Assessing Revisit Risk in Emergency Department Patients: A Machine Learning Approach (Preprint)

<sec> <title>BACKGROUND</title> Overcrowded emergency rooms might degrade the quality-of-care service and overload the clinic staff. Assessing unscheduled return visits (URVs) to the emergency department (ED) is a quality assurance procedure to identify ED discharged patients with a high likelihood of bounce-back, to ensure patient safety, and ultimately to reduce medical costs by decreasing the frequency of URVs. The field of machine learning (ML) has evolved considerably in the past decades and many ML applications have been deployed in various contexts. </sec> <sec> <title>OBJECTIVE</title> This study aims to develop an ML-assisted framework that identifies high-risk patients that may revisit the ED within 72 hours after the initial visit. Furthermore, this study evaluates different ML models, feature sets, and feature encoding methods in order to build an effective prediction model. </sec> <sec> <title>METHODS</title> This study proposes a machine learning (ML)-assisted system that extracts the features from both structured and unstructured medical data to predict patients who likely revisit the ED, where the structured data includes patients’ Electronic Health Records (EHRs) and the unstructured data is their medical notes (SOAP). A 5-year dataset consisting of 184,687 ED visits with 324,111 historical EHRs and the associated medical notes was obtained from a tertiary medical center, Kaohsiung Veterans General Hospital (KVGH), in Taiwan for evaluating the proposed system. </sec> <sec> <title>RESULTS</title> The evaluation results indicate that incorporating CNN-based feature extraction from unstructured ED physician narrative notes, combined with structured vital signs and demographic data, significantly enhances predictive performance. The proposed approach achieves an AUROC of 0.705 and a recall of 0.718, demonstrating its effectiveness in predicting URVs. These findings highlight the potential of integrating structured and unstructured clinical data to improve predictive accuracy in this context. </sec> <sec> <title>CONCLUSIONS</title> The study demonstrates a ML-assisted framework may be applied as a decision support tool to assist ED clinicians in identifying revisiting patients, although the model performance may not be sufficient for clinic implementation. Given the improvement in AUROC, the proposed framework should be further explored to a workable decision support tool to pinpoint ED patients with a high risk of revisit and provide them appropriate and timely care. </sec> <sec> <title>CLINICALTRIAL</title> This study is a retrospective analysis of anonymized patient records and does not involve patient enrollment or interventional procedures. Therefore, it does not meet the criteria for clinical trial registration as outlined by ICMJE. </sec>