The effect of window size and lead time on pre-impact fall detection accuracy using support vector machine analysis of waist mounted inertial sensor data

Abstract

Falls are a major cause of death and morbidity in older adults. In recent years many researchers have examined the role of wearable inertial sensors (accelerometers and/or gyroscopes) to automatically detect falls. The primary goal of such fall monitors is to alert care providers of the fall event, who can then commence earlier treatment. Although such fall detection systems may reduce time until the arrival of medical assistance, they cannot help to prevent or reduce the severity of traumatic injury caused by the fall. In the current study, we extend the application of wearable inertial sensors beyond post-impact fall detection, by developing and evaluating the accuracy of a sensor system for detecting falls prior to the fall impact. We used support vector machine (SVM) analysis to classify 7 fall and 8 non-fall events. In particular, we focused on the effect of data window size and lead time on the accuracy of our pre-impact fall detection system using signals from a single waist sensor. We found that our system was able to detect fall events at between 0.0625-0.1875 s prior to the impact with at least 95% sensitivity and at least 90% specificity for window sizes between 0.125-1 s.

TagFall: Towards Unobstructive Fine-Grained Fall Detection based on UHF Passive RFID Tags

“…The Methods: As for fall detection techniques, current fall detection systems mainly adopt supervised classificationbased method to detect a fall event, such as Support Vector Machine (SVM) [2,8,12], Neural Network [9] or Extreme Learning Machine [21], which have to tune plenty of parameters to achieve satisfied accuracy. But in our fall detection phase, we aim to mine the clustering patterns of RSSIs based on the variances of angle paired by data point of different actions when the environment is affected by diverse human activities.…”

“…Over the past decades, fall detection (FD) and prevention have been an active research area with several proposed solutions. Both wearable sensor based (e.g., inertial sensors [2], accelerometer [3,8], specialized cane [16]) and smart-phone based [5,14] fall detection techniques require the subject to be attached with sensors or phones, which might not be practical (e.g., sensors lost/damaged, or forget to carry by the elderly with dementia). Vision based fall detection systems [9,27,18] employ activity classification algorithms on a series of images recorded by a video camera, which is usually regarded as being privacy invasive and causes uncomfortable feeling to the elderly.…”

“…These sensors are capable of providing raw data with high precision and accuracy. Thus, smart phone based fall detection is promising and with good potential [1], which can integrate all sensors into one single mobile device (e.g., inertial sensors [2], tri-axial accelerometers [5] and gyroscopes [14]). However, smart-phone based fall detection systems share the same mechanism as wearable sensors based techniques.…”

A Survey Paper on Wearable Sensors based Fall Detection

“…By performing trial and error method, lead times 0.25 s or greater, sensitivity and specificity varied dramatically with choice of window size, indicating poor robustness of the classification performance. Therefore, the use of a target lead time around 0.1875 s or less, and window size 1 s or less for robust pre-impact fall detection [7].…”

Time-Frequency Linearization of Reactive Cortical Responses for the Early Detection of Balance Losses

“…Despite a fast detection, in most cases, these approaches provide accu-racy below the 90%. To improve the PIFD strategy discrimination capability, realizing more efficient threshold-based systems, several solutions use ML methodologies [10,14,23]. Nevertheless, the ML algorithms request for a prolonged classifier training period.…”