Developing an intelligent prediction system for successful aging based on artificial neural networks.

Area of Science:

• Gerontology
• Artificial Intelligence in Healthcare
• Biomedical Informatics

Background:

• Growing prevalence of disabilities in the elderly necessitates focused attention on this demographic.
• Limited research comprehensively addresses physical, mental, and disorder-related factors impacting elderly quality of life.
• Social Affect (SA) is influenced by multiple factors crucial for understanding elderly well-being.

Purpose of the Study:

• To develop an intelligent system for predicting Social Affect (SA) in the elderly.
• To investigate factors influencing elderly life quality using Artificial Neural Network (ANN) algorithms.
• To promote improved health outcomes and decision-making for the elderly population.

Main Methods:

• Analysis of 1156 Social Affect (SA) and non-SA cases.
• Application of statistical feature reduction to identify key predictive factors.
• Construction and evaluation of ANN models with varying hidden layer neurons (5, 10, 15, 20) using FF-BP algorithm.

Main Results:

• Identified 25 factors statistically correlated with Social Affect (SA) at P < 0.05.
• The best performing ANN configuration (25-15-1) achieved high accuracy: 0.92 (train), 0.86 (test), and 0.87 (validation).
• The Feed-Forward Back-Propagation (FF-BP) algorithm demonstrated superior performance in SA prediction.

Conclusions:

• The developed intelligent system, a Clinical Decision Support System (CDSS), is crucial for predicting SA.
• Informs geriatrics and healthcare policymakers for effective decision-making.
• Aids in promoting the quality of life for the elderly population.