Affiliations: [a] Neurology Department, Istanbul Medeniyet University Medical Faculty, Istanbul, Turkey | [b] Department of Computer Engineering, Istanbul Medeniyet University, Istanbul, Turkey
Correspondence:
[*]
Corresponding author: Abdulkadir Koçer, Neurology Department, Medical Faculty, Istanbul Medeniyet University, Istanbul, Turkey. Tel.: +90 505 426 28 28; E-mail:abdulkadirkocer@yahoo.com
Abstract: OBJECTIVE: Diagnosis of Parkinson's Disease (PD) by analyzing the resting
tremor were much studied by using different accelerometer based methods,
however the quantitative assessment of Hoehn and Yahr Scale (HYS) score with a
machine learning based system has not been previously addressed. In this
study, we aimed to propose a system to automatically assess the HYS score of
patients with PD.
METHODS: The system was evaluated and tested on a dataset containing 55
subjects where 35 of them were patients and 20 of them were healthy
controls. The resting tremor data were gathered with the 3 axis
accelerometer of the Nintendo Wii (Wiimote). The clinical disability of the
PD was graded from 1 to 5 by the HYS and tremor was recorded twice from the
more affected side in each patient and from the dominant extremity in each
control for a 60 seconds period. The HYS scores were learned with Support
Vector Machines (SVM) from the features of the tremor data.
RESULTS: Thirty-two of the subjects with PD were classified correctly and 18
of the normal subjects were also classified correctly by our system. The
system had average 0.89 accuracy rate (Range: 81-100% changing according
to grading by HYS).
CONCLUSIONS: We compared quantitative measurements of hand tremor in PD
patients, with staging of PD based on accelerometer data gathered using the
Wii sensor. Our results showed that the machine learning based system with
simple features could be helpful for diagnosis of PD and estimate HYS score.
We believed that this portable and easy-to-use Wii sensor measure might also
be applicable in the continuous monitoring of the resting tremor with small
modifications in routine clinical use.
