Early detection of self-care impairments in children with disabilities using an enhanced SE network optimized by ISCO algorithm

Children with disabilities frequently encounter considerable obstacles in acquiring self-care skills, which are vigorous for developing their independence and overall quality of life. The early detection of self-care deficits is important for prompt intervention and assistance. Nevertheless, current valuation techniques predominantly depend on manual evaluations, which can be subjective, labor-intensive, and often inadequate in recognizing subtle deficits. The purpose of this study is to create an artificial intelligence (AI)-enhanced self-care prediction system for children with disabilities, using machine learning to accurately and early identify self-care deficits. An innovative self-care prediction methodology is introduced based on Squeeze and Excitation Networks, refined through a modified metaheuristic algorithm known as the Improved Single Candidate Optimization Algorithm. This approach employs an extensive dataset to train the neural network, allowing it to discern intricate patterns and correlations. The experimental findings illustrate the efficacy of the proposed methodology, surpassing current techniques in predicting self-care deficits.