Using AI to contain COVID-19 and future epidemics in Malaysia and Sri Lanka with a focus on women, children, and underprivileged groups

 

With the emergency situation that arises with COVID-19, the intense containment strategies adopted by many countries had little or no consideration towards socio-economic ramifications or the impact on women, children, socioeconomically underprivileged groups. The existence of many adverse impacts raises questions on the approaches taken and demands proper analysis, scrutiny and review of the policies. Therefore, a framework was developed using the artificial intelligence (AI) techniques to detect, model, and predict the behaviour of the COVID-19 pandemic containment strategies, understanding the socio-economic impact of these strategies on identified diverse vulnerable groups, and the development of AI-based solutions, to predict and manage a future spread of COVID or similar infectious disease outbreaks while mitigating the social and economic toil. Based on generated behaviour and movements, AI tools were developed to conduct contact tracing and socio-economic impact mitigation actions in a more informed, socially conscious and responsible manner in the case of the next wave of COVID-19 infections or a different future infectious disease.

Since the year 2020, coronavirus disease 2019 (COVID-19) has emerged as the dominant topic of discussion in the public and research domains. Intensive research has been carried out on several aspects of COVID-19, including vaccines, its transmission mechanism, detection of COVID-19 infection, and its infection rate and factors. The awareness of the public related to the COVID-19 infection factors enables the public to adhere to the standard operating procedures, while a full elucidation on the correlation of different factors to the infection rate facilitates effective measures to minimize the risk of COVID-19 infection by policy makers and enforcers. Hence, this paper aims to provide a comprehensive and analytical review of different factors affecting the COVID-19 infection rate. Furthermore, this review analyses factors which directly and indirectly affect the COVID-19 infection risk, such as physical distance, ventilation, face masks, meteorological factor, socioeconomic factor, vaccination, host factor, SARS-CoV-2 variants, and the availability of COVID-19 testing. Critical analysis was performed for the different factors by providing quantitative and qualitative studies. Lastly, the challenges of correlating each infection risk factor to the predicted risk of COVID-19 infection are discussed, and recommendations for further research works and interventions are outlined.

The study provided an overview of changes in the educational system due to the COVID-19 pandemic among engineering undergraduates of Sri Lanka. Results show that students’ attendance in online classes improved over time compared to the initial pandemic period. Nearly 50% of students’ family income was impacted- either stopped or reduced due to the pandemic. Most students have issues regarding computing devices, internet connectivity, and the home environment, which are not conducive to learning at home. Under normal circumstances, engineering undergraduates in Sri Lanka have high exposure to modern technology and a diversity of instructional delivery, hence this student cohort was chosen for the study.

A handshake interaction localization model in real-time that may help mitigate the threat for transmitting COVID-19, is presented using computer vision in a non-intrusive technique. A real-time detection model (using YOLO/you only look once) is proposed to identify handshake interactions in realistic scenarios. YOLO can detect multiple interactions in a single frame. The model can be applied to public spaces to identify handshake interactions. The study is the first to use a human interaction localization model in a multi-person setting. YOLO is a convolutional neural network (CNN) for object detection in real-time.

Using data from the Epidemiological Department of Sri Lanka, a cluster analysis was carried out based on COVID-19 data and demographic data of districts, towards developing a mathematical model that can identify and describe socio-economic factors related to pandemic measures. Population and population density, monthly expenditure, and education level are suggested as main factors for policy makers consideration. Findings can support future evidence-based COVID-19 policies, and further utilized as a foundation for other epidemiological models. A challenge in the study was the presumed disparity between actual COVID-19 cases and observed COVID-19 cases, thereby depicting an inaccurate measure of COVID-19 severity.