid growth of natural language processing technologies, several solutions based on artificial intelligence have been proposed to aid in medical coding by automatically suggesting relevant codes for clinical documents. However, their effectiveness is still limited to simple cases, and it is not yet clear how much value they can bring in improving coding efficiency and accuracy. Objective: This study aimed to bring more efficiency to the coding process to improve the selection of codes by medical coders. To achieve this, we developed an innovative multimodal machine learning–based solution that, instead of predicting codes, detects the degree of coding complexity before coding is performed. The notion of coding complexity was used to better dispatch work among medical coders to eventually minimize errors and improve throughput. Methods: To train and evaluate our approach, we collected 2060 cases rated by coders in terms of coding complexity from 1 (simplest) to 4 (most complex). We asked 2 expert coders to rate 3.01% (62/2060) of the cases as the gold standard. The agreements between experts were used as benchmarks for model evaluation. A case contains both clinical text and patient metadata from the hospital electronic health record. We extracted both text features and metadata features, then concatenated and fed them into several machine learning models. Finally, we selected 2 models. The first used cross-validated training on 1751 cases and testing on 309 cases aiming to assess the predictive power of the proposed approach and its generalizability. The second model was trained on 1998 cases and tested on the gold standard to validate the best model performance against human benchmarks. Results: Our first model achieved a macro–F1-score of 0.51 and an accuracy of 0.59 on classifying the 4-scale complexity. The model distinguished well between the simple (combined complexity 1-2) and complex (combined complexity 3-4) cases with a macro–F1-score of 0.65 and an accuracy of 0.71. Our second model achieved 61% agreement with experts’ ratings and a macro–F1-score of 0.62 on the gold standard, whereas the 2 experts had a 66% (41/62) agreement ratio with a macro–F1-score of 0.67. Conclusions: We propose a multimodal machine learning approach that leverages information from both clinical text and patient metadata to predict the complexity of coding a case in the precoding phase. By integrating this model into the hospital coding system, distribution of cases among coders can be done automatically with performance comparable with that of human expert coders, thus improving coding efficiency and accuracy at scale.

Background: The aim of the present study was to apply machine learning (ML) techniques to define clusters relating patient demographics, compliance, and outcome variables in colorectal enhanced recovery after surgery (ERAS) patients and improve data-driven, predictive decision-making. Methods: To uncover inherent patient subgroups from the data without pre-defined labels, the unsupervised K-means clustering algorithm was utilized. This technique was selected for its effectiveness in partitioning patients into distinct groups by iteratively assigning them to the nearest cluster mean, thereby minimizing within-cluster variance across key variables. The top five recovery goals and the top 10 clinical outcome variables were defined based on clinical considerations (incidence and importance). In a second step, the cluster transition was traced by monitoring the transitions between clusters from demographic through compliance to outcome variables. Results: A total of 1381 patients were available for final analysis, revealing three clusters (low risk, n = 490, 36%; intermediate risk, n = 157, 11%; and high risk, n = 734, 53%) for demographic, two clusters (high compliance, n = 1011, 73%, and low compliance n = 370, 27%) for perioperative, and two clusters (good and poor outcomes) for the top five recovery goals and the top 10 clinical outcomes, respectively. The cluster transition for the top five recovery goals and the top 10 clinical outcomes revealed that most patients (488/490, 99.6%) of the low-risk demographic cluster had high perioperative compliance, and over 90% of them had favorable functional and clinical outcomes. Of the 2/3 of intermediate risk patients who had poor perioperative compliance, over 40% had a poor functional recovery, whereas 83% had good clinical outcomes. Of the high-risk demographic group, 100% (734/734) had low perioperative compliance, and over 40% of them had poor functional recovery. Conclusions: This ML-based analysis of demographic, compliance, and recovery clusters and associated cluster transition allowed us to identify patient clusters as a first step to tailored ERAS protocols aiming to improve compliance and outcomes.

Clinical notes contain valuable information for research and monitoring quality of care. Named Entity Recognition (NER) is the process for identifying relevant pieces of information such as diagnoses, treatments, side effects, etc., and bring them to a more structured form. Although recent advancements in deep learning have facilitated automated recognition, particularly in English, NER can still be challenging due to limited specialized training data. This exacerbated in hospital settings where annotations are costly to obtain without appropriate incentives and often dependent on local specificities. In this work, we study whether this annotation process can be effectively accelerated by combining two practical strategies. First, we convert usually passive annotation tasks into a proactive contest to motivate human annotators in performing a task often considered tedious and time-consuming. Second, we provide pre-annotations for the participants to evaluate how recall and precision of the pre-annotations can boost or deteriorate annotation performance. We applied both strategies to a text de-identification task on French clinical notes and discharge summaries at a large Swiss university hospital. Our results show that proactive contest and average quality pre-annotations can significantly speed up annotation time and increase annotation quality, enabling us to develop a text de-identification model for French clinical notes with high performance (F1 score 0.94).

