JMIR Publications

ABSTRACT

Background:

Prolonged hospital stays can lead to inefficiencies in healthcare delivery and unnecessary consumption of medical resources.

Objective:

This study aimed to identify key clinical variances associated with prolonged length of stay (PLOS) in clinical pathways using a machine learning model trained on real-world data from the ePath system.

Methods:

We analyzed data from 480 patients with lung cancer (mean age 68.3 ± 11.2 years, 51% men) who underwent video-assisted thoracoscopic surgery (VATS) at a university hospital between 2019 and 2023. PLOS was defined as a hospital stay exceeding nine days post-VATS. The variables collected between admission and postoperative day 4 were examined, and those significantly associated with PLOS in the univariate analysis were selected as predictors. Predictive models were developed using sparse linear regression methods (Lasso, ridge, elastic net) and decision-tree ensembles (random forest, XGBoost). The data were divided into derivation (earlier study period) and testing (later period) cohorts for temporal validation. The model performance was assessed using the area under the receiver operating characteristic curve (AUROC), Brier score, and calibration plots. Counterfactual analysis was used to identify key clinical factors influencing PLOS.

Results:

A three-dimensional heatmap illustrated the temporal relationships between clinical factors and PLOS based on patient demographics, comorbidities, functional status, surgical details, care processes, medications, and variances recorded on postoperative day 4. The ridge regression model achieved the best performance, with AUROC and Brier scores of 0.84 and 0.16, respectively, in the derivation cohort and 0.82 and 0.17, respectively, in the test cohort. The six key variables that increased PLOS risk included abnormal respiratory sounds, postoperative fever, arrhythmia, impaired ambulation, complications after drain removal, and pulmonary air leaks.

Conclusions:

A machine learning-based model using ePath data effectively identified critical variances in the clinical pathways associated with PLOS. This automated tool may enhance clinical decision-making and improve patient management.