The goal of the pilot projects is to broaden the research objectives of the hub by including projects that assess the effects of climate change on health, and projects that explore additional pathogens beyond SARS-CoV-2 and HIV. Additionally, the pilot projects seek to extend the geographical scope of INFORM Africa’s research focus to additional regions in Africa with the purpose of generating new data to bear on already existing data.
In December 2022, the INFORM Africa research hub published a call for letters of interest (LOIs), and subsequently conducted a series of reviews to select applicants eligible for a full proposal submission. The selection process was concluded in July, 2023, with the selection of five (5) projects from both internal and external applicants. The pilot projects are intended to run for a period of 12 months, with total funding of $30,000 for each project selected.
Title: Predicting COVID-19 variants in Africa through Active Learning and Hierarchical Classification
Principal Investigator: Joicymara Xavier
External collaborator: Dr. Pâmela M. Rezende (Data Science Manager, Nubank, Brazil)
An ongoing challenge in the COVID-19 pandemic is the timely and accurate classification of SARS-CoV-2 sequences with the plethora of available genomic data. Two popular tools for the dynamic classification of SARS-CoV-2 genetic lineages are Phylogenetic Assignment of Named Global Outbreak Lineages (PANGOLIN) and Nextclade (of Nextstrain). However, these tools are not designed to automatically identify new variants. Potential new variants need to be flagged by users and these alerts are then addressed by manual curation and assignment of new lineages. Fast identification of new variants is very important for public health response, particularly now that the sequencing rates are decreasing. Here, we propose to develop a new machine learning pipeline that classifies SARS-CoV-2 sequences and can detect new variants. We plan to use a hierarchical classification framework in tandem with an Active Learning (AL) technique to detect potential new SARS-CoV-2 variants in near real time. AL is a type of machine learning technique where, in addition to training on known truth data, the algorithm identifies the most relevant unknown data points and queries for human input. The output of this project will be directly usable by public health officials thereby enabling informed and timely public health responses. Furthermore, lessons learned in this classification task may be transferred to other pathogens.
Title: Vaccine coverage mapping: important tool for pandemic preparedness
Principal Investigators: Kenneth Enwerem and Victoria Etuk
The COVID-19 pandemic has revealed the need for a concerted and coordinated multi-sectoral approach to response initiatives, especially vaccination. While optimal vaccine coverage is critical for preventing or limiting transmission and spread of vaccine preventable diseases, experience with COVID-19 has demonstrated a poor uptake of free and effective vaccines (1,7). Despite ongoing sensitization programs, education, campaigns and incentives, Nigeria’s COVID vaccine uptake staggers at 57% of its eligible adult population of about 115 million people, with average vaccination rates in the South (40%) markedly lower than the north (71%). Nigeria has an ambitious target of fully vaccinating 70% of its eligible population, however, only 13 of 37 (35%) states have achieved at least 70% full vaccination of their eligible adult population, despite provision of free vaccines, mobile vaccination campaigns, and other vaccine uptake mobilization strategies (2,3,8,9). Mapping current COVID vaccination coverage with COVID incidence and spread will reveal inefficiencies in vaccine programs, aid detection of gaps in vaccine coverages across local government areas and sub-national regions to help the government to use its limited resources to plan better vaccination campaigns (4,5,10). This vaccine coverage mapping tool will be applied to assess the effectiveness of vaccines in preventing, limiting transmission and outbreaks or pandemics. It will be useful for monitoring vaccine effectiveness, improving immunisation programmes, ensuring equitable access and evaluating reporting efforts (6,7,11,12). By analyzing vaccination data and correlating it with COVID-19 incidence and spread, policymakers can develop more targeted interventions and better allocate resources to areas with lower vaccination rates. This approach will be useful in addressing the significant disparities in vaccine uptake across the country, particularly between the North and South regions. Ultimately, an effective multi-sectoral approach to vaccine distribution and uptake is critical to controlling the COVID-19 pandemic and preventing future outbreaks of vaccine-preventable diseases in Nigeria (13).
