Analyzing Habitual Patterns and Behavioral Discrepancies in Ambient Assisted Living: An LSTM-Based Predictive Model

Area of Science:

• Gerontology and Health Informatics
• Artificial Intelligence in Healthcare
• Sensor Technology and Data Analytics

Background:

• The aging global population necessitates advanced solutions for elderly safety and independent living.
• Ambient Assisted Living (AAL) technologies, including Passive Infrared (PIR) motion sensors, are crucial for monitoring daily activities and detecting potential health issues.
• Existing PIR sensor systems have limitations that can be overcome to improve occupancy prediction accuracy.

Purpose of the Study:

• To develop an unobtrusive in-home monitoring framework utilizing PIR sensors and Long Short-Term Memory (LSTM) networks.
• To enhance the accuracy of occupancy prediction by analyzing habitual activity patterns in elderly individuals.
• To identify deviations from normal routines that may signal health concerns or emergencies.

Main Methods:

• Deployment of strategically placed PIR sensors in a single-resident apartment.
• Application of Long Short-Term Memory (LSTM) networks for predictive modeling of elderly occupancy behavior.
• Forecasting occupancy over time segments of 15, 30, 45, and 60 minutes.
• Development and utilization of a habitual disaccordance metric to quantify pattern deviations.

Main Results:

• The LSTM-based predictive model achieved 93.0% training accuracy and 91.4% validation accuracy on 15-minute segments.
• The model demonstrated strong predictive performance for short-term occupancy forecasting.
• The habitual disaccordance metric effectively identified changes in daily routines and potential emergency situations.

Conclusions:

• Motion-sensor systems integrated with LSTM models provide a valuable tool for analyzing habitual patterns and behavioral discrepancies in elderly individuals.
• The framework can effectively detect routine changes or emergencies, thereby improving elderly care and safety.
• This technology supports independent living for seniors while ensuring timely intervention and reducing the risk of adverse health outcomes.