Fig. 3: Univariable correlation and multivariable linear regression of weekly tremor measures with clinical tremor scores.

In a Spearman’s correlation coefficients are shown with their significance level (*p < 0.05, **p < 0.01, ***p < 0.001, corrected using the false discovery rate method). Correlations with the rest tremor severity (3.17) and constancy (3.18) scores in the device-sided arm were assessed for ON and OFF scores separately. The other scores were averaged over ON and OFF motor states. In b the beta coefficients of the multivariable linear regression are shown with their 95% confidence intervals. Here, each weekly tremor measure was used as outcome and MDS-UPDRS tremor scores (averaged over ON and OFF conditions) as independent predictors. Beta coefficients were standardized by dividing them by the standard deviation of the weekly tremor measure, so that they represent the expected change in the measure (in standardized units) due to an increase in each MDS-UPDRS score of 1, with all other scores unchanged. The rest tremor constancy score (3.18) and patient-reported tremor score (2.10) were not included in b since these were not assessed for each body part separately. 3.15 = postural tremor severity, 3.16 = kinetic tremor severity, 3.17 = rest tremor severity, 3.18 = rest tremor constancy, DS device-sided, NDS non-device-sided.