Deep Learning Based Fall Detection Algorithms for Embedded Systems, Smartwatches, and IoT Devices Using Accelerometers

A robust multimodal detection system: physical exercise monitoring in long-term care environments

Introduction

Falls are a major cause of accidents that can lead to serious injuries, especially among geriatric populations worldwide. Ensuring constant supervision in hospitals or smart environments while maintaining comfort and privacy is practically impossible. Therefore, fall detection has become a significant area of research, particularly with the use of multimodal sensors. The lack of efficient techniques for automatic fall detection hampers the creation of effective preventative tools capable of identifying falls during physical exercise in long-term care environments. The primary goal of this article is to examine the benefits of using multimodal sensors to enhance the precision of fall detection systems.

Methods

The proposed paper combines time-frequency features of inertial sensors with skeleton-based modeling of depth sensors to extract features. These multimodal sensors are then integrated using a fusion technique. Optimization and a modified K-Ary classifier are subsequently applied to the resultant fused data.

Results

The suggested model achieved an accuracy of 97.97% on the UP-Fall Detection dataset and 97.89% on the UR-Fall Detection dataset.

Discussion

This indicates that the proposed model outperforms state-of-the-art classification results. Additionally, the proposed model can be utilized as an IoT-based solution, effectively promoting the development of tools to prevent fall-related injuries.

A Deep Learning Approach to Classify Sitting and Sleep History from Raw Accelerometry Data during Simulated Driving

Prolonged sitting and inadequate sleep can impact driving performance. Therefore, objective knowledge of a driver's recent sitting and sleep history could help reduce safety risks. This study aimed to apply deep learning to raw accelerometry data collected during a simulated driving task to classify recent sitting and sleep history. Participants (n = 84, Mean ± SD age = 23.5 ± 4.8, 49% Female) completed a seven-day laboratory study. Raw accelerometry data were collected from a thigh-worn accelerometer during a 20-min simulated drive (8:10 h and 17:30 h each day). Two convolutional neural networks (CNNs; ResNet-18 and DixonNet) were trained to classify accelerometry data into four classes (sitting or breaking up sitting and 9-h or 5-h sleep). Accuracy was determined using five-fold cross-validation. ResNet-18 produced higher accuracy scores: 88.6 ± 1.3% for activity (compared to 77.2 ± 2.6% from DixonNet) and 88.6 ± 1.1% for sleep history (compared to 75.2 ± 2.6% from DixonNet). Class activation mapping revealed distinct patterns of movement and postural changes between classes. Findings demonstrate the suitability of CNNs in classifying sitting and sleep history using thigh-worn accelerometer data collected during a simulated drive. This approach has implications for the identification of drivers at risk of fatigue-related impairment.

Artificial Intelligence-Driven Intrusion Detection in Software-Defined Wireless Sensor Networks: Towards Secure IoT-Enabled Healthcare Systems

Wireless Sensor Networks (WSNs) are increasingly deployed in Internet of Things (IoT) systems for applications such as smart transportation, telemedicine, smart health monitoring and fall detection systems for the elderly people. Given that huge amount of data, vital and critical information can be exchanged between the different parts of a WSN, good management and protection schemes are needed to ensure an efficient and secure operation of the WSN. To ensure an efficient management of WSNs, the Software-Defined Wireless Sensor Network (SDWSN) paradigm has been recently introduced in the literature. In the same vein, Intrusion Detection Systems, have been used in the literature to safeguard the security of SDWSN-based IoTs. In this paper, three popular Artificial Intelligence techniques (Decision Tree, Naïve Bayes, and Deep Artificial Neural Network) are trained to be deployed as anomaly detectors in IDSs. It is shown that an IDS using the Decision Tree-based anomaly detector yields the best performances metrics both in the binary classification and in the multinomial classification. Additionally, it was found that an IDS using the Naïve Bayes-based anomaly detector was only adapted for binary classification of intrusions in low memory capacity SDWSN-based IoT (e.g., wearable fitness tracker). Moreover, new state-of-the-art accuracy (binary classification) and F-scores (multinomial classification) were achieved by introducing an end-to-end feature engineering scheme aimed at obtaining 118 features from the 41 features of the Network Security Laboratory-Knowledge Discovery in Databases (NSL-KDD) dataset. The state-of-the-art accuracy was pushed to 0.999777 using the Decision Tree-based anomaly detector. Finally, it was found that the Deep Artificial Neural Network should be expected to become the next default anomaly detector in the light of its current performance metrics and the increasing abundance of training data.

