Power-law temporal autocorrelation of activity reflects severity of parkinsonism

Abstract

We aimed to obtain a reliable, objective scale representing disease severity for appropriate management of patients with Parkinson's disease (PD). Nineteen patients with PD at the Department of Neurology, Tokyo University Hospital, were classified into mild (n = 10) or severe groups (n = 9) depending on their Hoehn-Yahr scores, and wore accelerometers on their wrists for more than 6 consecutive days. During this time we monitored their subjective assessments of symptom severity and analyzed the power-law exponents (alpha) for local maxima and minima of fluctuations in the activity time series. Statistical comparisons were made between the severe and mild groups and of individual patients on "good condition" and "bad condition" days, as well as between days before and after antiparkinsonism medication. In all patients, the alpha for local maxima was always lower when parkinsonism was mild than when severe. Presence of tremor did not influence the alpha for local maxima. As the lower alpha value for local maxima of fluctuations in activity records reflects more frequent switching behavior from low to high physical activities or the severity of akinesia, actigraph monitoring of parkinsonism, and analysis of its power-law correlation may provide useful objective information for controlling parkinsonism in outpatient clinics and for evaluating new antiparkinsonism drugs.