Table 5.
Performance of random forest models trained on the magnitude features of each sensor type. Averaged AUROC values (with 95% CIs) for distinguishing between patients and controls with a motor score of zero and patients with a non-zero motor score.
| Sensor * | MDS-UPDRS III Tasks ** | |||||
|---|---|---|---|---|---|---|
| AR | OA | FN | HM | PS | FT | |
| Accelerometer | 0.65 (0.59–0.73) |
0.75 (0.68–0.82) |
0.66 (0.59–0.74) |
0.84 (0.76–0.92) |
0.51 (0.40–0.62) |
0.67 (0.57–0.77) |
| Gyroscope | 0.71 (0.64–0.78) |
0.77 (0.71–0.85) |
0.70 (0.63–0.77) |
0.72 (0.63–0.83) |
0.82 (0.74–0.91) |
0.77 (0.68–0.86) |
| Magnetometer | 0.57 (0.50–0.65) |
0.64 (0.56–0.72) |
0.63 (0.55–0.71) |
0.87 (0.80–0.95) |
0.51 (0.41–0.63) |
0.73 (0.64–0.83) |
| Accel + Gyro | 0.69 (0.62–0.76) |
0.77 (0.71–0.85) |
0.73 (0.66–0.80) |
0.86 (0.79–0.94) |
0.85 (0.77–0.93) |
0.71 (0.62–0.81) |
| Accel + Magn | 0.69 (0.62–0.76) |
0.75 (0.68–0.82) |
0.68 (0.61–0.76) |
0.90 (0.85–0.97) |
0.51 (0.41–0.63) |
0.73 (0.64–0.83) |
| Gyro + Magn | 0.70 (0.64–0.77) |
0.77 (0.71–0.84) |
0.68 (0.62–0.76) |
0.90 (0.84–0.97) |
0.80 (0.72–0.89) |
0.72 (0.63–0.82) |
| Accel + Gyro + Magn |
0.72 (0.65–0.79) |
0.78 (0.71–0.85) |
0.72 (0.65–0.79) |
0.92 (0.86–0.98) |
0.82 (0.74–0.90) |
0.75 (0.66–0.84) |
* Performance for each sensor type and all sensor combinations. Binary classification with motor scores {0} vs. {1} for AR and OA tasks, {0} vs. {1,2,3} for FN, HM, and FT tasks, and {0} vs. {1,2,2,3,4} for PS tasks. ** AUROC values with the best result in each column are highlighted in bold.