Accelerometric Classification of Resting and Postural Tremor Amplitude
Christina van der Linden,
Thea Berger,
Gregor A. Brandt,
Joshua N. Strelow,
Hannah Jergas,
Juan Carlos Baldermann,
Veerle Visser-Vandewalle,
Gereon R. Fink,
Michael T. Barbe,
Jan Niklas Petry-Schmelzer,
Till A. Dembek
Affiliations
Christina van der Linden
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Thea Berger
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Gregor A. Brandt
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Joshua N. Strelow
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Hannah Jergas
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Juan Carlos Baldermann
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Veerle Visser-Vandewalle
Department of Stereotactic and Functional Neurosurgery, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Gereon R. Fink
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Michael T. Barbe
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Jan Niklas Petry-Schmelzer
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Till A. Dembek
Department of Neurology, University Hospital Cologne and Faculty of Medicine, University of Cologne, 50937 Cologne, Germany
Clinical rating scales for tremors have significant limitations due to low resolution, high rater dependency, and lack of applicability in outpatient settings. Reliable, quantitative approaches for assessing tremor severity are warranted, especially evaluating treatment effects, e.g., of deep brain stimulation (DBS). We aimed to investigate how different accelerometry metrics can objectively classify tremor amplitude of Essential Tremor (ET) and tremor in Parkinson’s Disease (PD). We assessed 860 resting and postural tremor trials in 16 patients with ET and 25 patients with PD under different DBS settings. Clinical ratings were compared to different metrics, based on either spectral components in the tremorband or pure acceleration, derived from simultaneous triaxial accelerometry captured at the index finger and wrist. Nonlinear regression was applied to a training dataset to determine the relationship between accelerometry and clinical ratings, which was then evaluated in a holdout dataset. All of the investigated accelerometry metrics could predict clinical tremor ratings with a high concordance (>70%) and substantial interrater reliability (Cohen’s weighted Kappa > 0.7) in out-of-sample data. Finger-worn accelerometry performed slightly better than wrist-worn accelerometry. We conclude that triaxial accelerometry reliably quantifies resting and postural tremor amplitude in ET and PD patients. A full release of our dataset and software allows for implementation, development, training, and validation of novel methods.