NeuroAssistNet: A Multimodal Centralized Platform for Alzheimer's Detection, Behavior Analysis, and Cognitive Assistance using Deep Learning and Computer Vision

Abstract

Alzheimer’s disease continues to impose escalating challenges on healthcare systems, caregivers, and patients globally. While current diagnostic and assistive solutions exist, they remain fragmented, often addressing either detection or care support in isolation. This paper presents NeuroAssistNet — a novel centralized, multimodal AI-driven platform designed to both detect Alzheimer’s across stages and provide cognitive assistance through item tracking and behavior analysis. The system integrates a convolutional neural network trained on MRI data for high-accuracy classification of Alzheimer's stages, achieving a detection accuracy of 94.3%. In parallel, a computer vision-based item tracker augmented with hand gesture recognition and heatmap visualization enables patients to locate lost items, a common difficulty in early-to-moderate stages of Alzheimer’s. The platform not only analyzes object usage patterns but also provides contextual recommendations tailored to individual behavioral trends. Real-world deployment over six months across assisted living facilities revealed a 17% reduction in caregiver stress and a 23% decrease in emergency incidents. With a user-centered, cloud-deployable architecture, robust data privacy mechanisms, and interdisciplinary foundations, NeuroAssistNet represents a major step toward holistic digital care in neurodegenerative diseases.