Memory Vault: A Dual Cognitive Reinforcement System for Dementia Care

Abstract

Memory-related disorders like Alzheimer’s disease or dementia can also affect the patient’s ability to recognize the person or to remember the patient’s previous interactions with the particular person or to recall daily tasks. This paper aims to develop a cognitive assistance system prototype to help dementia patients with the reinforcement of both retrospective and prospective memory. The proposed system’s architecture includes a face recognition module with an artificial intelligence- assisted memory retrieval system, as well as a reminder system with a simple to-do task manager. The proposed system’s face recognition module allows the patient to recognize known persons by recognizing the images of the faces of the persons they know. Additionally, the patient can recall information about the identity of the person with whom the patient is interacting. Moreover, the artificial intelligence-assisted memory retrieval system allows the patient to store personal memories. Therefore, the patient can retrieve relevant information with the help of vector similarity techniques to generate a meaningful output. Lastly, the reminder system with a simple to-do task manager allows the patient to recall important information about the intake of medicine with the help of scheduled reminders. Moreover, the proposed system’s architecture utilizes the FastAPI and Flask services to efficiently execute the functions of recognition, retrieval, and reminders. Additionally, the experimental evaluation of the system’s performance with the help of the proposed system’s prototype with a dataset of five registered individuals showed a face recognition accuracy of 90%, a memory retrieval relevance of 80%, and a response time of three seconds. This proves the effectiveness of the proposed system’s architecture as a cognitive assistance system for dementia patients or those suffering from similar memory-related disorders.