Computer Sciences and Information Technology
Title : | Design and Development of an Interactive System for Early Diagnosis of Autism Spectrum Disorder (ASD) using Deep Learning and Federated Learning Models for Rural Community in Chhattisgarh State |
Area of research : | Computer Sciences and Information Technology |
Focus area : | Artificial Intelligence in Healthcare |
Principal Investigator : | Dr. Chandrashekar jatoth, National Institute Of Technology (NIT) Raipur, Chhattisgarh |
Timeline Start Year : | 2023 |
Timeline End Year : | 2026 |
Contact info : | chandrashekar.jatoth@gmail.com |
Equipments : | Digital Voice Recorder -Black
Computational AI - System |
Details
Executive Summary : | Autism Spectrum Disorder (ASD) is a developmental disorder that describes certain challenges associated with communication (verbal and non-verbal), social skills, and repetitive behaviours. In India, rural communities face challenges in proper access to the resources for screening, diagnosing, treatment and services for individuals with ASD. The factors that contribute towards it include unawareness, low accessibility to health care professionals, and geographical and cultural aspects. These factors together lead to delay in screening or no diagnosis for ASD. Presently, the screening of ASD is done manually with the help of questionnaires provided by doctors, volunteers, and special educators, however these methods are time consuming and might miss crucial information required for the effective screening. To overcome the above-mentioned limitation of the current screening methods, we are proposing the project on developing and validating a mobile/web application based on video analysis and questionnaires based on gold-standard diagnostic instruments using deep learning feature extraction model. We collect home videos of children from rural communities to make an effective screening process of ASD. This application also helps in creating awareness of the ASD affected rural population and provide them the guidance to approach the proper channel for diagnosis confirmation and available therapies. The proposed application will be user-friendly, easy access, and cost-effective which includes questionnaires and videos by using latest technologies such as PHP REST API and Flutter for application development, deep learning and Federated Learning models for feature extraction and model building using machine learning algorithms to capture the screening information more effectively among the rural population. To the best of our knowledge this would be the first kind of study of developing and validating Interactive system for effective screening of ASD for rural communities and having large scope for commercialization in the global market along with the Societal benefit. In addition, Deep Learning and Federated Learning Models are applied to effectively analyses the audio and video captured during screening sessions to increase the sensitivity, specificity and accuracy rates. |
Total Budget (INR): | 27,66,710 |
Organizations involved