Comparison of Walking Protocols and Gait Assessment Systems for Machine Learning-Based Classification of Parkinson’s Disease

Abstract

Early diagnosis of Parkinson’s diseases (PD) is challenging; applying machine learning (ML) models to gait characteristics may support the classification process. Comparing performance of ML models used in various studies can be problematic due to different walking protocols and gait assessment systems. The objective of this study was to compare the impact of walking protocols and gait assessment systems on the performance of a support vector machine (SVM) and random forest (RF) for classification of PD. 93 PD and 103 controls performed two walking protocols at their normal pace: (i) four times along a 10 m walkway (intermittent walk-IW), (ii) walking for 2 minutes on a 25 m oval circuit (continuous walk-CW). 14 gait characteristics were extracted from two different systems (an instrumented walkway—GAITRite; and an accelerometer attached at the lower back—Axivity). SVM and RF were trained on normalized data (accounting for step velocity, gender, age and BMI) and evaluated using 10-fold cross validation with area under the curve (AUC). Overall performance was higher for both systems during CW compared to IW. SVM performed better than RF. With SVM, during CW Axivity significantly outperformed GAITRite (AUC: 87.83 ± 7.81% vs. 80.49 ± 9.85%); during IW systems performed similarly. These findings suggest that choice of testing protocol and sensing system may have a direct impact on ML PD classification results and highlight the need for standardization for wide scale implementation.

Turning Detection During Gait: Algorithm Validation and Influence of Sensor Location and Turning Characteristics in the Classification of Parkinson’s Disease

“…Of the gait and turning characteristics included in the random forest classifier, angular velocity was the most important in the classification of PD. We have previously used gait characteristics in the classification of PD [ 48 , 49 ], resulting in 73–93% accuracy. In addition, PLS-DA trained on the gait characteristics gave a classification accuracy of 70.42–88.73% (AUC: 78.4–94.5%) with a sensitivity of 72.84–90.12% and specificity of 60.3–86.89% [ 30 ].…”

Towards remote healthcare monitoring using accessible IoT technology: state-of-the-art, insights and experimental design

“…In brief, use of the continuous wavelet transform helps identify timings of the initial (heel strike) and final contact (toe off) for each step from the vertical acceleration of MEMS-based wearables, such as the AX3. The AX3 has been widely used for validated gait analysis studies in various clinical cohorts [ 86 – 89 ]. Those contact times coupled with the inverted pendulum model [ 90 ], which estimates change in height of the wearable due to attachment near the wearers’ centre of mass, provide pragmatic gait characteristics.…”

Digital Biomarkers for the Early Detection of Mild Cognitive Impairment: Artificial Intelligence Meets Virtual Reality

“…The great amount of data collected with VR and kinematics implies the need for a computational method of analysis that is able to extract meaning from a large amount of data. Thus, Machine Learning appears to be a viable solution, as shown by previous research employing this technique to discriminate between normal and pathological gait alteration and for diagnostic purposes as well (Pogorelc et al, 2012;Zhang and Wang, 2012;Eskofier et al, 2013;Costa et al, 2016;Mannini et al, 2016;Caldas et al, 2017;Farah et al, 2017;Ur Rehman et al, 2019). "How" to Analyze Them?…”

“…Including an evaluation where the elderly person him/herself performs IADL might be essential for establishing more precise criteria ( Díaz-Mardomingo et al, 2017 ; Seo et al, 2017 ). Several authors have tried to refine early MCI detection by combining two out of the three variables considered in this paper: either behavioral alteration (IADL, gait) within a VR environment ( Lee et al, 2003 ; Seo et al, 2017 ; Kim et al, 2019 ; Eraslan Boz et al, 2019 ), gait kinematics extracted and analyzed by means of ML, which will be discussed further ( Begg and Kamruzzaman, 2005 ; Pogorelc et al, 2012 ; Zhang and Wang, 2012 ; Eskofier et al, 2013 ; Akl et al, 2015 ; Costa et al, 2016 ; Mannini et al, 2016 ; Caldas et al, 2017 ; Farah et al, 2017 ; Ur Rehman et al, 2019 ), or ML techniques for predicting MCI evolution ( Filipovych and Davatzikos, 2011 ; Williams and Weakley, 2013 ; Moradi et al, 2014 , 2015 ; Bratić et al, 2018 ; Grassi et al, 2018 , 2019 ; Graham et al, 2020 ). Thus, to our best knowledge, this is the first paper proposing an integration of VR, gait kinematics, and ML in order to refine early detection of MCI following a dimensional approach in line with the most recent diagnostic systems and possibly providing information on disease progression.…”