Characterization and classification of Parkinson’s disease patients based on symbolic dynamics analysis of heart rate variability

Abstract Background: Parkinson’s disease (PD) is a chronic and progressive neurodegenerative disorder characterized by deterioration of the substantia nigra, resulting in a deficiency of dopamine. PD is considered a movement disorder associated with numerous non-motor symptoms related to Autonomic Nervous System failures which can precede the motor ones. Therefore, their awareness could be helpful  in the diagnosis of PD at an early stage. Methods: Heart Rate Variability (HRV) is assessed by time and frequency domain indices, and by nonlinear indices based on symbolic dynamics and multiscale symbolic entropy. The features obtained were used to classify between PD patients and control volunteers using a support vector machine. Volunteers performed cardiovascular autonomic reflex tests: active standing, post- hyperventilation and controlled breathing. Results: Temporal and frequency indices showed significantly lower values in PD patients compared to control volunteers. Symbolic dynamics and multiscale symbolic entropy results suggest a decrease in the complexity of the HRV signal in PD patients, in contrast with a more variable pattern of words for control volunteers. During controlled breathing differences between groups were found with most of the indices computed. Additionally, classification process achieves good separability during cardiorespiratory maneuvers (>95% of accuracy) and features based on symbolic dynamics showed high discrimination between groups. Conclusions: The results found in this work suggest that the proposed methodological approach can classify PD patients in an early disease stage from healthy controls and give additional information about the cardiorespiratory system, which could be useful for diagnosis and follow up of PD patients.