MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641076139 A) filed by Vellore Institute Of Technology on June 19, 2026, for Multimodal Parkinson'S Disease Screening System With Adaptive Fuzzy Inference And Quantized Classification.
Inventors include Chandra Segar; and Vinothini S.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: ABSTRACT MULTIMODAL PARKINSON'S DISEASE SCREENING SYSTEM WITH ADAPTIVE FUZZY INFERENCE AND QUANTIZED CLASSIFICATION A computer-implemented method (100) for early-stage Parkinson's Disease screening comprises acquiring (102) multimodal digital biomarker data comprising voice data, gait data, and handwriting data; preprocessing (104) each modality to generate preprocessed modality data; extracting (106) feature vectors comprising a voice feature vector, a gait feature vector, and a handwriting feature vector using modality-specific deep learning encoders; computing (108) reliability indicators for each modality based on signal quality metrics; inputting (110) the reliability indicators into a gradient-optimized fuzzy inference engine comprising learnable membership functions to generate dynamic modality weights, wherein the learnable membership functions are differentiable and updated via gradient descent; fusing (112) the feature vectors using the dynamic modality weights to generate a Trimodal Feature Vector; classifying (114) the Trimodal Feature Vector using a quantized machine learning classifier to generate a Parkinson's Disease probability score; and generating (116) an explainable output comprising feature attribution information. (Fig. 1)
Disclaimer: Curated by HT Syndication.
