Fig. 1

Data processing description and example results. a Data processing flow from measurement to analysis. Patients wore a three-axis accelerometer on the waist for more than 10 h per day and METs were measured every 10 s. METs data were extracted and lined up over time (transition of METs). METs data were then arranged from highest to lowest values (contribution of METs). Next, percentile values on a day were extracted (in this example case, the 95th percentile value was 2.8 METs). Finally, time, area and number of maximal values (peaks) over specific percentile values (in this example case, the 95th percentile value = 2.8 METs) and specific absolute METs values (e.g., 2 METs) were calculated. b Example results in a patient who had bradykinesia without involuntary movement (transition and contribution of METs, and 95 percentile value) are shown for the days on which the worst or best UPDRS-3 [On] ratings were observed. Transitions of METs appeared to be higher, longer, and more frequent on the patient’s best day compared with the patient’s worst day. c Example results of a patient with bradykinesia and involuntary movement (dyskinesia), suggesting a correlation between transition of METs with symptom diary states (“On” and “Off”) and the elevated baseline of METs (1.5–2.0 METs) during dyskinesia. The patient’s condition was checked every 1–2 h(s) objectively using UPDRS-3