Random forest predictive modeling of prolonged hospital length of stay in elderly hip fracture patients

• DOI: 10.3389/fmed.2024.1362153
• Publication Date: 2024-05-17T04:28:46Z
• Authors: Hao Liu; Fei Xing; Jiabao Jiang; Zhao Chen; Zhou Xiang; Xin Duan
• Keywords: Hip Fracture; Lasso

Background In elderly individuals suffering from hip fractures, a prolonged hospital length of stay (PLOS) not only heightens the probability patient complications but also amplifies mortality risks. Yet, most fracture patients present compromised baseline health conditions. Additionally, PLOS leads to increased expenses for treatment and care, while diminishing turnover rates. This, in turn, jeopardizes prompt allocation beds urgent cases. Methods A retrospective study was carried out October 2021 November 2023 on 360 who underwent surgical at West China Hospital. The 75th percentile total cohort’s duration, which 12 days, used define (PLOS). cohort divided into training testing datasets with 70:30 split. predictive model developed using random forest algorithm, its performance validated compared Lasso regression model. Results Out patients, 103 (28.61%) experienced PLOS. Random Forest classification dataset, identifying 10 essential variables. achieved perfect set, an area under curve (AUC), balanced accuracy, Kappa value, F1 score 1.000. model’s assessed AUC 0.846, accuracy 0.7294, value 0.4325, 0.6061. Conclusion This aims develop prognostic predicting delayed discharge thereby improving this population. By utilizing machine learning models, clinicians can optimize medical resources devise effective rehabilitation strategies geriatric patients. method potentially improve bed rates, providing latent benefits healthcare system.