Advancing Geriatric Care: Machine Learning Algorithms and AI Applications for Predicting Dysphagia, Delirium, and Fall Risks in Elderly Patients

Abstract

Background Information: As the older population increases, dysphagia, delirium, and fall hazards considerably affect morbidity and death. Applications of machine learning (ML) and artificial intelligence (AI) offer potential improvements in early prediction and preventive tactics within geriatric care.
Objectives: The objective of this study is to create prediction models utilising machine learning algorithms to detect dysphagia, delirium, and fall risks in older patients, thereby enabling prompt treatments and enhancing patient outcomes.
Methods: Applied logistic regression, Random Forest, and CNN models both independently and in ensemble configurations to forecast risk variables for dysphagia, delirium, and falls, utilising clinical and sensor data to improve predictive accuracy in elderly patient care.
Results: The ensemble model attained superior predictive performance, exhibiting an accuracy of 93%, precision of 91%, recall of 89%, F1-score of 90%, and AUC-ROC of 92%, exceeding the performances of individual models.
Conclusion: Ensemble machine learning methods improve prediction accuracy for assessing risks in geriatrics, facilitating proactive management of dysphagia, delirium, and falls in aged care environments.