An Explainable Spatial-Temporal Graphical Convolutional Network to Score Freezing of Gait in Parkinsonian Patients

Abstract

Freezing of gait (FOG) is a poorly understood heterogeneous gait disorder seen in patients with parkinsonism which contributes to significant morbidity and social isolation. FOG is currently measured with scales that are typically performed by movement disorders specialists (i.e., MDS-UPDRS), or through patient completed questionnaires (N-FOG-Q) both of which are inadequate in addressing the heterogeneous nature of the disorder and are unsuitable for use in clinical trials The purpose of this study was to devise a method to measure FOG objectively, hence improving our ability to identify it and accurately evaluate new therapies. A major innovation of our study is that it is the first study of its kind that uses the largest sample size (>30 h, N = 57) in order to apply explainable, multi-task deep learning models for quantifying FOG over the course of the medication cycle and at varying levels of parkinsonism severity. We trained interpretable deep learning models with multi-task learning to simultaneously score FOG (cross-validated F1 score 97.6%), identify medication state (OFF vs. ON levodopa; cross-validated F1 score 96.8%), and measure total PD severity (MDS-UPDRS-III score prediction error ≤ 2.7 points) using kinematic data of a well-characterized sample of N = 57 patients during levodopa challenge tests. The proposed model was able to explain how kinematic movements are associated with each FOG severity level that were highly consistent with the features, in which movement disorders specialists are trained to identify as characteristics of freezing. Overall, we demonstrate that deep learning models’ capability to capture complex movement patterns in kinematic data can automatically and objectively score FOG with high accuracy. These models have the potential to discover novel kinematic biomarkers for FOG that can be used for hypothesis generation and potentially as clinical trial outcome measures.

Development of a Tremor Detection Algorithm for use in an Academic Movement Disorders Center

“…However, we know that signal processing approaches like correlation across body regions provide additional diagnostic insight for discriminating, for example, parkinsonian from orthostatic tremors [28]. With full body data, end-to-end machine learning approaches (e.g., [9]) have significant potential to discover these and other features automatically. Other more subtle tremor features like distractibility [28] seem more likely to be characterized in full body data.…”

“…We compared algorithm performance using a database of 2,272 recordings made during standard clinical exams of N = 50 arbitrarily selected clinic patients. Aspects of the testing paradigm have been described previously [9,20,21]; more detail is provided below. In 42 patients (84%) the primary diagnosis was either Parkinson’s disease or essential tremor.…”

“…We compared the ability of six separate processing to recreate clinician ratings of tremor presence/absence based on F 1 score, which is a harmonic mean of precision and recall [9]: …”

“…Despite this, technologies for identifying tremor have progressed towards standardization and clinical uptake very slowly [1]. In the research domain, various technologies measure human motion, including body-worn sensors [3,7,8], 3D motion capture [9,10], and -most recently -pose recognition from monocular video [11][12][13]. Digitizing tablets are often used for assessing tremor during tasks like spiral drawing [14,15] and for discriminating tremor from bradykinesia during finger tapping [16].…”

A Hybrid Protection Scheme for the Gait Analysis in Early Dementia Recognition

“…Regular walking is characterized by a rhythmic, fluid gait, without apparent efforts of joint movement, with freely swinging legs and an upright posture, accompanied by movements of the head, trunk, and arms [ 9 , 10 ]. Recent efforts show the potential of machine learning (ML) approaches for the early detection of neurodegenerative diseases through gait analysis [ 11 , 12 , 13 , 14 ], but there are many concerns about the security of these approaches [ 15 , 16 , 17 , 18 ]. In healthcare applications, data privacy is a significant challenge.…”