Background: Computed tomography (CT)-based sarcopenia is a promising predictor of postoperative complications and recovery. However, studies on the longitudinal evolution of skeletal muscle markers are lacking and findings regarding its correlation with survival are still not clear. Methods: We conducted a retrospective single-center cohort study of consecutive patients undergoing curative-intent colon cancer surgery. Skeletal muscle area (SMA), skeletal muscle index (SMI), skeletal muscle radiation attenuation (SMRA), and intermuscular adipose tissue (IMAT) area and index (IMATI) were measured on a single axial CT slice at the third lumbar vertebral level before surgery and at 6, 12, and 24 months after. Descriptive statistics were used to report the evolution over time of CT-based sarcopenia markers. Their correlation with overall survival was analyzed using Cox’s proportional hazards regression analysis. Results: The final cohort included 102 patients (65.7% males) with a mean age of 66 ± 13 years. Eighty-five (86.7%) patients were alive at 24 months, and forty-five (45.9%) underwent a CT scan at all time points. CT-based sarcopenia markers remained statistically stable over 2 years. Age (HR 1.07, 95% CI 1.00–1.14) and ASA score (HR 2.4, 95% CI 1.00–5.7) were negative independent predictive factors. Patients with larger differences ($\mathrm{\Delta }$) of IMAT area and IMATI at 12 months, HR 0.79 (95% CI 0.67–0.93) and 0.49 (95% CI 0.30–0.80), respectively, had a lower mortality. Conclusions: CT-based markers of skeletal muscle quantity (SMA, SMI) and quality (IMAT area, IMATI) remained statistically stable after curative-intent colon cancer surgery. No preoperative CT-based sarcopenia markers were predictive of overall survival. Larger cohorts are needed to generalize these initial findings.

Background: Large language models (LLMs) are increasingly being deployed in health care, but their use and deployment in many real-world hospital environments pose significant challenges and concerns. In particular, state-of-the-art commercial models store or process data externally, which is often in conflict with ensuring patient data protection. At the same time, using LLMs locally is limited by the lack of available computing infrastructure. Small open-source LLMs that do not require substantial computing resources could offer a practical way to resolve these tensions, but their medical utility in real-world local contexts, especially in non-English languages, has not been sufficiently evaluated. Objective: This study aimed to evaluate the feasibility of small, locally deployable open-source LLMs for clinically relevant tasks in a resource-constrained hospital setting and to propose a reproducible framework for institution-specific evaluation before deployment. Methods: We evaluated 6 open-source LLMs ranging from 8B to 24B parameters (from the Mistral, Phi4, Falcon3, Llama3.1, and Meditron3 families) in a zero-shot setting across 7 tasks covering 4 clinical use cases: information extraction, medical text translation, text generation, and clinical decision support. We used deidentified French clinical data from a Swiss tertiary hospital, including discharge letters, clinical notes, and structured electronic health records. Performance was assessed using task-specific metrics, such as precision, recall, F1-score, embedding-based semantic similarity, recall-oriented understudy for gisting evaluation (ROUGE) score, readability indices, and human review by clinicians. Results: Model performance varied substantially between tasks. In the simplest retrieval task (needle-in-the-haystack), several models performed strongly, with Llama3.1 achieving an F1-score of 99.81% and Mistral-small achieving 99.71%. In contrast, performance was poor in more complex tasks. For detecting protected health information, the best-performing LLMs achieved only modest overall macro–F1-scores (0.33-0.34), substantially below a fine-tuned Robustly Optimized BERT Pretraining Approach (RoBERTa) baseline (0.94). In the task of extracting immune-related adverse events from discharge notes, the highest overall macro–F1-score was 0.35 with Phi4. For medical text translation, Phi4 ranked highest in embedding-based evaluation, whereas Meditron3-Phi4 performed the worst, with clinician reviews identifying hallucinations in 55% of its outputs. In the task of summarizing discharge letters, quality was low across all models, with the best penalized ROUGE score reaching only 0.169 with Llama3.1. In the tasks of generating patient-friendly discharge note summaries and clinical decision support, clinician ratings generally ranged from dissatisfied to neutral, and no model achieved consistently satisfactory performance. Conclusions: Small open-source LLMs appear feasible for simple retrieval-oriented tasks in local hospital deployments but are currently inadequate for more complex applications, such as clinical decision support, deidentification, extraction of adverse events, and medical summarization. These findings highlight the importance of locally grounded evaluation tailored to specific use cases and the need for robust institutional evaluation frameworks to ensure safe and reliable deployment.

BACKGROUND Large Language Models (LLMs) offer promise for healthcare but face challenges of scale, privacy, and limited evidence in non-English settings. Smaller, locally deployable LLMs remain underexplored. OBJECTIVE To assess the feasibility of small open-source LLMs (1–24B parameters) in French-language clinical tasks and provide a reproducible hospital-based evaluation framework. METHODS Six state-of-the-art small LLMs from the Mistral, Phi-4, Llama-3.1, Meditron-3, Falcon 3 model families were tested in a zero-shot setting on de-identified discharge letters across seven use cases, including information extraction, translation, summarization, and clinical decision support. Performance was measured with F1 scores, readability indices, embedding similarity, and structured clinician reviews. RESULTS The models achieved high recall in simple retrieval tasks (up to 99.6%) but showed poor performance in detection of protected health information, adverse-event extraction, summarization, and decision support. Translation quality varied, with general-purpose models outperforming medical-focused models. CONCLUSIONS In localized resource-constrained deployments, small LLMs are suitable for basic tasks but insufficient for complex reasoning or clinical decision-making. Our framework supports context-specific evaluation for safe adoption in hospitals.