Early Detection of Parkinson’s Disease using Voice and Motion Data

Parkinson’s disease (PD) is a degenerative disorder of the nervous system that quite often results in speech, movement, and fine motor functions getting impaired and the gradual decline of an individual’s quality of life. The importance of early detection of PD cannot be overemphasized as it helps to facilitate the right time medical intervention and slow down the disease progression. This research introduces a non, invasive, inexpensive, and easy, to, understand early detection system that uses data obtained from voice, motion, and handwriting to figure out neurological problems that the patient might not have identified but which are noticeable in the early stages of the disease. Parkinson’s disease, related speech abnormality may be lessened pitch change, monotonous speech, and vocal tremor whereas motion problems are accompanied with tremors, slow movements, irregular gait, and loss of motor coordination. Analysis of handwriting additionally unveils micrographia and irregular stroke dynamics that are associated with fine motor dysfunction. This paper focuses on the extraction of acoustic features, for example, Mel, Frequency Cepstral Coefficients (MFCCs), as well as motion, related temporal features, and handwriting trajectory features for further analysis. These deep learning architectures, for example, Convolutional Neural Networks (CNNs) as well as Long Short, Term Memory (LSTM) networks, are used to understand the spatial and temporal patterns present in the multimodal data. The ultimate goal of the proposed system is to be able to provide a reliable early screening tool as well as to help the healthcare professionals in the early diagnosis of Parkinson’s disease.