MistraL - Mitigation strategies for communities with COVID-19 transmission in Lesotho using artificial intelligence on chest x-rays and novel rapid diagnostic tests

The COVID-19 pandemic is straining health systems globally. Public health and primary care interventions implemented by high-income countries may be unsuccessful or infeasible in low- and middle-income countries (LMIC). Effective mitigation strategies need to be adapted to the specific situation faced by countries with limited resources and fragile healthcare systems. These strategies should be designed to protect healthcare workers, community members, essential health care services -e.g. TB and HIV programmes-, and should offer the best possible care to COVID-19 patients in settings with very limited intensive care infrastructure. Our consortium unites experts from diverse backgrounds to make research and translation an integral part of the COVID-19 outbreak response in the Kingdom of Lesotho, a LMIC with one of the highest HIV and TB prevalence in the world and a very limited capacity for PCR-based confirmatory SARS-CoV-2 testing. Swiss TPH will lead the multidisciplinary MistraL project on novel mitigation strategies for communities in resource-limited settings with local COVID-19 transmission (WP1). The overall aim of WP1 is to create and maintain efficient managerial structures, roles and workflows to achieve the intended project milestones and deliverables. In order to respond to emerging scientific evidence and to the dynamic of the COVID-19 epidemic in Lesotho, Swiss TPH will use an agile monitoring and evaluation approach by scrutinizing the selection of tools and possible alternatives. This will ensure that the best possible technologies are implemented to maximize the impact for patients and health system. In this pragmatic study on new COVID-19, TB and HIV screen and triage centres, we will describe processes and outcomes through individual and population-level data as well as data on operational characteristics. The first main objective is to validate and further develop the diagnostic performance of an artificial intelligence software tool, namely CAD4COVID, for automatic assessment of chest radiographs in COVID-19 suspected cases (WP2, Radboud University Medical Center). We will also validate CAD4COVID in conjunction with blood markers measurable at point-of-care, if currently emerging data proof their diagnostic value for the diagnosis of COVID-19. Additionally, a digital x-ray image repository will be established to allow developers improve and validate Computer Aided Diagnosis software solutions. The second main objective is to select the most promising antigen-based SARS-CoV-2 RDT and to evaluate the diagnostic performance and the operational characteristics of the novel RDTs under pragmatic conditions (WP3, FIND). The RDT will be chosen from FIND's R&D activities based on the following key criteria: cost, infrastructure requirements, ease of use, time to test result, timely availability on the market and expected test performance. During the project, we will revisit the selection of tools and consider alternatives if we have evidence of improved performance or operational characteristics. SolidarMed, a long-standing partner of the Ministry of Health in Lesotho, will be responsible for implementing and operating the COVID-19, TB and HIV screen and triage centres at two hospitals, one in Butha-Buthe and one Mokhotlong District (WP4). SolidarMed will be able to even build up a third centre, which will be used for RDT and CAD4COVID validation, through collaboration with the TB TRIAGE+ project. The innovative and comprehensive MistaL project will directly improve the health of the people in Lesotho during the COVID-19 pandemic and will potentially become a role model for other LMICs.