Enhancing Delirium Prediction and Prevention in Elderly Patients Through Machine Learning-Based Analysis.

Objective: This study aimed to identify predictors of delirium within 24 hours of admission in elderly patients using machine learning (ML) models and evaluate their performance.

Methods: This prospective cohort study was conducted among patients aged 65 years and older admitted to the general medical unit of Sultan Qaboos University Hospital, Muscat, Oman, from January 2022 to May 2023. Clinical and demographic data were collected and analysed using 4 ML models: logistic regression, random forest, gradient boosting and support vector machine. Model performance was evaluated using accuracy, precision, recall, F1 score and area under the curve-receiver operating characteristic (AUC-ROC) metrics. Cross-validation was performed to assess model robustness and feature importance analysis was conducted to identify key predictors.

Results: A total of 327 patients were included in this study. The random forest model demonstrated the best performance, achieving an accuracy of 96.9%, an F1 score of 97.2%, and an AUC-ROC of 98.4%. Cross-validation confirmed the model's stability. Feature importance analysis identified acute kidney injury, respiratory failure, dementia, stroke and decompensated heart failure as the most influential predictors of delirium.

Conclusion: ML models, particularly the random forest model, exhibited strong predictive performance in identifying patients at risk of delirium within 24 hours of admission. These findings support the potential of ML in enhancing early delirium detection and guiding targeted preventive strategies. Future research should focus on external validation to confirm the model's applicability across different healthcare settings.