Rules Determine Therapy-Dependent Relationship in Symptoms Development of Parkinson's Disease Patients

摘要

We still do not have cure for neurodegenerative disorders (ND) such as Parkinson's disease (PD). Recent findings demonstrated that neurodegenerative processes related to ND have long periods without symptoms that effects lack in effective therapy when ND is diagnosed. Neurologists estimate PD progression on the basis of their tests: Hoehn and Yahr (H&Y) and Unified Parkinson's Disease Rating (UPDRS) scales, but results of these tests are partly school-dependent. We have previously proposed that eye movement tests can give objective and precise measure of the PD progression. In this study, we have recorded reflexive saccades in patients with different disease stages and with different treatments. We put together patients' demographic data, results of neurological and eye movements' tests. In order to estimate effectiveness of different therapies we have placed data in information tables, discretized and used data mining (RS - rough set theory) and machine learning (ML). In end-effect we have obtained rules that determine longitudinal course of disease progression in different group of patients. By using of ML and RS rules obtained for the first visit of BMT/DBS/POP (only on medication/ recent DBS surgery/earlier DBS surgery) patients we have predicted UPDRS values in next year (two visits) with the global accuracy of 70% for both BMT visits; 56% for DBS, and 67, 79% for POP second and third visits. We have used rules obtained in BMT patients to predict UPDRS of DBS patients; for first session DBSW1: global accuracy was 64%, for second DBSW2: 85% and the third DBSW3: 74% but only for DBS patients during stimulation-ON. We could not predict UPDRS in DBS patients during stimulation-OFF visits and in all conditions of POP patients. We have compared rules in BMT patients with POP group and found many contradictory rules. It means that long-term brain electrical stimulation has changed brain mechanisms.