Since falls are a major public health problem among older people, the number of systems aimed at detecting them has increased dramatically over recent years. This work presents an extensive literature review of fall detection systems, including comparisons among various kinds of studies. It aims to serve as a reference for both clinicians and biomedical engineers planning or conducting field investigations. Challenges, issues and trends in fall detection have been identified after the reviewing work. The number of studies using context-aware techniques is still increasing but there is a new trend towards the integration of fall detection into smartphones as well as the use of machine learning methods in the detection algorithm. We have also identified challenges regarding performance under real-life conditions, usability, and user acceptance as well as issues related to power consumption, real-time operations, sensing limitations, privacy and record of real-life falls.

No abstract

Despite being a major public health problem, falls in the elderly cannot be detected efficiently yet. Many studies have used acceleration as the main input to discriminate between falls and activities of daily living (ADL). In recent years, there has been an increasing interest in using smartphones for fall detection. The most promising results have been obtained by supervised Machine Learning algorithms. However, a drawback of these approaches is that they rely on falls simulated by young or mature people, which might not represent every possible fall situation and might be different from older people's falls. Thus, we propose to tackle the problem of fall detection by applying a kind of novelty detection methods which rely only on true ADL. In this way, a fall is any abnormal movement with respect to ADL. A system based on these methods could easily adapt itself to new situations since new ADL could be recorded continuously and the system could be re-trained on the fly. The goal of this work is to explore the use of such novelty detectors by selecting one of them and by comparing it with a state-of-the-art traditional supervised method under different conditions. The data sets we have collected were recorded with smartphones. Ten volunteers simulated eight type of falls, whereas ADL were recorded while they carried the phone in their real life. Even though we have not collected data from the elderly, the data sets were suitable to check the adaptability of novelty detectors. They have been made publicly available to improve the reproducibility of our results. We have studied several novelty detection methods, selecting the nearest neighbour-based technique (NN) as the most suitable. Then, we have compared NN with the Support Vector Machine (SVM). In most situations a generic SVM outperformed an adapted NN.

Falls are one of the leading causes of mortality among the older population, being the rapid detection of a fall a key factor to mitigate its main adverse health consequences. In this context, several authors have conducted studies on acceleration-based fall detection using external accelerometers or smartphones. The published detection rates are diverse, sometimes close to a perfect detector. This divergence may be explained by the difficulties in comparing different fall detection studies in a fair play since each study uses its own dataset obtained under different conditions. In this regard, several datasets have been made publicly available recently. This paper presents a comparison, to the best of our knowledge for the first time, of these public fall detection datasets in order to determine whether they have an influence on the declared performances. Using two different detection algorithms, the study shows that the performances of the fall detection techniques are affected, to a greater or lesser extent, by the specific datasets used to validate them. We have also found large differences in the generalization capability of a fall detector depending on the dataset used for training. In fact, the performance decreases dramatically when the algorithms are tested on a dataset different from the one used for training. Other characteristics of the datasets like the number of training samples also have an influence on the performance while algorithms seem less sensitive to the sampling frequency or the acceleration range.