An Analysis on Sensor Locations of the Human Body for Wearable Fall Detection Devices: Principles and Practice.

Area of Science:

• Biomedical Engineering
• Human-Computer Interaction
• Machine Learning

Background:

• Falls pose significant risks, especially to the elderly, necessitating effective detection methods.
• Current wearable fall detection devices face challenges with ergonomics and user-friendliness for older adults.
• Optimizing sensor placement and reducing node count are crucial for practical wearable fall detection systems.

Purpose of the Study:

• To identify the optimal body location for wearable sensors to maximize fall detection accuracy.
• To reduce the number of sensor nodes in wearable devices for improved ergonomics.
• To evaluate the performance of various machine learning techniques for fall detection based on sensor location.

Main Methods:

• A large dataset of 2520 tests was collected using triaxial sensors (accelerometer, gyroscope, magnetometer) placed on six body locations: head, chest, waist, right-wrist, right-thigh, and right-ankle.
• Six machine learning techniques (k-NN, BDM, SVM, LSM, DTW, ANNs) were applied to analyze sensor data from single and multiple sensor configurations (63 combinations).
• The sensitivity and accuracy of each sensor location and machine learning combination were systematically investigated.

Main Results:

• The waist region demonstrated the highest fall detection sensitivity at 99.96% when utilizing the k-nearest neighbor (k-NN) classifier.
• The right-wrist sensor placement achieved a maximum sensitivity of 97.37%, despite its common use in current wearable technology.
• The study evaluated 378 unique sensor configuration and machine learning technique combinations.

Conclusions:

• The waist is the most effective location for wearable sensor placement in fall detection systems, offering superior accuracy and reliability.
• Optimizing sensor placement, particularly on the waist, significantly enhances fall detection performance for elderly individuals.
• This research provides valuable insights for designing more ergonomic and effective wearable fall detection devices.