MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202621066843 A) filed by Vishwakarma Institute Of Technology on May 27, 2026, for A System For Early Detection Of Autism Spectrum Disorder.
Inventors include Cholke, Puja; Jagtap, Ayush Vinod; Khanchandani, Deep Mukesh; Kandpal, Om Lalit; Masram, Tanishq Vinod; and Pareek, Shyam Dhanraj.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: The present invention relates to a system for early detection of autism spectrum disorder (ASD) using deep learning techniques applied to facial images and video data. The system automates the diagnostic process, reducing reliance on manual and time-consuming clinical assessments and enabling faster and more accurate screening. Convolutional neural network architectures including InceptionV3, VGG-19, EfficientNet-B4, Xception, and NasNetMobile are used for feature extraction and classification of autistic and non-autistic traits. The system integrates Explainable AI using LIME to highlight facial regions influencing predictions, improving transparency and trust. A frontend interface developed using HTML and JavaScript enables user interaction, while a Flask-based backend manages processing and communication with the model. Deployment is achieved through AWS Lambda for real-time serverless inference, ensuring scalability. Google Cloud Datastore is used for secure data storage, providing an efficient, accessible, and interpretable platform supporting early ASD detection and timely intervention.
Disclaimer: Curated by HT Syndication.
