Intellectual Property India Publishes Patent Application for A Multi-Stage Automated Medical Diagnostic System For Neurodegenerative Disease Classification Using Hardware-Accelerated Recurrent Architectures

MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641111374 A) filed by Sr University on September 16, 2026, for A Multi-Stage Automated Medical Diagnostic System For Neurodegenerative Disease Classification Using Hardware-Accelerated Recurrent Architectures.

Inventors include Krishna Kishore Maaram; and Dr. Shanker Chandre.

The application for the patent was published on September 25, 2026, under issue no. 39/2026.

Abstract: A MULTI-STAGE AUTOMATED MEDICAL DIAGNOSTIC SYSTEM FOR NEURODEGENERATIVE DISEASE CLASSIFICATION USING HARDWARE- ACCELERATED RECURRENT ARCHITECTURES Alzheimer’s Disease (AD) is a progressive neurodegenerative disorder affecting cognition, memory, and daily functioning. Early and accurate classification of Alzheimer’s Disease (AD), Mild Cognitive Impairment (MCI), and Cognitively Normal (CN) conditions is essential for timely intervention and patient care. The present research proposes a three-stage deep learning framework incorporating optimized feature selection, advanced activation functions, and multimodal imaging. In the first stage, Adaptive Weber Distribution–Flower Pollination Algorithm (AWD-FPA) selects relevant features, which are classified using a Gated Recurrent Unit (GRU) with Rectified Linear Unit (ReLU) activation, achieving 97.80% accuracy for AD vs CN and 97.21% for multiclass classification. The second stage employs a Stacked Long Short-Term Memory (SLSTM) network with Sech activation and Crisscross Strategy based Parrot Optimization Algorithm (CSPOA) for hyperparameter optimization, achieving over 99% accuracy in binary classifications and 98.25% in multiclass classification using multimodal PET and MRI data. The third stage integrates functional and structural imaging using Aquila Optimizer–Arithmetic Optimization Algorithm (AO-AOA) feature selection and Stacked GRU with Leaky ReLU, achieving accuracies of 96.57%, 97.96%, and 95.61% for MCI vs CN, AD vs CN, and MCI vs AD, respectively. The proposed framework demonstrates improved accuracy, robustness, and computational efficiency for AD classification.

Disclaimer: Curated by HT Syndication.