Automatic fall risk assessment with Siamese network for stroke survivors using inertial sensor‐based signals

Abstract

Fall is a major threat to stroke survivors with the problems of gait and balance disorders in the rehabilitation phase following severe consequences on quality of life and a heavy burden to their families. Many solutions have been proposed to assess fall risk for elders based on inertial sensor‐based signals, however, there still exists a great challenge of transferring them from elderly populations to the stroke‐survivors populations as gait disorder patterns are significant difference between elders and stroke survivors. In this study, we conduct a pilot study to collect inertial sensor‐based signals from stroke survivors when they performed the timed up and go test, and build an automatic fall risk assessment model with the architecture of Siamese network, with a merit of mitigating the problem of small sample size. Specifically, the proposed automatic fall risk assessment model consists of two parallel convolutional neural networks, each of which is composed of three convolutional layers, two max‐pooling layers, and three fully connected layers. To utilize the space relation among accelerator‐based and gyroscope‐based signals, two‐dimensional discrete wavelet transform extracts image‐like features, wavelet coefficients, from inertial sensor‐based signals as the input. Experimental results show that the proposed fall risk assessment model has achieved a promising results, which outperform cutting‐edge methods with a big margin. The proposed fall risk assessment model with low computational complexity and limited memory consuming can be deployed on an embedded system to provide fall risk assessment service for stroke survivors in point‐of‐care environments or community settings.

Prediction of Wellness Condition for Community-Dwelling Elderly via ECG Signals Data-Based Feature Construction and Modeling

“…For example, wearable health trackers can be used to monitor gait speed, sleep time, heartbeats, daily diet, and energy consumption, etc. Most existing studies involve chronic conditions [ 4 , 5 , 6 ], rehabilitation [ 7 , 8 ], cardiovascular disease [ 9 , 10 , 11 ], fall [ 12 , 13 , 14 ], and general wellness [ 15 , 16 , 17 , 18 ].…”

“…Nowadays, single-lead ECG signals are conveniently available via wearable or portable health-monitoring devices without the limits of time and location. Fan et al developed a one-day-forward forecasting method of wellness for a community-dwelling elderly population based on single-lead short ECG signals [ 14 ]. Despite the inspiring results of wellness prediction from these research, the matter of how to extract multiple useful features from the raw data for wellness prediction is still a challenge due to the dynamic characteristics of ECG signals.…”

Time series change detection using reservoir computing networks for remote sensing data

“…When calculating the error, the chain rule is generally used to calculate the gradients. The current state depends not only on the current input but also on previous input states; these operations increase the computational cost of training the network 1–3 …”

“…The current state depends not only on the current input but also on previous input states; these operations increase the computational cost of training the network. [1][2][3] Recursive neural networks 4,5 have a feedback connection mechanism and are suitable for dynamic data processing. Recursive connections allow the network to capture temporal features in continuous inputs, and the temporal dependence of the inputs can be embedded in their dynamic behavior.…”

An Indoor Fall Monitoring System: Robust, Multistatic Radar Sensing and Explainable, Feature-Resonated Deep Neural Network

“…Regarding the former, commonly used sensors include inertial measurement units (IMUs). The potential of wearable sensors has been explored in fall risk assessment [8] and outdoor fall detection [9]. Fall risk assessment with wearable sensing systems requires a series of instructions under supervision [8], [9], and the need for bodily contact makes it less usable for certain individuals.…”