The impact of social media messages on Parkinson’s disease treatment: detecting genuine sentiment in patient notes

Abstract

Parkinson’s Disease(PD), one of the most serious neurodegenerative diseases that known huge controversy on social networks. Following medical lexicons, few approaches have been extended to leverage sentiment information that obviously reflects the patient’s health status in terms of related-narratives observations. It is been crucial to analyze online narratives and detect sentiment in patients’ self-reports. In this paper, we propose an automatic concept-level neural network method to distilling genuine sentiment in patients’ notes as medical polar facts into true positives and true negatives. Towards building emotional Parkinsonism assisted method from Parkinson’s Disease daily narratives di- gests, we characterize polar facts of defined medical configuration space through distributed biomedical representation at the concept-level as- sociated with real-world entities, which are operated to quantifying the emotional status of the speaker context. We conduct comparisons with state-of-art neural networks algorithms and biomedical distributed systems. Finally, as a result, we achieve an 85.3% accuracy performance, and the approach shows a well-understanding of medical natural language concepts.