Parameters | AUROC | 95% CI | P-value |
---|
Area |
Over 1.5 METs | 0.956 | 0.913–0.999 | 0.21 |
Over 2 METs | 0.932 | 0.884–0.981 | 0.011 |
Over 2.5 METs | 0.859 | 0.785–0.933 | < 0.001 |
Over 3.0 METs | 0.677 | 0.598–0.756 | < 0.001 |
Time |
Over 1.5 METs | 0.959 | 0.918–1.000 | reference |
Over 2 METs | 0.943 | 0.897–0.989 | 0.062 |
Over 2.5 METs | 0.872 | 0.799–0.946 | < 0.01 |
Over 3.0 METs | 0.678 | 0.595–0.760 | < 0.001 |
Peaks |
Over 1.5 METs | 0.942 | 0.881–1.000 | 0.16 |
Over 2 METs | 0.930 | 0.862–0.998 | 0.041 |
Over 2.5 METs | 0.870 | 0.782–0.957 | 0.010 |
Over 3.0 METs | 0.713 | 0.622–0.804 | < 0.001 |
- AUROC Area under receiver operating curve, CI Confidence interval; MET, metabolic equivalents of task. (A) We compared the performance o f METs parameters as predictive models for the “On” state, with estimates of the area under the receiver operating characteristic curve (AUROC). The performance of time over 2 METs (AUROC: 0.779, 95% CI: 0.733–0.824) was higher than that of time and peaks over 2.5 METs and was significantly higher than the others. (B) We compared the performance of METs parameters as predictive models for the “dyskinesia” state, with estimates of AUROC. The performance of time over 1.5 METs (AUROC: 0.959, 95% CI: 0.918–1.000) was higer than that of area and peaks over 1.5 METs and time over 2 METs, and significantly higher than the others