Predicting Axial Impairment in Parkinson's Disease through a Single Inertial Sensor

Abstract

Background: Current telemedicine approaches lack standardised procedures for the remote assessment of axial impairment in Parkinson's disease (PD). Unobtrusive wearable sensors may be a feasible tool to provide clinicians with practical medical indices reflecting axial dysfunction in PD. This study aims to predict the postural instability/gait difficulty (PIGD) score in PD patients by monitoring gait through a single inertial measurement unit (IMU) and machine-learning algorithms.

Methods: Thirty-one PD patients underwent a 7-m timed-up-and-go test while monitored through an IMU placed on the thigh, both under (ON) and not under (OFF) dopaminergic therapy. After pre-processing procedures and feature selection, a support vector regression model was implemented to predict PIGD scores and to investigate the impact of L-Dopa and freezing of gait (FOG) on regression models.

Results: Specific time- and frequency-domain features correlated with PIGD scores. After optimizing the dimensionality reduction methods and the model parameters, regression algorithms demonstrated different performance in the PIGD prediction in patients OFF and ON therapy (r = 0.79 and 0.75 and RMSE = 0.19 and 0.20, respectively). Similarly, regression models showed different performances in the PIGD prediction, in patients with FOG, ON and OFF therapy (r = 0.71 and RMSE = 0.27; r = 0.83 and RMSE = 0.22, respectively) and in those without FOG, ON and OFF therapy (r = 0.85 and RMSE = 0.19; r = 0.79 and RMSE = 0.21, respectively).

Conclusions: Optimized support vector regression models have high feasibility in predicting PIGD scores in PD. L-Dopa and FOG affect regression model performances. Overall, a single inertial sensor may help to remotely assess axial motor impairment in PD patients.

Keywords: Levodopa; Parkinson’s disease; dimensionality reduction; feature extraction; freezing of gait (FOG); gait; machine learning; postural instability and gait difficulty score (PIGD); time up and go; wearable sensors.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1

Sketch of sensor position, together with an exploded view of the neMEMSI device.

Figure 2

Schematic of the Kalman filter, together with input and output. Raw gyroscope (a) and accelerometer (b) readings are input to the Kalman filter (c) to provide an estimate of orientation (d).

Figure 3

Absolute value of the CWT plotted as a function of time and frequency. Yellow zones represent the walking segments of the signal.

Figure 4

Orientation signal (blue) and walking bouts detected by the algorithm (orange).

Figure 5

Schematic of the analysis performed. The processing was performed for both pharmacological conditions, different populations, different sizes of the feature set, and different dimensionality reduction methods. ON: under dopaminergic therapy; OFF: not under dopaminergic therapy; FOG: patients with Parkinson’s disease and freezing of gait; FOG−: patients with Parkinson’s disease without freezing of gait; PCA: principal component analysis.

Figure 6

Correlation plot between the average principal harmonic height for the x-axis angular velocity (DH height 4) and PIGD score. Data and PIGD score refer to patients OFF (left) and ON (right) therapy. PIGD: postural instability and gait difficulty.

Figure 7

The effect of therapy on the regression model performances, in terms of correlation coefficient (left) and root mean square error (right). ON: under dopaminergic therapy; OFF: not under dopaminergic therapy.

Figure 8

Regression results for patients under different pharmacological condition. Data are plotted using a scatter plot and the regression line is reported as the best fit line. (a) Patients under dopaminergic therapy. (b) Patients not under dopaminergic therapy.

Figure 9

The effect of freezing of gait on the regression model performance, in terms of correlation coefficient (left) and root mean square error (right). Performance refers to patients under dopaminergic therapy. FOG+: patients with freezing of gait; FOG−: patients without freezing of gait.

Figure 10

The effect of freezing of gait on the regression model performance, in terms of correlation coefficient (left) and root mean square error (right). Performance refers to patients not under dopaminergic therapy. FOG+: patients with freezing of gait; FOG−: patients without freezing of gait.

Figure 11

Pearson correlation coefficient between features and PIGD score. DH ratio and DH frequency refer to the component θx; Min and vPeaks refer to the component ωx; RMSE and Range refer to the component αy.