4 minute read
Have you had COVID
It is observed that the pandemic and quarantine have a disadvantageous impact on mental health as an increase of psychiatric symptoms and of mental health problems in the general population is expected as a fallout of the measures since most health professionals in isolation units are seldom trained nor supported for their mental health care in Italy. Mental health services worldwide are not prepared to manage the short- and long-term consequences of pandemic. With a Coronavirus anxiety scale, [15] came up with a brief mental health screener to measure COVID-19 related anxieties premised on the basis that mental health issues of people impacted by the pandemic have not been adequately addressed by relevant authorities. The 5-item scale screener, based on 775 adults with anxiety over the virus demonstrated compact dependability and validity as elevated CAS scores were found to be connected with COVID-19 diagnosis, impairment, alcohol/drug coping, negative religious coping, extreme hopelessness, suicidal ideation etc. by discriminating well between persons with and without dysfunctional anxiety using an optimized cut score of ≥ 9 (90% sensitivity and 85% specificity). The work of [10] seeks to discover the implications of fear emotion on students' and teachers' technology adoption during COVID19 pandemic. The study adopts Google Meet as an educational social platform in private higher education institutes and data from the study were analyzed with partial least squares structural equation modeling (PLS-SEM) with machine learning supervised algorithm. The J48 classifier performed better in predicting the dependent variable and the introduction of the fear of COVID-19 was an improvement on existing literature which similarly seeks to understudy adoption of technologies amidst the pandemics. A cross-national longitudinal study to predict fear and perceived health towards COVID-19 was conducted by [10] using factors such as worrying about shortages in supplies, perceived vulnerability to disease (PVD) and sex etc. Result shows case counts does not elicit adaptive responses to environmental threats while [16] investigated the pervasiveness of nosophobia, and readiness of people to seek medical attention amidst the COVID-19 epidemic in Calabar of Cross River State in Nigeria. The study shows that nosophobia is associated with age and healthcare seeking attitude while gender and education seldom play significant role in its pathology. It also discovered that fear varied with respect to the type of diseases hence the need for sensitization of the public. In [17], the author identified exposure to avalanche of news coverage about diseases and its risks with recurrent exposure to people with severe illnesses as factors that trigger nosophobia amongst citizens of India and in [12], authors conceptualized and designed a novel COVID-19 stress scale to efficiently determine the stress severity of the virus by measuring fear of getting infected, fear of contact with contaminated articles or planes, disease-related xenophobic anxiety, socio-economic anxiety of the outbreak (e.g., loss of job), compulsive examination and reassurance-seeking regarding possible pandemic-related threats, and traumatic stress symptoms about the pandemic like hallucination and disturbing thought; hence the design of the Stress Scales to quantity the aforementioned features. The work of [18] investigated the predictors of COVID-19 fear and observed that incremental fear is related to risks around loved ones and anxieties related to their health while regular social media use is another fear threshold with respect to the pandemic. In [19], a topic modelling approach and sentiment analysis was implemented on information flow on twitter during the coronavirus outbreak using the LDA topic modelling preprocessor identifying the most relevant and accurate subjects related to the virus. The model cofirms the prevalence of fear in negative sentiments and the positive sentiments depicts trust in government establsihments to tame the trend. A noun only approach for topic modelling was the thrust of [20] by comparing three topic models trained on news datasets generated from news corpus, lemmatised scope of the news item, and noun only corpus. the experiemntal result shows that excluding all other words except now improved topic semantic occurrence.
III. RESEARCH METHODOLOGY
Advertisement
The research methodology framework implemented in this study is as presented in this section comprising of REPTree supervised machine learning and opinion mining through sentiment analysis and topic modelling. The data analytics is executed on the content of data captured from the opinions gathered from teaching and non-teaching staff of tertiary educational institutions. The study adopted LSI and LDA methods for the topic modelling and sentiment analysis encapsulated in a 6-phase approach as presented in Fig. 1 which includes data acquisition, data preprocessing, machine learning, Topic modelling, sentiment analysis, and result analysis phases.
A. Data Capturing A total of 2376 data features acquired through Google form survey tool, constituting 10232 text corpus, is captured for this study, representing opinions of teaching and non-teaching staff of twenty-one (21) TEIs in Nigeria. A total of 15-no feature attributes (survey responses) is acquired per respondent while the last question (SN 15 on Table 1) is an open-ended question to capture textual phrases for the topic modelling and sentiment analysis phases described in sections 0 and 0 respectively.
