Skip to main content

Table 2 Number of data clips scored for tremor and bradykinesia used in the supervised machine learning models

From: Role of data measurement characteristics in the accurate detection of Parkinson’s disease symptoms using wearable sensors

Score

No. Clips with Tremor Score (%)

No. Clips with Bradykinesia Score (%)

BioStampRC

Watch

BioStampRC

Watch

0

12,143 (73.8%)

10,485 (73.1%)

5487 (45.0%)

4845 (44.8%)

> 0

4302 (26.2%)

3854 (26.9%)

6697 (55.0%)

5979 (55.2%)

1

2684 (16.3%)

2346 (16.4%)

4764 (39.1%)

4213 (38.9%)

2

1274 (7.7%)

1233 (8.6%)

1835 (15.1%)

1676 (15.5%)

3

344 (2.0%)

275 (1.9%)

98 (0.8%)

90 (0.8%)

4

0 (0.0%)

0 (0.0%)

0 (0.0%)

0 (0.0%)

Total

16,445

14,339

12,184

10,824

  1. Number of 5-s clips for tremor and bradykinesia symptoms, by score and device type. Fewer clips are available with bradykinesia scores because not all tasks involved enough movement for the clinician to assess bradykinesia