Search

• The study compares the performance of 13 artificial intelligence-based fall detection algorithms when applied to accelerometer signals captured at the waist of the test subject with a wearable device. • Unlike most related studies, which just use a single data repository to test their proposals, the analysis employs up to 11 different public laboratory-based datasets, as well as two databases with signals measured during real falls, and four repositories with mobility samples gathered during long-term monitoring campaigns. • The study considers not only the typical intra-dataset evaluation procedure followed by the literature but also cross-dataset evaluation (i.e. using data from a different repository for testing than that considered for training). • The performance metrics show that the behavior of all algorithms significantly degrades under cross-dataset evaluation conditions, both when tested with samples from another dataset generated in the laboratory and with real fall or long-term monitoring samples. • The results show that the algorithms tend to overfit the patterns of emulated falls (obtained from laboratory based datasets) used for training. Thus they are not able to extrapolate their learning correctly to conventional movements or new fall patterns, which reflects in a high percentage of undetected falls and a high hourly rate of false alarms. • The obtained results clearly question the evaluation procedures of fall detectors used so far in the literature, highlighting the importance of testing this type of detectors in realistic scenarios with signals from real falls and actual daily life movements. The evaluation of fall detection systems based on wearables is controversial as most studies in the literature benchmark their proposals against falls that are simulated by experimental subjects under unrealistic laboratory conditions. In order to systematically investigate the suitability of this procedure, this paper evaluates a wide set of artificial intelligence algorithms used for fall detection, when trained with a large number of datasets containing acceleration samples captured during the emulation of falls and ordinary movements and then tested with the signals of both actual falls and long-term traces collected from the constant monitoring of users during their daily routines. The results, based on a large number of repositories, show a remarkable degradation in all performance metrics (sensitivity, specificity and false alarm hourly rate) with respect to the typical case in which the detectors are tested with the same types of laboratory movements for which they were trained. [ABSTRACT FROM AUTHOR]

Due to the boom of wireless handheld devices such as smartwatches and smartphones, wearable Fall Detection Systems (FDSs) have become a major focus of attention among the research community during the last years. The effectiveness of a wearable FDS must be contrasted against a wide variety of measurements obtained from inertial sensors during the occurrence of falls and Activities of Daily Living (ADLs). In this regard, the access to public databases constitutes the basis for an open and systematic assessment of fall detection techniques. This paper reviews and appraises twelve existing available data repositories containing measurements of ADLs and emulated falls envisaged for the evaluation of fall detection algorithms in wearable FDSs. The analysis of the found datasets is performed in a comprehensive way, taking into account the multiple factors involved in the definition of the testbeds deployed for the generation of the mobility samples. The study of the traces brings to light the lack of a common experimental benchmarking procedure and, consequently, the large heterogeneity of the datasets from a number of perspectives (length and number of samples, typology of the emulated falls and ADLs, characteristics of the test subjects, features and positions of the sensors, etc.). Concerning this, the statistical analysis of the samples reveals the impact of the sensor range on the reliability of the traces. In addition, the study evidences the importance of the selection of the ADLs and the need of categorizing the ADLs depending on the intensity of the movements in order to evaluate the capability of a certain detection algorithm to discriminate falls from ADLs. [ABSTRACT FROM AUTHOR]

The progress in the field of inertial sensor technology and the widespread popularity of personal electronics such as smartwatches or smartphones have prompted the research on wearable Fall Detection Systems (FDSs). In spite of the extensive literature on FDSs, an open issue is the definition of a common framework that allows a methodical and agreed evaluation of fall detection policies. In this regard, a key aspect is the lack of a public repository of movement datasets that can be employed by the researchers as a common reference to compare and assess their proposals. This work describes UMAFall, a new dataset of movement traces acquired through the systematic emulation of a set of predefined ADLs (Activities of Daily Life) and falls. In opposition to other existing databases for FDSs, which only include the signals captured by one or two sensing points, the testbed deployed for the generation of UMAFall dataset incorporated five wearable sensing points, which were located on five different points of the body of the participants that developed the movements. As a consequence, the obtained data offer an interesting tool to investigate the importance of the sensor placement for the effectiveness of the detection decision in FDSs. [ABSTRACT FROM AUTHOR]

• Most studies on wearable fall detection systems are evaluated with falls emulated in a laboratory. • Clinical trials show that wearable fall detectors underperform in actual application scenarios. • Two repositories containing real falls are compared with 18 well-known datasets of simulated falls. • Fall dynamics are characterized with 15 statistics, computed from the accelerometry components. • The comparison reveals the noteworthy discrepancies between real-world and mimicked falls. • Dynamics of real and fake falls may differ more than those of real falls and other daily activities. • These results call into question the evaluation procedures utilized by the literature to date. Automatic fall detection is one of the most promising applications of wearables in the field of mobile health. The characterization of the effectiveness of wearable fall detectors is hampered by the inherent difficulty of testing these devices with real-world falls. In fact, practically all the proposals in the literature assess the detection algorithms with 'scripted' falls that are simulated in a controlled laboratory environment by a group of volunteers (normally young and healthy participants). Aiming at appraising the adequacy of this method, this work systematically compares the statistical characteristics of the acceleration signals from two databases with real falls and those computed from the simulated falls provided by 18 well-known repositories commonly employed by the related works. The results show noteworthy differences between the dynamics of emulated and real-life falls, which undermines the testing procedures followed to date and forces to rethink the strategies for evaluating wearable fall detectors. [ABSTRACT FROM AUTHOR]

