COVIDaction Data Challenge Epi-Model Winners Announced
The COVIDaction Data Challenge has made two awards to organisations building or assisting in the uptake and use of epidemiological models, which remain one of the best tools for understanding the spread and impact of COVID-19.
Of the many high-quality epidemiological models submitted to the Data Challenge, two stood out for their unique approach to leveraging data and evidence that will benefit lower- and middle-income countries (LMICs). The applicants that will receive awards under the Data Challenge are:
- Bio Nano Consulting, in association with MRC-GIDA at Imperial College London, for the COVID-19 Scenario Analysis Tool. This leading epidemiological prediction tool covers 135 countries and is particularly useful to LMICs in the fight against COVID-19.
- The UN OCHA Centre for Humanitarian Data, in partnership with Johns Hopkins University Applied Physics Laboratory, for a sub-national model of COVID-19 spread and severity for countries experiencing humanitarian crises. The model enables humanitarian stakeholders to simulate a COVID-19 outbreak and see estimates for cases, severe cases (hospitalisations) and deaths.
Underscoring the importance of these investments, Magdalena Banasiak, Senior Innovation Adviser at DFID, who manages the COVIDaction programme, noted that, “Epidemiological models are needed to assess the spread of COVID-19 across the continent. They are particularly essential for LMICs where reliable testing data is lacking, providing a tool for predicting contagion patterns to determine the best course of action for a country’s population.”
A timely example of why these models are needed was observed on August 7, when the number of confirmed cases of COVID-19 in Africa surpassed one million. However, this figure is derived from testing data, which we know to be nearly universally inadequate outside of a few developed countries. Indeed, many epidemiological models projected that Africa likely crossed the one million case threshold as early as April. Relying on testing data alone would leave most African countries without a proper understanding of their actual disease burden.
In the absence of widespread testing, the supply of timely epidemiological data to LMICs, such as that facilitated by the two awardees, is vital as individuals are often less able to maintain quarantines or social distancing due to the nature of their economies and living environments. This makes data on potential hotspots more important, as noted in a Foreign Policy report, “… the [public policy] measures used in many developed countries could have adverse effects in low-income nations — potentially endangering more lives than they save.”
Sample actions that an LMICs could take, based on information gleaned from an epidemiological model, include targeted distribution or subsidising soap for handwashing and prohibiting large gatherings — measures that the evidence suggests will have a positive impact in suppressing the pandemic.
Overview of the Awardees
The Awardee: Bio Nano Consulting, in association with MRC-GIDA at Imperial College London
Bio Nano Consulting supports the Covidsim programme, a deterministic Susceptible, Exposed, Infectious, Recovered (SEIR) model, based on a modelling framework developed by Imperial College London. The model currently includes over 135 LMICs from Asia, Latin America, the Middle East, and Africa and has received positive feedback from WHO and DFID.
“All countries have been challenged by the COVID-19 pandemic, but it is those living in the lowest-income settings that are likely to be hardest hit. By making these tools available, countries with more limited resources can compare the outputs from epidemiological projections with the wider effects that interventions may have on their population to help to design locally appropriate responses that maximise health and well-being,” says Azra Ghani, Professor in Infectious Disease Epidemiology, MRC Centre for Global Infectious Disease Analysis, Imperial College London.
Features of the model include country-specific demography and contact patterns relevant to LMICs; age-specific outcomes based on current best estimates of fraction requiring hospitalisation and infection fatality ratio from China and UK analyses; explicit incorporation of healthcare needs and capacity including modifications to outcomes where healthcare is exceeded and for LMIC settings based on expert clinical opinion convened by the UK’s Scientific Advisory Group for Emergencies (SAGE) 4; and statistically robust model fitting to generate start date and the reproduction number (R0) for each country based on time series of reported deaths. In short, the current model provides simulation outputs, by country, that allow comparison of the epidemic trajectory, and healthcare demand over time.
The award will allow significantly quicker development of the software, which allows LMICs to access and use the model, and the development of a new “multiple intervention simulation tool” (including scenarios such as use of face masks and social distancing) to help governments and public health officials in LMICs to develop lockdown exit strategies and de-escalation plans. It is also hoped that the model will inform those developing drugs and vaccines, enabling them to explore longer-term scenarios.
“Bio Nano Consulting is honoured to have been selected as one of the grant recipients through the COVIDaction Data Challenge. The covidsim.org site has already proven itself to be an important online tool for LMICs, helping government, healthcare officials, strategic planners, and academic researchers to model, plan, prepare for, and manage the COVID-19 global pandemic more effectively,” noted David Sarphie, CEO of Bio Nano Consulting.
For updates visit covidsim.org and follow @bionanoc, @MRC_Outbreak and @imperialcollege on Twitter.
The Awardee: The United Nations Office for the Coordination of Humanitarian Affairs (UN OCHA) — Centre for Humanitarian Data, in partnership with Johns Hopkins University Applied Physics Laboratory
The UN OCHA Centre for Humanitarian Data has developed a subnational COVID -19 model for countries with ongoing humanitarian operations. The model includes estimates for cases, severe cases and deaths, and is based on a SEIR model of infectious disease dynamics.
The model parameters include the R0 rate, case fatality rate (CFR), and estimated probabilities that an individual person may contract Covid-19. It includes spatio-temporal elements of transmission, including transportation data and contact data between persons of different age groups. It also includes country-specific risk factors and comorbidities such as age, food insecurity, access to healthcare and other pre-existing conditions (such as malnutrition, diabetes, tuberculosis, cardiovascular diseases and chronic respiratory diseases).
The model provides COVID-19 case count estimates at the district level over the span of four weeks. Following testing and technical improvements, the model is now being used to inform decision-making in several countries, including South Sudan, Sudan, Afghanistan, the Democratic Republic of the Congo and Somalia.
The award will allow the team to improve the model by addressing data gaps and expanding it to additional countries such as Iraq and Ethiopia. Financial support will enable technical training and handover of the model from the Johns Hopkins University Applied Physics Laboratory, and ensure that the code and documentation is made available as a public good. Model adaptations could also be made to focus on specific populations in need, such as those experiencing food insecurity or displacement. Discussions are also underway to use the model to inform humanitarian (and development) needs in the coming months as secondary impacts of the crisis are felt.
“Our goal is for the model to be used by decision-makers to prepare for and adjust on-going humanitarian operations so that those most vulnerable can be protected and more lives can be saved. We are grateful for the support that this award offers and look forward to making all of our work accessible as a public good for others to use and learn from,” said Sarah Telford, Lead of the UN OCHA Centre for Humanitarian Data
For updates visit https://centre.humdata.org and follow @humdata on Twitter.
