Surface Web Merits for SARS-CoV-2 Pandemic in Iraq

Abstract

Background: Data on SARS-CoV-2 from developing countries is not entirely accurate, demanding incorporating digital epidemiology data on the pandemic.

Objectives: To reconcile non-Bayesian models and artificial intelligence connected with digital and classical (non-digital) epidemiological data on SARS-CoV-2 pandemic in Iraq.

Results: Baghdad and Sulaymaniyah represented statistical outliers in connection with daily cases and recoveries, and daily deaths, respectively. Multivariate tests and neural networks detected a predictor effect of deaths, recoveries, and daily cases on web searches concerning two search terms, "كورونا" and "Coronavirus" (Pillai's Trace value=1, F=1106915.624, Hypothesis df=3, Error df=12, p-value<0.001, Partial Eta Squared=1). Using hierarchical clustering, we identified distinctive aggregates involving the Iraqi capital, Kurdistan region, and the south of Iraq. Three search terms were most prevalent among Iraqi web users, including "كورونا", "كوفيد-19", and "Coronavirus". Significant bivariate correlations were all positive except for those involving the search term "لقاح كورونا". Al-Muthanna governorate residents were least interested in data on SARS-CoV-2 vaccines.Methods: Our study design is longitudinal, for the period from 24 February 2020 to 25 September 2020. We retrieved data from the Iraqi Ministry of Health on the daily cases, recoveries, and deaths from SARS-CoV-2, and incorporated collateral data from Google Trends using five search terms, "Coronavirus", "كورونا", "COVID-19", "كوفيد-19", and "لقاح كورونا". The search terms "كورونا", "كوفيد-19", and "لقاح كورونا" represent the Arabic translations for "Coronavirus", "COVID-19", and "COVID-19 Vaccine". We implemented multivariate tests and machine learning to scrutinize the spatio-temporal trends of the pandemic in Iraq and interpret the causality influencing Iraqis to seek digital knowledge, via the web, on SARS-CoV-2.

Conclusion: Our analyses were triumphant in syncretizing non-Bayesian and machine learning models, using two forms of epidemiology data on the pandemic in Iraq. We opine that the current study is exquisite and precious for decision-makers at the Iraqi Ministry of Health.