Machine learning models in enhancing prediction of health-related indices among older adults: A scoping review

Abstract Background Early prediction of health indices based on these statuses is crucial in enhancing older people's status and achieving an efficient preventive strategy. A predictive analysis will be essential in better predicting health-related indices. Objective This scoping review aims to investigate machine learning models in predicting health-related indices among older adults. Eligibility criteria The articles were written in English, and various health-related indices measuring physical, mental, and social factors among older adults were included. Sources of evidence This scoping review was conducted according to the PRISMA extension for scoping reviews. Any publications that adhered to eligibility criteria and were indexed in PubMed, Scopus, and Web of Sciences were leveraged. Data charting and analysis A data extraction form, including general and specific data fields, was leveraged to elicit data from articles. The data on leveraging machine learning models in health-related indices among older adults were analyzed using narrative synthesis. Results After screening 355 articles from scientific databases, 14 were considered to be included in this current review. Random Forest performed superiorly to other models for successful aging. Artificial neural networks obtained highly predictive insights into the quality of life. QUEST performed well in terms of life satisfaction. SVM obtained satisfactory performance in predicting health-related behaviors. XG-Boost had favorable predictive performance efficiency in healthy aging. Health status is indicated by a relatively good predictive performance based on ANNs. Conclusion This review demonstrated that various machine learning algorithms and data types significantly impact predictive ability and preventive strategies in clinical environments.