Title: Patterns and associations between Social Vulnerability Index and HIV and COVID-19 epidemics
Principal Investigator: Olanrewaju Lawal
The Social Vulnerability Index (SVI) has the potential to be a critical tool for Ministries of Health and Public Health organizations to prepare for and manage infectious disease outbreaks or natural disasters in Africa. However, SVI availability selected African countries at the 3rd-order administrative level through three specific aims. AIM 1 develops and shares SVIs for selected African countries using recent population-based HIV impact assessments (P for the African continent is limited and often, when available, is only at the national or regional level. The overall goal of this project is to leverage pre-existing African data sources to develop and use SVIs for HIAs) and other pre-existing data sources. AIM 2 identifies patterns and associations between SVI and HIV and COVID-19 epidemics. AIM 3 builds the capacity of researchers and public health officials to use their countries’ indexes.
Title: A Training Platform for Human Mobility and Health Data, Analytics, and Data-Driven Modeling- Preparing Future Data Scientists Crossing the Disciplines
Principal Investigator: Dr. Chenfeng Xiong
Enabled by the multidisciplinary collaboration of the INFORM-Africa (“Role of Data Streams in Informing Infection Dynamics in Africa”) team, the proposed research team is in the process of producing individual level trip data that reveals large-scale time-dependent travel patterns and human mobility. While this unique dataset has huge potential to enable innovation in pandemic tracking/prediction, public health, and climate change research and trigger new collaborations, the data remains as an engineering product that requires significant domain knowledge of transportation, data science, and statistical modeling to comprehend. To bridge this gap between engineering and health for INFORM-Africa team, the larger DSI consortium, and the even broader research community, we propose a pilot project focusing on technology and knowledge transfer of transportation data and related supplementary analytical tools.
The following two activities are proposed:
1) Develop an online training module for geospatial analysis and modeling using personal trips data. The module shall include a data user guide, a detailed tutorial on how to use the data, and a series of video lectures regarding analyses, visualization, modeling, and fusion using such data with other supplementary information such as GIS, health records, and climatic data.
2) Start seed collaboration that can bring interested researchers on INFORM Africa team on board to work closely with Villanova University research team on topics of mutual interests including retrospective analysis of five-waves of COVID-19 outbreak in South Africa, climate change and mobility, and 2022 Ebola outbreak, using transportation data products.
Title: Understanding drivers of COVID-19 resilience in Africa and plan for future pandemics: A data science approach using the CARTA Evidence data
Principal Investigator: Otwombe, Kennedy
The announcement that COVID-19 was a pandemic created major challenges for both resource rich and poor countries. In wealthy countries, infections and mortality increased fast creating fears about the impact of the pandemic in resource poor settings such as Africa. Thus, the leaders and scientists in Africa started strategizing on the best mitigation plans to alleviate the risks of COVID-19. However, the much-anticipated high rates of infection and deaths in Africa were not observed raising questions about resilience and pandemic preparedness. However, these were hard to understand or quantify without reliable COVID-19 data from Africa. Furthermore, a well-structured, reliable and open-access portal hosting COVID-19 related peer-reviewed publications and databases specific to Africa was unavailable. There is a paucity of data from Africa that addresses context specific drivers of resilience and are sensitive to cultural realities such as the influence of traditional healers and their remedies, traditional religious practices, myths and misinformation, rumours and ethnicity amongst others. Furthermore, there is an urgent need for the African continent to develop a general policy framework for future pandemic preparedness derived from reliable sources of data and analysed using the most recent technological advances like AI led by a dedicated team of stakeholders such as the Africa CDC, political leadership, religious leaders, scientists, healthcare workers and other interested groups. Therefore, this secondary data analysis project aims to use artificial intelligence (AI), specifically, machine learning (ML) approaches to collate and curate COVID-19 data from Africa in the CARTA Evidence (https://carta-evidence.org/) portal, a site containing peer-reviewed public health articles and instructions on how to access associated datasets; to evaluate the primary drivers of COVID-19 resilience in the African continent; and to evaluate the level of pandemic preparedness through lessons learnt from the management of COVID-19.