A convolution neural network approach for fall detection based on adaptive channel selection of UWB radar signals

Abstract

According to the World Health Organization and other authorities, falls are one of the main causes of accidental injuries among the elderly population. Therefore, it is essential to detect and predict the fall activities of older persons in indoor environments such as homes, nursing, senior residential centers, and care facilities. Due to non-contact and signal confidentiality characteristics, radar equipment is widely used in indoor care, detection, and rescue. This paper proposes an adaptive channel selection algorithm to separate the activity signals from the background using an ultra-wideband radar and to generalize fused features of frequency- and time-domain images which will be sent to a lightweight convolutional neural network to detect and recognize fall activities. The experimental results show that the method is able to distinguish three types of fall activities (i.e., stand to fall, bow to fall, and squat to fall) and obtain a high recognition accuracy up to 95.7%.

The Effect of using Dimensionality Reduction Compared with Type of Algorithm on Detecting Patient Fall: Triage Case Study

“…Several recent studies have been conducted on various systems for fall detection, relying on wearable sensors, vision-based, DL algorithms, and ML techniques [22][23][24][25][26].…”

Convolutional neural network for wearable fall detection systems

“…The input layer received segmented data from a 3-axis accelerometer. A batch normalization layer followed the input layer to make the data distribution consistent (Wang et al, 2022). Three convolution layers consisted of 16, 32, and 64 kernels, with a ReLU activation function connected to the batch normalization layer (Gadaleta and Rossi, 2018).…”

Revolutionizing Geriatric Care: Radar and Digital Twin Technology for Fall Severity Detection

“…As the aging demographic continues to reshape societies across the globe, the challenges within the healthcare system, particularly in Long-Term Care (LTC) facilities, have become increasingly pronounced. Approximately 9.1% of the global population, 701 million individuals, is estimated to consist of seniors aged 65 and above [1]. Care homes across North America find themselves struggling with extreme staffing and resource shortages, often resulting in alarming levels, leading to preventable deaths, closures of care homes, surging costs, and a pervasive negative impact on the quality of life for residents [2].…”

“…This taxing phenomenon poses a serious threat to the prompt response to critical alerts, jeopardizing the well-being of seniors. One of the sources contributing to alarm fatigue is falls, as they stand as a primary cause of death and injury among seniors [1,4,5]. Various technologies, such wearable, nonwearable, and hybrid systems, have been developed for fall detection [6].…”

“…Radar sensors emerge as a promising solution [19] due to their efficacy in preserving privacy, ensuring accuracy, operating independently of illumination levels, and detecting even slow falls that might be missed by wearables [1,4,13,14,20,21,22,23,24,25]. Machine learning algorithms such as Support Vector Machines(SVM) [13], Random Forests, K-Nearest Neighbours(KNN), K-Means, Linear Discriminant Analysis(LDA) [26] and and deep learning such as Long Short-Term Memory(LSTM) [27], Convolutional Neural Network(CNN) [22] and Gated recurrent unit(GRU) [28] have been used for fall detection [22].…”