Artificial Intelligence Distinguishes Pathological Gait: The Analysis of Markerless Motion Capture Gait Data   Acquired by an iOS Application (TDPT-GT)

Abstract

Distinguishing pathological gait is challenging in neurology because of the difficulty of capturing total body movement and its analysis. We aimed to obtain a convenient recording with an iPhone and establish an algorithm based on deep learning. From May 2021 to November 2022 at Yamagata University Hospital, Shiga University, and Takahata Town, patients with idiopathic normal pressure hydrocephalus (n = 48), Parkinson’s disease (n = 21), and other neuromuscular diseases (n = 45) comprised the pathological gait group (n = 114), and the control group consisted of 160 healthy volunteers. iPhone application TDPT-GT captured the subjects walking in a circular path of about 1 meter in diameter, a markerless motion capture system, with an iPhone camera, which generated the three-axis 30 frames per second (fps) relative coordinates of 27 body points. A light gradient boosting machine (Light GBM) with stratified k-fold cross-validation (k = 5) was applied for gait collection for about 1 min per person. The median ability model tested 200 frames of each person’s data for its distinction capability, which resulted in the area under a curve of 0.719. The pathological gait captured by the iPhone could be distinguished by artificial intelligence.

Fluctuations in Upper and Lower Body Movement during Walking in Normal Pressure Hydrocephalus and Parkinson’s Disease Assessed by Motion Capture with a Smartphone Application, TDPT-GT

“…As a result, circle walking likely induced gait disorders more sensitively than straight walking, making it a potentially valuable assessment tool. Until now, circular walking in conjunction with TDPT-GT has been effective in recording pathological gait, as demonstrated in previous studies [32][33][34]. The present study further supports this.…”

“…However, the association between dexterity in the upper limbs and the fluctuation of motor control during walking is still unclear. Our previous study, using the same motion-capture system, used AI to differentiate between pathological gait and normal gait [34], and AI emphasized utilizing the trunk positions of the body [34]. Thus, we speculated that the trunk-posture adaptation [61] and/or the subjective vertical position [62] play a role in the movement of total or other body positions and contribute to gait or balance (Figure 6).…”

“…As a result, our goal is to make these beneficial sensors as easily accessible as possible for patients in need. The system we employed was a smartphone-based markerless motioncapture system using artificial intelligence (AI), named TDPT-GT (Three-Dimensional Pose Tracker for Gait Test), which essentially consisted of taking videos of the patients on an iPhone [32][33][34]. A noninvasive motion-capture application allowed us to use this system in the general room for outpatients.…”

“…The patient setup required only several minutes [32], and, for the analysis, a minimal recording time of a few seconds was sufficient, as seen in the current study. Among recent gait-analysis methods, our system stands out for its superior availability and convenience in clinical applications [34]. Hence, we aimed to establish a connection between the application of this measurement tool [32] and its meaningful clinical applications.…”

“…Hence, we aimed to establish a connection between the application of this measurement tool [32] and its meaningful clinical applications. The initial trial involved extracting two-dimensional characteristics of pathological gait [33], while the second trial delved into the use of AI for distinguishing various gait patterns [34]. This manuscript represents our third publication, focusing on sensing the pathology of the temporospatial aspect of patients' gait.…”