The Impact of COVID-19 on Parkinson’s Disease Patients from Social Networks

Abstract

AbstractMultiple examples of unsafe and incorrect treatment recommendations shared everyday. The challenge, however, to provide efficient, credible, and quick relevant access to reliable insight. Providing computational tools for Parkinson’s Disease (PD) using a set of data-objects that contain medical information is very desirable for alleviating the symptoms that can help to discover the risk of this disease at an early stage. In this paper, we propose an automatic CNN-clustering aspect-based identification method for drug mentions, events, treatments from daily PD narratives digests. Therefore, a BiLSTM-based Parkinson classifier is developed regarding both varied emotional states and common senses reasoning, which further used to seek the impactful COVID-19 insights. The embedding strategy characterized polar facts through concept-level distributed biomedical representation associated with real-world entities, which are operated to quantifying the emotional state of the speaker context in which aspect are extracted. We conduct comparisons with neural networks state-of-art algorithms and biomedical distributed systems. Finally, as a result, the classifier achieves an accuracy of 85.3%, and facets of this study may used in many health-related concerns such as: Analyzing change in health status, unexpected situations or medical conditions, and outcome or effectiveness of a treatment. KeywordsCOVID-19 pandemicNeural networksParkinson’s diseaseSentiment analysisSocial networks