BERT Learns From Electroencephalograms About Parkinson’s Disease: Transformer-Based Models for Aid Diagnosis

Medicine is a complex field with highly trained specialists extensive knowledge that continuously needs updating. Among them all, those who study the brain can perform tasks due to structure of this organ. There are neurological diseases such as degenerative ones whose diagnoses essential in very early stages. Parkinson’s disease one them, usually having confirmed diagnosis when it already developed. Some physicians have proposed using electroencephalograms non-invasive method for prompt diagnosis. The problem these tests data analysis relies on clinical eye experienced professional, which entails situations escape human perception. This research proposes use deep learning techniques combination develop These models demonstrated their good performance managing massive amounts data. Our main contribution apply from Natural Language Processing, particularly an adaptation BERT models, being last milestone area. model choice similarity between texts and be processed sequences. Results show best uses 64 channels people without resting states finger-tapping tasks. In terms metrics, has values around 86%.