Utilization of Digital Network Learning and Healthcare for Verbal Assessment and Counselling During Post COVID-19 Period

Abstract

The COVID-19 pandemic has created a pressing need to integrate information from disparate sources to assist decision-makers. Social media is important in this respect; however, to make sense of the textual information it provides and automate large amounts of data, natural language processing methods are needed. Social media posts are often noisy, yet they may provide valuable insights regarding the severity and prevalence of the disease in the population. In particular, machine learning techniques for triage and diagnosis could better understand what social media may offer in this respect. In an effort that provides investigators and some other relevant individuals with new knowledge on the signs, seriousness, and incidence of COVID-19, this study intends to construct an end-to-end natural language processing pipeline for assessment and identification of COVID-19 using patient-authored status updates. The information retrieval pipeline retrieves COVID-19 ailments and similar theories such as severity, duration, negations, and bodily parts from patients’ messages using potential artificial fields. In the next step of the pipeline, an unsupervised rule-based algorithm establishes relationships between ideas. Following that, the retrieved concepts and relationships create two alternative adversarial examples of each post. These vectors are used individually to create gradient boosting machine learning that categorizes patients into three groups and diagnoses COVID-19. When the models are trained on human labelled data, we report Macro- and Micro-averaged F1 scores in the range of 71–96 percent and 61–87 percent, respectively, to triage and diagnose COVID-19. Our results show that when the models are trained using projected labels from notion extraction and directive classifiers, identical efficiency may be attained, resulting in end-to-end machine learning. We also compare the most common symptoms disclosed in another COVID-19 dataset to the essential traits identified by our diagnostic machine learning models. We discovered that the most important characteristics are not usually the most common. Our preliminary findings demonstrate that a machine learning pipeline can automatically triage and diagnose COVID-19 patients using natural language narratives, allowing us to provide further knowledge on the disease’s seriousness and prevalence via the lenses of social networking sites.