Fig. 5.

Correlations between the 7 metrics computed with deep learning algorithms and averaged MDS-UPDRS scores. F, frequency of the repetitions; ΔT_sweep, comparison of the duration of the three first/last sweeps; ΔA_sweep, comparison of the amplitude of the three first/last sweeps; ΔS_sweep, comparison of the speed of the three first/last sweeps; ΔT_fit, period variation, which represents the change in period computed from the fit parameters; ΔA_fit, amplitude variation, which represents the change in amplitude computed from the fit parameters; L, fatigue parameter which represents the maximal change in amplitude during the whole protocol. ns: not significant; ^*<0.05, ^**<0.005, ^***<0.0005 for the significance of the p-values. LR, linear regression; SC, Spearman correlation.