Emotional discourse analysis of COVID-19 patients and their mental health: A text mining study

• Published in: PloS one Vol. 17; no. 9; p. e0274247
• Main Authors: Deng, Yu, Park, Minjun, Chen, Juanjuan, Yang, Jixue, Xie, Luxue, Li, Huimin, Wang, Li, Chen, Yaokai
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 16.09.2022; Public Library of Science (PLoS)
• Subjects: Analysis; Automatic control; Biology and Life Sciences; Care and treatment; Coronaviruses; COVID-19; Data Mining; Emotional factors; Emotions; Emotions - physiology; Fear; Happiness; Humans; Medicine and Health Sciences; Mental Health; Narratives; Patients; Social Sciences; Social support; China
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0274247

COVID-19 has caused negative emotional responses in patients, with significant mental health consequences for the infected population. The need for an in-depth analysis of the emotional state of COVID-19 patients is imperative. This study employed semi-structured interviews and the text mining method to investigate features in lived experience narratives of COVID-19 patients and healthy controls with respect to five basic emotions. The aim was to identify differences in emotional status between the two matched groups of participants. The results indicate generally higher complexity and more expressive emotional language in healthy controls than in COVID-19 patients. Specifically, narratives of fear, happiness, and sadness by COVID-19 patients were significantly shorter as compared to healthy controls. Regarding lexical features, COVID-19 patients used more emotional words, in particular words of fear, disgust, and happiness, as opposed to those used by healthy controls. Emotional disorder symptoms of COVID-19 patients at the lexical level tended to focus on the emotions of fear and disgust. They narrated more in relation to self or family while healthy controls mainly talked about others. Our automatic emotional discourse analysis potentially distinguishes clinical status of COVID-19 patients versus healthy controls, and can thus be used to predict mental health disorder symptoms in COVID-19 patients.