COVID-19 caught the world off-guard. Amongst other countries, Sri Lanka and Malaysia had to implement various COVID-19 containment strategies to combat the rising number of cases, deaths, high case-fatality rate and high infectivity rate. Government measures have ranged from no containment — social-distancing — movement restrictions — lockdown — and even complete shut-down. In the midst of it all, some individuals and groups in Sri Lanka and Malaysia have experienced psychosocial stress, discomfort and discrimination due to some of the measures adopted. 

Some identified diverse groups, including women, children, and the socio-economically underprivileged affected, raised questions about the approaches taken and called for proper policy review. As part of our research, we intend to assess how COVID-19 has impacted these vulnerable groups, as well as how threat assessment, containment and socio-economic impacts can be analyzed through AI-informed strategies. We envision that the results and insights will be beneficial in formulating evidence-based responses to address not only the progression of the ongoing COVID-19 crisis, but also future disease outbreaks. 

Our research team consists of engineering, medical and social sciences academics from Sri Lanka and Malaysia, namely from the University of Peradeniya and the Institute of Policy Studies in Sri Lanka; and Universiti Tenaga Nasional, The National University of Malaysia, Management and Science University in Malaysia.

Our Approach

For the purposes of our research in Sri Lanka, we will assess the impact of the pandemic over four different time periods: before the lockdown; first wave (from 27.01.2020 to 03.10.2020); second wave (from 04.10.2020 to 14.04.2021); and the third wave (from 15.04.2021 to present). A survey will be conducted nationwide in Sri Lanka, involving roughly 3000 households. To enable more timely and accurate planning and decision-making, our team will leverage artificial intelligence to analyze, assess and generate insights from the survey. In addition, several AI techniques will be utilized to evaluate the effectiveness of COVID-19 containment strategies through pattern analysis of infection data.

Focused on forecasting transmissions and reducing the spread of the virus through policy and public health interventions, AI and data science models can help predict outbreak hotspots, including providing relevant information on timing and population size. Insights gathered through these approaches can better inform government strategies and policies, particularly for effective allocation of limited resources. Whilst epidemiological tools such as contact tracing can help to limit the scope and spread of outbreaks, as part of our research we recognize the importance of oversight to check assumptions in modelling to ensure veracity, reliability, and accountability to address potential biases and harms.

Expected Outcomes

The project team will develop a set of data-driven, integrated tools to analyze the socio-economic impact of different containment strategies. Potential beneficiaries include policymakers, virologists, medical professionals and other relevant decision makers coordinating COVID-19 responses. Although our research mainly focuses on COVID-19, we envision that the methodologies and designs will be replicated to address other infectious disease outbreaks. Furthermore, as the research work will focus on women, children, and socially underprivileged groups, gender and inclusiveness will be intersectional. 

More specifically, we’ll be working towards the following outcomes:

• Socio-economic impact of the current wave of COVID-19 on the three groups identified;
• Broad database of socio-economic and ancillary data pertaining to the COVID-19 pandemic, highlighting the impact on the three groups identified;
• Data trajectories, stratified along job categories of the three groups;
• Specialized AI for each study population and corresponding behavioral trajectories from past and present data, labelled based on the occupation of the study population under multiple government containment strategies;
• AI model incorporating collision emulation and a spatio-temporal forecasting;
• AI model to provide the susceptibility of each group and their vulnerability (in terms of health and socio-economic impact);
• Impact Study, contact tracing strategy and threat level assessment incorporating local outbreak data; and 
• Evidence-based recommendations to inform decisions and actions at the policy level to minimize the disease spread.

As we progress and share our learnings, we hope our project will become a focal point for AI-based social, medical, and policy research in Sri Lanka, Malaysia, and other countries. In the coming months, we will share key findings and knowledge products on our website. We are thankful to Canada’s International Development Research Centre (IDRC) and the Swedish International Development Cooperation Agency (Sida) for supporting our research through the Global South AI4COVID program.