Machine Learning-Supported Efficient VTE Risk Assessment using Routinely Collected Electronic Health Record Data
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Venous thromboembolism (VTE) is a leading cause of preventable inpatient mortality, while the real-world performance of mandated risk assessment and the potential for automating using electronic health record (EHR) data remain unclear. We analysed 577,904 admissions and 726,896 VTE assessment forms across five NHS hospitals between 2015 and 2025 to evaluate assessment completion, concordance with structured EHR data, clinical validity, and feasibility of EHR-based automation assisted by machine learning. Overall completion was high (96.7%), and timely completion improved from 47.4% in 2015 to 90.5% in 2024. Agreement between forms and EHR data was good for common risk factors, but low-prevalence variables were often under-documented in the forms. Despite these discrepancies, form-derived thrombosis risk was associated with increased VTE incidence (OR 3.31, 95% CI 2.81–3.90). Machine learning models using first-14-hour EHR data achieved discrimination comparable to clinician-recorded variables (AUROC 0.709 vs 0.704), supporting real-time EHR-integrated assessment pre-population and decision support.
Author summary
We studied whether information already stored in hospital electronic health records could make required venous thromboembolism (VTE) risk assessments quicker and more reliable. VTE refers to potentially serious blood clots in the deep veins or lungs. We examined more than half a million admissions across five NHS hospitals over ten years. Most assessments were eventually completed, but many were not finished within the recommended 14-hour window. Information entered manually by clinicians often agreed with existing electronic records for common risk factors, but uncommon factors and bleeding risks were missed more often. Even with these documentation differences, the assessments identified patients who were more likely to develop VTE. We also tested machine-learning models using information available during the first 14 hours of admission. These models performed about as well as models based on clinician-completed forms. Our findings suggest that hospital systems could pre-fill parts of the assessment using data already recorded, while leaving clinicians to verify the information and make the final decision. This approach could reduce repetitive manual data entry, improve timely completion, and help clinicians identify patients who may benefit from preventive treatment.
