Parkinson's disease (PD) is the second-most common neurodegenerative disease and affects about 2-3% of the population of age 65 years or over, worldwide. One of the symptoms that often occurs in patients with PD is depression in about 40-50% of them and is very common in early stages of the development of PD. Numerous studies have been conducted with various results in identifying risk factors for depression in PD patients. While, the mechanism of depression is not yet known in detail in PD patients. In this study, a decision tree method was used to differentiate PD patients who underwent depression from those who did not and identify risk factors associated with the depression. We propose the Synthetic Minority Over-sampling TEchnique (SMOTE) to handle imbalanced class in the data. Data on 257 patients with early stage of PD in the Parkinson's Progression Markers Initiative (PPMI) database were used. The overall important risk factors associated with depression in patients with early-stage of PD are alpha synuclein (α-syn) levels, gender, SEADL (Schwab & England - Activities on Daily Living) score, STAI (State & Trait Anxiety Inventory) - State score, putamen binding ratio on the left side of the brain, RBDSQ (REM Sleep Behaviour Disorder-Questionnaire) score, and age when diagnosed with PD. The accuracy, precision, and recall of the model are 95.18%, 92.15%, and 94.12%, respectively. Moreover, the AUC and F1 score are 0.949, and 0.9312, respectively, supporting the high accuracy of the resulting model.
|Journal||Journal of Physics: Conference Series|
|Publication status||Published - 7 Jan 2021|
|Event||10th International Conference and Workshop on High Dimensional Data Analysis, ICW-HDDA 2020 - Sanur-Bali, Indonesia|
Duration: 12 Oct 2020 → 15 Oct 2020