Gait temporal parameters estimation in toddlers using inertial measurement units: A comparison of 15 algorithms

Abstract

Background

Children’s motor development can be evaluated through the analysis of gait temporal parameters and their variability. This requires the detection of gait events in a real-world environment, which can be achieved using inertial measurement units. Algorithms have been previously developed for healthy adults; however, the performance of these algorithms in the detection of gait events in toddlers has not been analysed.

Research question

Can inertial measurement units be used to analyse gait temporal parameters in toddlers?

Methods

Fifteen previously published algorithms using sensors attached on the lower-back or the ankles were used to identify gait events and calculate gait temporal parameters. A total of 1388 initial and 1388 final foot contacts collected from 15 toddlers were included in the analysis. The performance of the algorithms was compared against a GAITRite mat in terms of accuracy and precision. Accuracy in the measurement of gait temporal parameters was evaluated using Bland Altman limits of agreement for repeated measurements, and precision was assessed through the evaluation of correctly identified, falsely identified and missed events.

Results

From our results, no algorithm emerged as a best option from all those analysed. Algorithms using the ankle sensors provide higher accuracy and perfect precision when using only angular velocity about the medio-lateral axis. The best algorithms using the sensor attached at the lower-back use the resultant or global acceleration that reduces the effect of the sensor’s alignment. These lower-back-based algorithms compared to the best ankle-based ones have similar accuracy for the calculation of stride time and higher accuracy for step time; however, they do not have perfect precision.

Significance

Inertial measurement units can support research analysing the temporal parameters of toddlers’ gait in controlled environments, and may allow future studies in natural, free-living environments that can improve the monitoring of gait in young children.