MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641079986 A) filed by Sona College Of Technology on June 29, 2026, for Multimodal Resnet-Retinaface System For Early Autism Detection And Severity Classification.
Inventors include Suji Priya J; Karolin R; and Harini T M.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: ABSTRACT The present invention relates to a multimodal artificial intelligence-based system for early detection and severity stratification of Autism Spectrum Disorder (ASD). The system integrates facial image analysis and behavioral questionnaire data to improve diagnostic accuracy and reduce subjectivity in traditional screening methods. Facial images are processed using a RetinaFace-based detection and alignment module followed by a ResNet-based deep convolutional neural network for feature extraction. In parallel, caregiver questionnaire responses are encoded and analyzed using machine learning models to generate behavioral risk scores. A multimodal fusion module combines visual and behavioral features using weighted or attention-based mechanisms to produce a unified Autism Risk Score. The system further classifies individuals into severity levels including normal, mild, moderate, and severe ASD, along with confidence scores. The invention enables non-invasive, scalable, and real-time screening suitable for clinical and remote healthcare applications, thereby supporting early intervention and improved developmental outcomes.
Disclaimer: Curated by HT Syndication.