A Novel Feature Set Extraction Based on Accelerometer Sensor Data for Improving the Fall Detection System

Because falls are the second leading cause of injury deaths, especially in the elderly according to WHO statistics, there have been a lot of studies on developing a fall detection and warning system. Many approaches based on wearable sensors, cameras, Infrared sensors, radar, etc., have been proposed to detect falls efficiently. However, it still faces many challenges due to noise and no clear definition of fall activities. This paper proposes a new way to extract 44 features based on the time domain, frequency domain, and Hjorth parameters to deal with this. The effect of the proposed feature set has been evaluated on several classification algorithms, such as SVM, k-NN, ANN, J48, and RF. Our method achieves a relative high performance (F1-Score metric) in detecting fall and non-fall activities, i.e., 95.23% (falls), 99.11% (non-falls), and 96.16% (falls), 99.90% (non-falls) for the MobileAct 2.0 and UP-Fall datasets, respectively.

A study on the impact of the users' characteristics on the performance of wearable fall detection systems

Wearable Fall Detection Systems (FDSs) have gained much research interest during last decade. In this regard, Machine Learning (ML) classifiers have shown great efficiency in discriminating falls and conventional movements or Activities of Daily Living (ADLs) based on the analysis of the signals captured by transportable inertial sensors. Due to the intrinsic difficulties of training and testing this type of detectors in realistic scenarios and with their target audience (older adults), FDSs are normally benchmarked against a predefined set of ADLs and emulated falls executed by volunteers in a controlled environment. In most studies, however, samples from the same experimental subjects are used to both train and evaluate the FDSs. In this work, we investigate the performance of ML-based FDS systems when the test subjects have physical characteristics (weight, height, body mass index, age, gender) different from those of the users considered for the test phase. The results seem to point out that certain divergences (weight, height) of the users of both subsets (training ad test) may hamper the effectiveness of the classifiers (a reduction of up 20% in sensitivity and of up to 5% in specificity is reported). However, it is shown that the typology of the activities included in these subgroups has much greater relevance for the discrimination capability of the classifiers (with specificity losses of up to 95% if the activity types for training and testing strongly diverge).

Development of an Anomaly Alert System Triggered by Unusual Behaviors at Home

In many countries, the number of elderly people has grown due to the increase in the life expectancy of the population, many of whom currently live alone and are prone to having accidents that they cannot report, especially if they are immobilized. For this reason, we have developed a non-intrusive IoT device, which, through multiple integrated sensors, collects information on habitual user behavior patterns and uses it to generate unusual behavior rules. These rules are used by our SecurHome system to send alert messages to the dependent person’s family members or caregivers if their behavior changes abruptly over the course of their daily life. This document describes in detail the design and development of the SecurHome system.

Innovative IoT Solutions and Wearable Sensing Systems for Monitoring Human Biophysical Parameters: A Review

Digital and information technologies are heavily pervading several aspects of human activities, improving our life quality. Health systems are undergoing a real technological revolution, radically changing how medical services are provided, thanks to the wide employment of the Internet of Things (IoT) platforms supporting advanced monitoring services and intelligent inferring systems. This paper reports, at first, a comprehensive overview of innovative sensing systems for monitoring biophysical and psychophysical parameters, all suitable for integration with wearable or portable accessories. Wearable devices represent a headstone on which the IoT-based healthcare platforms are based, providing capillary and real-time monitoring of patient’s conditions. Besides, a survey of modern architectures and supported services by IoT platforms for health monitoring is presented, providing useful insights for developing future healthcare systems. All considered architectures employ wearable devices to gather patient parameters and share them with a cloud platform where they are processed to provide real-time feedback. The reported discussion highlights the structural differences between the discussed frameworks, from the point of view of network configuration, data management strategy, feedback modality, etc.

IoT-Based Applications in Healthcare Devices

The last decade has witnessed extensive research in the field of healthcare services and their technological upgradation. To be more specific, the Internet of Things (IoT) has shown potential application in connecting various medical devices, sensors, and healthcare professionals to provide quality medical services in a remote location. This has improved patient safety, reduced healthcare costs, enhanced the accessibility of healthcare services, and increased operational efficiency in the healthcare industry. The current study gives an up-to-date summary of the potential healthcare applications of IoT- (HIoT-) based technologies. Herein, the advancement of the application of the HIoT has been reported from the perspective of enabling technologies, healthcare services, and applications in solving various healthcare issues. Moreover, potential challenges and issues in the HIoT system are also discussed. In sum, the current study provides a comprehensive source of information regarding the different fields of application of HIoT intending to help future researchers, who have the interest to work and make advancements in the field to gain insight into the topic.