Pre-Impact Fall Detection with CNN-Based Class Activation Mapping Method.

Min Wang,Jingyi Shi,Diansheng Chen

doi:10.3390/s20174750

Abstract

In this paper, we report our improvement on the prediction accuracy of pre-impact fall detection by applying a learning-based method on the real-time data from an IMU (inertial measurement unit)-sensor mounted on the waist, making it possible to achieve a high accuracy on a wearable device with the extracted features. Using the fixed threshold method is difficult for achieving satisfactory detection accuracy, due to various characteristics and behaviors in the movement of different individuals. In contrast, one could realize high-accuracy detection with machine learning-based methods, but it is difficult to apply them in the wearable devices due to the high hardware requirement. Our method merges the two methods above. We build a convolutional neural network (CNN) with a class activation mapping (CAM) method, which could highlight the class-specific region in the data and obtain a hot map of the fall data. After training on the MobiAct dataset, the model could achieve high-accuracy detection (95.55%) and obtain the region with high contributions to the classification. Then, we manually extract effective features and characteristics of this region and form our special threshold method, achieving pre-impact fall detection in real-world data. Consequently, our method achieves accuracy of 95.33% and a detection time of within 400 ms.

Highlights

Falls are a major threat to people’s health, since they are the cause of disability and injury-related deaths, especially for the elderly
Based on the contributing region on the hot maps, we found that the characteristics of support vector machine (SVM) learned by convolutional neural network (CNN), similar to the results verified on the MobiAct dataset, contributed to identify a pre-impact fall, mostly in the early fall phase
The CNN-LSTM model reported by Hassan et al [20] achieved 96.75% accuracy on the MobiAct dataset, which was claimed as a state-of-the-art approach

Summary

Introduction

Falls are a major threat to people’s health, since they are the cause of disability and injury-related deaths, especially for the elderly. The injuries caused by falls are serious, including fractures, joint dislocations, and cognitive decline, which pose high costs to both family and public health care. Accelerometer-based fall detection systems are widely used in all fall stages, including pre-impact, impact, and post-impact [2]. Some studies used a gyroscope along with an accelerometer integrated as a portable inertial sensor, whose data contains posture information, to detect a fall before impact [4]. These pre-impact fall detection systems either use a threshold-based [5] or a machine learning-based method [6]

Objectives

Methods

Results

Discussion

Conclusion