A Survey on Recent Advances in Wearable Fall Detection Systems

Table 1

Threshold-based systems for fall detection using wearable devices.

Reference	Year	Dataset used	Sensors used	Sensor placement	Methodology	Observed performance
[21]	2009	Generated from experiments	Accelerometer, gyroscope	Chest, thigh	Three-step algorithm based on activity intensity analysis, posture analysis, and transition analysis, based on signals reported by accelerometer and gyroscope	Sensitivity = 91%; Specificity = 92%
[18]	2011	Generated from experiments	3D accelerometer	Not specified	Algorithm based on first differences and first derivatives of sum of accelerometer readings along X, Y, and Z directions	Algorithm is reliable, simple, and real time
[19]	2015	Generated from experiments	Accelerometer	Waist	Quaternion algorithm using sum acceleration and angle information	Better sensitivity and specificity than threshold-based algorithms
[20]	2015	Generated from experiments	Accelerometer, HRV sensor	Not specified	Analysis of signals from accelerometer for movement detection and HRV sensor for stress detection	Accuracy = 96% to 100% (depending on the type of movement)
[25]	2017	Generated from experiments	3-Axis accelerometer, 3-axis gyroscope, 3-axes magnetometer	Shoulder, waist, and foot	Threshold-based method, applied to acceleration and Euler’s angle (yaw, pitch, and roll), run on a mobile phone	Accuracy = 100%; Specificity = 91.1% (shoulder), 100% (waist), 78.5% (foot); Sensitivity = 100% (for all three placements)
[27]	2017	Generated from experiments	G-force sensor	Smartphone	3-Phase detection based on thresholds to identify falls and smartphone drops	Specificity = 72% when compared to the specificity of 31% with 2-phase threshold-based algorithm
[23]	2017	Generated from experiments	MEMS accelerometers, RF signals	Waist + network of fixed motes within the home	Signal analysis based on threshold-based methods	Not specified