Fall Prediction and Prevention Systems: Recent Trends, Challenges, and Future Research Directions

Abstract

Fall prediction is a multifaceted problem that involves complex interactions between physiological, behavioral, and environmental factors. Existing fall detection and prediction systems mainly focus on physiological factors such as gait, vision, and cognition, and do not address the multifactorial nature of falls. In addition, these systems lack efficient user interfaces and feedback for preventing future falls. Recent advances in internet of things (IoT) and mobile technologies offer ample opportunities for integrating contextual information about patient behavior and environment along with physiological health data for predicting falls. This article reviews the state-of-the-art in fall detection and prediction systems. It also describes the challenges, limitations, and future directions in the design and implementation of effective fall prediction and prevention systems.

A Systematic Review on the Use of Wearable Body Sensors for Health Monitoring: A Qualitative Synthesis

“…Intrinsic factors include age, fall history, mobility impairments, sleep disturbances, and neurological disorders", pp. 1 [76]. It is reported in [77] that 35% of non-institutionalized adults had abnormal gait and that sleep disturbances are very common among older people.…”

“…Yet, the majority of the studies focusing on gait and fall in this review were simulations that include none or few old participants. This shortcoming is also discussed in [76], "It is evident that existing systems have mainly been tested in laboratory environments with controlled conditions and do not include frequent fallers and aging adults as test subjects.[..] future work should focus on longitudinal studies of fall detection and prediction systems in real-life conditions on a diverse group that includes frequent fallers, aging adults, and persons with neurological disorders."…”

Quantitative Mobility Assessment for Fall Risk Prediction in Dementia: A Systematic Review

“…As such, linear and other commonly used statistical models such as logistic regression, ANOVA, or χ 2 test cannot extract useful information from episodes of human motion data and are unlikely to result in accurate prediction of falls [45]. Machine learning algorithms, including dynamic temporal models such as recurrent neural network or hidden Markov models, may be better tools to detect gait and balance anomalies in individuals' walking patterns that are predictive of impending falls [46].…”

“…and failures in their data collection. Others in the literature have reported that older adults transferred the wearable sensors between various body locations due to discomfort [46,47]. Moreover, at present, the algorithms used to process data from wearable sensors can have difficulty identifying gait episodes that are asymmetric or irregular, which limits their utility in the advanced dementia population.…”

Deep Recurrent Neural Networks for Edge Monitoring of Personal Risk and Warning Situations

“…System. Different deep learning methods have been successfully used for fall detection [2,[25][26][27][28]. Analyzing those systems, we see that all these methods rely either on models with a huge number of parameters or on remote communication.…”