Correlations between Motor Symptoms across Different Motor Tasks, Quantified via Random Forest Feature Classification in Parkinson’s Disease

Abstract

BackgroundObjective assessments of Parkinson’s disease (PD) patients’ motor state using motion capture techniques are still rarely used in clinical practice, even though they may improve clinical management. One major obstacle relates to the large dimensionality of motor abnormalities in PD. We aimed to extract global motor performance measures covering different everyday motor tasks, as a function of a clinical intervention, i.e., deep brain stimulation (DBS) of the subthalamic nucleus.MethodsWe followed a data-driven, machine-learning approach and propose performance measures that employ Random Forests with probability distributions. We applied this method to 14 PD patients with DBS switched-off or -on, and 26 healthy control subjects performing the Timed Up and Go Test (TUG), the Functional Reach Test (FRT), a hand coordination task, walking 10-m straight, and a 90° curve.ResultsFor each motor task, a Random Forest identified a specific set of metrics that optimally separated PD off DBS from healthy subjects. We noted the highest accuracy (94.6%) for standing up. This corresponded to a sensitivity of 91.5% to detect a PD patient off DBS, and a specificity of 97.2% representing the rate of correctly identified healthy subjects. We then calculated performance measures based on these sets of metrics and applied those results to characterize symptom severity in different motor tasks. Task-specific symptom severity measures correlated significantly with each other and with the Unified Parkinson’s Disease Rating Scale (UPDRS, part III, correlation of r2 = 0.79). Agreement rates between different measures ranged from 79.8 to 89.3%.ConclusionThe close correlation of PD patients’ various motor abnormalities quantified by different, task-specific severity measures suggests that these abnormalities are only facets of the underlying one-dimensional severity of motor deficits. The identification and characterization of this underlying motor deficit may help to optimize therapeutic interventions, e.g., to “automatically” adapt DBS settings in PD patients.

Motion Biomarkers Showing Maximum Contrast Between Healthy Subjects and Parkinson's Disease Patients Treated With Deep Brain Stimulation of the Subthalamic Nucleus. A Pilot Study

“…In a previous study, we showed that machine-learning approaches significantly correlate with known clinical measures such as the UPDRS (Kuhner et al, 2017).…”

“…There is already evidence that jerk is abnormal in PD (Teulings et al, 1997;Hogan and Sternad, 2009). Other methods of data reduction by feature extraction involve signal processing methods, e.g., wavelet analysis (Joshi et al, 2017), stochastic models, like the Hidden Markov Model (Joshi et al, 2017), or machine-learning algorithms (Wouda et al, 2016), i.e., using Random Forests (Wahid et al, 2015;Kuhner et al, 2016Kuhner et al, , 2017.…”

“…When experimenting with machine-learning algorithms for feature extraction in PD subjects (Kuhner et al, 2016(Kuhner et al, , 2017, we came up with a very simple question that machine-learning classification methods alone cannot answer, namely, which signals or features best describe the difference between healthy subjects on the one hand and PD patients on the other hand.…”

Optimizing Clinical Assessments in Parkinson's Disease Through the Use of Wearable Sensors and Data Driven Modeling

“…For instance, these models have been used to predict the optimal stimulation frequency after the DBS implantation surgery based on the Unified Parkinson's disease Rating Scale (UPDRS III) scores collected at or before surgery (Khojandi et al, 2017 ) These models have also shown promise when used on the data collected from wearable sensors. Kuhner et al ( 2017 ) captured 3D inertial measurements from a motion capture suit consisting of magnetometers, accelerometers, and gyroscopes in PD patients with DBS switched-off or -on, as well as healthy controls. Utilizing an RF model with probability distribution on data derived from various clinical tasks, they demonstrated the ability to detect PD patients off DBS from healthy subjects with high sensitivity and specificity.…”

Application of Normalized Compression Distance and Lempel-Ziv Jaccard Distance in Micro-electrode Signal Stream Classification for the Surgical Treatment of Parkinson’s Disease

“…DBS technique has been known for 30 years, and over the past few years a number of interdisciplinary works concerning the use of machine learning methods have appeared, extending its functionality. Kuhner et al (2017) used Random Forests with probability distributions to detect abnormal motor behaviour of PD patients performing several different motor tasks in two clinical conditions (DBS switch: off, on). Trevathan et al (2017) Shamir et al (2015) proposed a clinical support decision system to provide effective stimulation and adequate drug dosages, based on three machine learning methods, which included Support Vector Machines, Naïve Bayes, and Random Forest.…”