Artificial Neural Networks: Case for Real-time Accident Detection and Emergency Response in Elderly Care

Abstract

The number of senior citizens is rising, and so is the demand for care services. Among the care services, one of the most crucial ones that senior citizens require is sudden fall detection. (Yoo & Oh, 2018). Between February and December 2021, approximately 20.9% of admissions to the Kenyatta National Hospital were due to falls (Omondi et al., 2023). Due to their high frequency and serious side effects among the elderly population, falls pose a serious public health concern. Physical injury, a reduction in independence, and a significant financial strain on families and healthcare systems are a few of these. Falls are especially dangerous in a population that is getting older. For this reason, automatic fall detection is crucial in order to provide older fall victims with prompt medical attention. The current fall detection techniques have limited practical uses in wearable fall detection devices due to their low robustness or high computing cost (Yu et al., 2023). The aim of this research is developing an AI solution that uses the Artificial Neural Network Model for real-time fall detection and emergency response, specifically tailored for the elderly population. The proposed solution was developed using sensor data collected from simulated fall experiments, which was used to train the ANN for accurate fall detection and was integrated into an embedded system architecture. The model's performance was 100% accurate, with 1 as the F1 score, 21 ms latency, and 3.1 KB RAM use. Despite promising results, the system faces limitations including challenges in real-world scenarios, connectivity dependencies, potential false positives/negatives, battery life restrictions, and reliance on timely human response. This study promotes elderly care through an AI-driven system that enhances fall detection accuracy and response time, addressing key challenges in fall-related incidents. In conclusion, this project demonstrates the potential of ANNs in creating an efficient, accurate, and privacy-preserving fall detection system for elderly care, paving the way for improved safety and emergency response in resource-constrained wearable devices.