Audcare: Advanced Care for Children with Autism Spectrum Disorder Through a Mobile Application for Predicting and Managing Meltdowns Using CNN-LSTM

Children with Autism Spectrum Disorder (ASD) often experience unpredictable meltdowns due to complex behavioral and environmental factors. A lack of accessible information on prevention and management strategies, along with a shortage of specialized healthcare professionals, complicates support for families. This study develops a mobile application that predicts meltdowns, featuring a recommendation engine for prevention strategies and a location-based service for nearby healthcare providers. The app utilizes the CNN-LSTM model (MoViNets) to classify tantrums and generate tailored strategies. The MoViNets model achieved training and validation accuracies of 82.03 % and 87.30%, respectively, with an overall classification accuracy of 90%. Evaluations from therapy centers indicated strong functionality (mean of 4.33), usability (4.19), reliability (4.14), and security (4.40), resulting in an overall mean rating of 4.27, signifying a well-rounded system in terms of functionality and user experience.