MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202621043586 A) filed by Prof. Nilam Thorat; Pranav Dawange; Bhavik Chhoriya; Ronit Jain; and Rudrasingh Dikshit on April 06, 2026, for Computer-Implemented Method For Skin Lesion Risk Assessment Using Explainability-Driven Feature Fusion.
Inventors include Prof. Nilam Thorat; Pranav Dawange; Bhavik Chhoriya; Ronit Jain; and Rudrasingh Dikshit.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: The present invention relates to a computer-implemented method for assessing risk associated with skin lesions using artificial intelligence. The method comprises processing an input image using a trained machine learning model to generate a predicted lesion class and a classification confidence score. An explainable artificial intelligence technique is applied to generate a spatial activation map, from which an attention score is computed as a numerical representation of activation intensity. The classification confidence score, predicted lesion class, attention score, and patient metadata are combined to form a feature vector, which is processed using a secondary machine learning model to generate a risk classification and an associated risk score. A deterministic override rule is applied to assign a highest-risk classification for predefined critical lesion classes irrespective of the output of the secondary model. In an embodiment, the secondary model is trained using labels generated through a deterministic function based on outputs of the primary model.
Disclaimer: Curated by HT Syndication.
