Smart Assistant For Alzhimer’s Patients

Abstract

Alzheimer’s disease (AD) presents growing challenges in patient care, safety, and daily functioning. This paper introduces a secure, IoT-based wearable assistant system designed to support individuals with mild to moderate AD. The system integrates facial recognition via a Convolutional Neural Network (CNN) to distinguish between family and non-family members, enhancing memory recall and social interaction. It employs steganography to securely embed identity and relational data within images, ensuring privacy. Additional features include GPS-based location tracking with geofence alerts, voice communication through Google Assistant, and psychological status monitoring via conversational analysis. To further enhance patient safety and autonomy, the system incorporates a Panic Assistant that enables immediate alerts in emergency situations, and a Daily Routine Voice Prompter that delivers scheduled voice reminders for essential activities such as medication, meals, hydration, and hygiene. These additions aim to reduce anxiety, improve adherence to routines, and foster independence. The prototype, implemented on a Raspberry Pi 3 platform, demonstrates high accuracy in facial classification, secure data transmission, and responsive voice interaction, offering a comprehensive and compassionate solution for AD patient assistance.