Feature selection in Parkinson´s disease: A rough sets approach

Parkinson´s disease is a neurodegenerative disorder with a long time course and a significant prevalence, which increases significantly with age. Although the etiology is currently unknown, the disease presents with neurodegeneration of regions of the basal ganglia. the onset occurs later in life, and the disease progresses slowly. The disease is diagnosed clinically, requiring the identification of several factors such as distal resting tremor, rigidity, and bradykinesia. The common thread throughout the range of symptoms is motor dysfunction, and recent reports have focused on dysphonia, the impairment in voice production as a diagnostic measure. In this paper, a number of features associated with speech have been collected through clinical studies from both healthy and people with Parkinson´s (PWP) and analysed in order to determine if one or more of them can be used to diagnose PWP. The feature set is analysed using the rough sets paradigm, which maps feature vectors associated with objects onto decision classes. The results from applying rough sets is a set of rules that map features via rules into a decision support system - performing classification of objects. the results FOM this study indicate that a subset of typical voice derived features is adequate to differentiate healthy from PWP with 100% accuracy. These result are important in that they imply that a diagnosis can be automated and performed remotely. This work will be extended to determine if this approach can be utilised with the same effectiveness for the diagnosis of parkinsonism disorders - a collection-diseases with Parkinson´s like symptoms.