Assessing artificial intelligence ability in predicting hospitalization duration for pleural empyema patients managed with uniportal video‑assisted thoracoscopic surgery: a retrospective observational study
Files
Date
2025-05-09
Authors
Issa Alnajjar
Baraa Alshakarnah
Tasneem AbuShaikha
Tareq Jarrar
Abed Al‑Raheem Ozrail
Yousef Abu Asbeh
Journal Title
Journal ISSN
Volume Title
Publisher
Springer Nature (BioMed Central)
Abstract
This retrospective observational study evaluated the ability of artificial intelligence and machine-learning models to predict hospital length of stay in patients with pleural empyema managed with uniportal video-assisted thoracoscopic surgery. Data from 56 patients were analyzed using two predictive models: a Random Forest Regressor and a literature-informed weighted model. Both models demonstrated poor predictive accuracy, with mean absolute errors exceeding four days and negative R-squared values. These findings highlight the challenges of length-of-stay prediction in pleural empyema due to significant clinical variability and current limitations of AI-based models. Future multicenter studies with larger datasets and more advanced modeling approaches are recommended to improve predictive performance.
Description
This retrospective observational study evaluates the performance of artificial intelligence models in predicting hospital length of stay among pleural empyema patients undergoing uniportal VATS. The findings demonstrate current limitations of AI-based predictive tools and emphasize the need for larger multicenter datasets and improved machine-learning strategies to enhance clinical applicability.
Keywords
Citation
Alnajjar I, Alshakarnah B, AbuShaikha T, Jarrar T, Ozrail AA, Abu Asbeh Y. Assessing artificial intelligence ability in predicting hospitalization duration for pleural empyema patients managed with uniportal video-assisted thoracoscopic surgery: a retrospective observational study. BMC Surg. 2025;25:218. doi:10.1186/s12893-025-02959-w.