Theory of Mind Helps to Predict Neurodegenerative Processes in Parkinson's Disease

摘要

Normally, it takes many years of theoretical and clinical training for a physician to be the movement disorder specialist. It takes additional multiple years of the clinical practice to handle various "non-typical" cases. The purpose of our study was to predict neurodegenerative disease development by abstract rules learned from experienced neurologists. Theory of mind (ToM) is human's ability to represent mental states such as emotions, intensions or knowledge of others. ToM is crucial not only in human social interactions but also is used by neurologists to find an optimal treatment for patients with neurodegenerative pathologies such as Parkinson's disease (PD). On the basis of doctors' expertise, we have used supervised learning to build AI system that consists of abstract granules representing ToM of several movement disorders neurologists (their knowledge and intuitions). We were looking for similarities between granules of patients in different disease stages to granules of more advanced PD patients. We have compared group of 23 PD with attributes measured three times every half of the year (G1V1, G1V2, G1V3) to other group of 24 more advanced PD (G2V1). By means of the supervised learning and rough set theory we have found rules describing symptoms of G2V1 and applied them to G1V1, G1V2. and G1V3. We have obtained the following accuracies for all/speed/emotion/cognition attributes: G1V1: 68/59/53/72%; G1V2: 72/70/79/79%; G1V3: 82/92/71/74%. These results support our hypothesis that divergent sets of granules were characteristic for different brain's parts that might degenerate in non-uniform ways with Parkinson's disease progression.