A low-cost stand-alone platform for measuring motor behavior across developmental applications

Abstract

Motion tracking provides unique insights into motor, cognitive, and social development by capturing subtle variations into how movements are planned and controlled. Here, we present a low-cost, wearable movement measurement platform, KiD, specifically designed for tracking the movements of infants and children in a variety of natural settings. KiD consists of a small, lightweight sensor containing a nine-axis inertial measurement unit plus an integrated processor for computing rotations. Measurements of three-dimensional acceleration using KiD compare well with those of current state-of-the-art optical motion capture systems. As a proof of concept, we demonstrate successful classification of different types of sinusoidal right arm movements using KiD.

Keywords: Behavioral neuroscience; Biological sciences; Neuroscience; Techniques in neuroscience.

Graphical abstract

Figure 1

Overview of the KiD platform (Upper panel) The KiD platform being worn by a child while performing an elliptical movement, and exploded view of the KiD tracking unit. (Lower panel) Overview of the KiD user interface (UI), and block diagram of the potential use of KiD as a stand-alone platform for measuring motor behavior across developmental applications.

Figure 2

Similarity between the magnitude of motion acceleration measured using KiD and MoCap during horizontal, vertical, elliptical, and figure eight movements (A) Representation of movement performed by an exemplar participant. (B) Magnitude of motion acceleration (Am) measured by KiD and MoCap across samples of an exemplar participant. (C) Scatterplot of the Am measured by MoCap against the Am measured by KiD across individual participants at intervals of 10% of the normalized movement time. Each dot represents the average acceleration for each subject. Black line represents the line of equality between MoCap and KiD acceleration. Red line is a trend line (least squares line) passing through the observed values.

Figure 3

Classification of movement types (A) Confusion matrix of SVM classifier trained and tested on MoCap data. Rows represent the percentage of true movement-type labels. Columns represent the percentage of predicted movement-type labels. (B) Permutation null distribution obtained by classifying MoCap with shuffled movement-type labels. The permutation null distribution is represented by the gray histograms. The red line represents the observed accuracy. (C) Confusion matrix of SVM classifier trained on MoCap data and tested on KiD data. (D) Confusion matrix of SVM classifier trained and tested on KiD data. (E) Permutation null distribution obtained by classifying KiD with shuffled movement-type labels. The permutation null distribution is represented by the gray histograms. The yellow line represents the observed accuracy. (F) Confusion matrix of SVM classifier trained on KiD data and tested on MoCap data.