Detecting Gait Events from Accelerations Using Reservoir Computing

Abstract

Segmenting the gait cycle into multiple phases using gait event detection (GED) is a well-researched subject with many accurate algorithms. However, the algorithms that are able to perform accurate and robust GED for real-life environments and physical diseases tend to be too complex for their implementation on simple hardware systems limited in computing power and memory, such as those used in wearable devices. This study focuses on a numerical implementation of a reservoir computing (RC) algorithm called the echo state network (ESN) that is based on simple computational steps that are easy to implement on portable hardware systems for real-time detection. RC is a neural network method that is widely used for signal processing applications and uses a fast-training method based on a ridge regression adapted to the large quantity and variety of IMU data needed to use RC in various real-life environment GED. In this study, an ESN was used to perform offline GED with gait data from IMU and ground force sensors retrieved from three databases for a total of 28 healthy adults and 15 walking conditions. Our main finding is that despite its low complexity, ESN is robust for GED, with performance comparable to other state-of-the-art algorithms. Our results show the ESN is robust enough to obtain good detection results in all conditions if the algorithm is trained with variable data that match those conditions. The distribution of the mean absolute errors (MAE) between the detection times from the ESN and the force sensors were between 40 and 120 ms for 6 defined gait events (95th percentile). We compared our ESN with four different state-of-the-art algorithms from the literature. The ESN obtained a MAE not more than 10 ms above three other reference algorithms for normal walking indoor and outdoor conditions and yielded the 2nd lowest MAE and the 2nd highest true positive rate and specificity when applied to outdoor walking and running conditions. Our work opens the door to using the ESN as a GED for applications in wearable sensors for long-term patient monitoring.

Inertial Sensor Location for Ground Reaction Force and Gait Event Detection Using Reservoir Computing in Gait

“…Matrix W in , W and the parameter , were selected according to the ESN hyperparameters (HP) model according to the first optimization study published in [ 42 ]. HP were selected to optimize computing capability according to the CHARC metrics.…”

“…The numerical implementation of RC called the echo states network (ESN), is one of the most important and used RC methods [ 34 , 37 , 38 ]. ESN [ 37 ], and has proved to be promising in biomechanics applications including gesture recognition [ 39 ], muscle drive-in actuation [ 40 ], exoskeleton control [ 41 ] and GED [ 42 ]. ESN previously demonstrated that sensors on the lower limb are excellent for GED prediction (MAE not more than 10 ms) [ 42 ].…”

“…ESN [ 37 ], and has proved to be promising in biomechanics applications including gesture recognition [ 39 ], muscle drive-in actuation [ 40 ], exoskeleton control [ 41 ] and GED [ 42 ]. ESN previously demonstrated that sensors on the lower limb are excellent for GED prediction (MAE not more than 10 ms) [ 42 ]. However, the best sensor location for the prediction of integrated GED and GRF tasks using the same light reservoir computing algorithm as presented in [ 42 ], remains unaddressed.…”

“…ESN previously demonstrated that sensors on the lower limb are excellent for GED prediction (MAE not more than 10 ms) [ 42 ]. However, the best sensor location for the prediction of integrated GED and GRF tasks using the same light reservoir computing algorithm as presented in [ 42 ], remains unaddressed. The novelty of this study is that it demonstrates the best sensor location for an integrated task of GED and multi-dimensional GRF prediction in able-bodied and MKOA individuals for the first time and via an ESN approach.…”

Fatigue Effect on Minimal Toe Clearance and Toe Activity during Walking

“…Bamberg et al [ 44 ] embedded an accelerometer, a gyroscope, and a ground force reaction sensor in a shoe, while Hegde et al [ 45 ] embedded pressure and acceleration sensors in a shoe. Such shoes with multiple sensors have been applied in personal navigation systems [ 46 ], anomaly detection [ 47 ], activity detection [ 48 ], and the detection of gait events [ 49 ].…”