Data mining using SPECT can predict neurological symptom development in Parkinson´s patients

We have compared in Parkinson´s diseases patients neurological data with the local cerebral blood flow measured by the Single-Photon Emission Computed Tomography. Most of our patients underwent Deep Brain Stimulation surgery or were qualified for one in relation to the advanced disease progression. Local cerebral blood flow in different areas has correlated to the Unified Parkinson´s Disease Rating Scale (UPDRS). We have used two different data mining methods: WEKA and Rough Set Exploration System to explore these correlations. We have demonstrated that cerebral blood flow changes gave good predictions for the UPDRS IV (84 %) that suggest that a general state of Parkinson Disease are stronger related to the cerebral blood flow than to only motor symptoms.