Fig. 5From: Role of data measurement characteristics in the accurate detection of Parkinson’s disease symptoms using wearable sensorsEffect of magnitude-only features. Relationship between feature computation time and model performance (AUROC) for (a) Binary and (b) Multiclass models of tremor and bradykinesia. Shaded regions depict a 95% confidence interval of the mean AUROC. Features derived from the signal magnitude alone can achieve similar performance to combined tri-axial and magnitude features at a fraction of the computation timeBack to article page