Analysis of GRU-Based Platform to Prevent the Accident from Lonely Death

Due to recent advances in medicine, industry, and science, the number of diseases that humans can overcome has gradually increased. As a result, rapid population aging has occurred globally. In addition, as the aging population increases, the issue of lonely death is also emerging as a social concern. Recently, lonely deaths have been occurring across all age groups, not just among the elderly population. In particular, cases of long-term neglect are gradually increasing. As a result, there is a growing need for technology that can continuously detect the movement of individuals living alone and respond immediately in emergency situations. Therefore, in this paper, we propose a platform to prevent lonely deaths by predicting user movement patterns. The proposed system's core technology uses AI (Artificial Intelligence) to predict user movement patterns and make inferences based on the predicted values. Due to the diverse movement patterns in each household, we have developed a time series prediction technology for single-person households to prevent lonely deaths using data collected from sensors and AI. For this purpose, a total of three PIR(Passive Infrared) sensors were installed inside the testbed, and a Raspberry Pi was utilized as a server. In addition, the GRU(Gated Recurrent Unit) was utilized to monitor and predict the hourly movement of single-person households in real time. Through experiments, it has been confirmed that user movement patterns can be inferred using GRU.