Granular Computing (GC) Demonstrates Interactions Between Depression and Symptoms Development in Parkinson's Disease Patients.

There is high frequency incidence of depressive symptoms in neurodegenerative diseases (ND) but reasons for it is not well understood. Parkinson’s disease (PD) is often evoked by strong emotional event and related to reduced level of dopamine (reward hormone). Similarly to PD, in older (over 65 year of age) subjects with late onset Alzheimer’s disease (LOAD) have first symptoms related to depression (95%). Present work is devoted to the question if evaluation of depression can help to predict PD symptoms? We have gathered results of: neurological (disease duration, values of Unified Parkinson’s Disease Rating Scale (UPDRS)), neuropsychological (depression – Beck test, PDQ39 (life quality), Epworth (sleep problems)) and eye movement (RS – reflexive saccadic) tests. We have tested 24 PD patients only with medical treatment (BMT-group), and 23 PD with medical and recent DBS (deep brain stimulation DBS-group), and 15 older DBS (POP-group) treatments during one and half year with testing every six months (W1, W2, W3). From rules found with help of GC (RST-rough set theory) in BMTW1 (patients BMT during first visit W1) we have predicted UPDRS in BMTW2 and BMTW3 with accuracies (acc.) 0.765 (0.7 without Beck result) and 0.8 (0.7 without Beck result). By using BMTW1 rules we could predict disease progression (UPDRS) of another group of patients – DBSW1 group with accuracy of 0.765 but not DBSW2/W3 patients. By using DBSW2 rules we could predict UPDRS of DBSW3 (acc. = 0.625), POPW1 (acc. = 0.77), POPW2 (acc. = 0.5), POPW3 (acc. = 0.33). By adding depression attribute and by using GC we could make better predictions of disease progressions in many different groups of patients than without it.