Data mining and neural network simulations can help to improve deep brain stimulation effects in parkinson’s disease

Parkinson’s Disease (PD) is primary related to substantia nigra degeneration and, thus, dopamine insufficiency. L-DOPA as a precursor of dopamine is the standard medication in PD. However, disease progression causes L-DOPA therapy efficiency decay (on-off symptom fluctuation), and neurologists often decide to classify patients for DBS (Deep Brain Stimulation) surgery. DBS treatment is based on stimulating the specific subthalamic structure: subthalamic nucleus (STN) in our case. As STN consists of parts with different physiological functions, finding the appropriate placement of the DBS electrode contacts is challenging. In order to predict the neurological effects related to different electrodecontact stimulations, we have tracked connections between the stimulated part of STN and the cortex with the help of diffusion tensor imaging (DTI). By changing a contacts number and amplitude of stimulus (proportional in size to stimulated area), we have determined connections to cortical areas and related neurological effects. We have applied data mining methods to predict which contact (and at what amplitude) should be stimulated in order to improve a particular symptom. We have compared different data mining methods: Wekas Random Forest classifier and Rough Set Exploration System (RSES). We have demonstrated that the Weka classifier was more accurate when predicting the effects of stimulations on general neurological improvements, while RSES was more accurate when using specific neurological symptoms. We have simulated other effects of stimulation related to the interruption of pathological oscillation in the basal ganglia found in PD. Our model represents possible STN neural population with inhibitory and excitatory connections that have pathologically synchronized oscillations. High-frequency electrical stimulation has interrupted synchronization. something that is also observed in PD patients.