Falls are a major cause of health and psychological problems as well as hospitalization costs among older adults. Thus, the investigation on automatic Fall Detection Systems (FDSs) has received special attention from the research community during the last decade. In this area, the widespread popularity, decreasing price, computing capabilities, built-in sensors and multiplicity of wireless interfaces of Android-based devices (especially smartphones) have fostered the adoption of this technology to deploy wearable and inexpensive architectures for fall detection. This paper presents a critical and thorough analysis of those existing fall detection systems that are based on Android devices. The review systematically classifies and compares the proposals of the literature taking into account different criteria such as the system architecture, the employed sensors, the detection algorithm or the response in case of a fall alarms. The study emphasizes the analysis of the evaluation methods that are employed to assess the effectiveness of the detection process. The review reveals the complete lack of a reference framework to validate and compare the proposals. In addition, the study also shows that most research works do not evaluate the actual applicability of the Android devices (with limited battery and computing resources) to fall detection solutions. [ABSTRACT FROM AUTHOR]

Falls are a foremost source of injuries and hospitalization for seniors. The adoption of automatic fall detection mechanisms can noticeably reduce the response time of the medical staff or caregivers when a fall takes place. Smartphones are being increasingly proposed as wearable, cost-effective and not-intrusive systems for fall detection. The exploitation of smartphones' potential (and in particular, the Android Operating System) can benefit from the wide implantation, the growing computational capabilities and the diversity of communication interfaces and embedded sensors of these personal devices. After revising the state-of-the-art on this matter, this study develops an experimental testbed to assess the performance of different fall detection algorithms that ground their decisions on the analysis of the inertial data registered by the accelerometer of the smartphone. Results obtained in a real testbed with diverse individuals indicate that the accuracy of the accelerometry-based techniques to identify the falls depends strongly on the fall pattern. The performed tests also show the difficulty to set detection acceleration thresholds that allow achieving a good trade-off between false negatives (falls that remain unnoticed) and false positives (conventional movements that are erroneously classified as falls). In any case, the study of the evolution of the battery drain reveals that the extra power consumption introduced by the Android monitoring applications cannot be neglected when evaluating the autonomy and even the viability of fall detection systems. [ABSTRACT FROM AUTHOR]

• Impact of sampling frequency on wearable fall detection system (based on accelerometer signals) is normally neglected. • Sampling rate is normally arbitrarily chosen although it can heavily impact on the battery lifetime of the wearable. • The paper assesses the influence of the sampling rate on the effectiveness of a deep learning classifier (a convolutional neural network) directly fed with an observation window of accelerometry signals. • The study reveals that a frequency of 15–20 Hz may be enough for a proper operation of a wearable (specificity/ sensitivity higher than 95%). • Results are confirmed when the power spectrum of the signals is investigated and when the series are low pass filtered. • The systematic study is based on the massive analysis of 15 public datasets (related literature typically utilizes just 1 or 2 datasets). Last decade has witnessed a major research interest on wearable fall detection systems. Sampling rate in these detectors strongly affects the power consumption and required complexity of the employed wearables. This study investigates the effect of the sampling frequency on the efficacy of the detection process. For this purpose, we train a convolutional neural network to directly discriminate falls from conventional activities based on the raw acceleration signals captured by a transportable sensor. Then, we analyze the changes in the performance of this classifier when the sampling rate is progressively reduced. In contrast with previous studies, the detector is tested against a wide set of public repositories of benchmarking traces. The quality metrics achieved for the different frequencies and the analysis of the spectrum of the signals reveal that a sampling rate of 20 Hz can be enough to maximize the effectiveness of a fall detector. [ABSTRACT FROM AUTHOR]

Due to the serious impact of falls on the quality of life of the elderly and on the economical sustainability of health systems, the study of new monitoring systems capable of automatically alerting about falls has gained much research interest during the last decade. In the field of Human Activity Recognition, Fall Detection Systems (FDSs) can be contemplated as pattern recognition architectures able to discriminate falls from ordinary Activities of Daily Living (ADLs). In this regard, the combined application of cellular communications and wearable devices that integrate inertial sensors offers a cost-efficient solution to track the user mobility almost ubiquitously. Inertial Measurement Units (IMUs) typically utilized for these architectures, embed an accelerometer and a gyroscope. This paper investigates if the use of the angular velocity (captured by the gyroscope) as an input feature of the movement classifier introduces any benefit with respect to the most common case in which the classification decision is uniquely based on the accelerometry signals. For this purpose, the work assesses the performance of a deep learning architecture (a convolutional neural network) which is optimized to differentiate falls from ADLs as a function of the raw data measured by the two inertial sensors (gyroscope and accelerometer). The system is evaluated against on a well-known public dataset with a high number of mobility traces (falls and ADL) measured from the movements of a wide group of experimental users. [ABSTRACT FROM AUTHOR]