We were excited to have had a sharing session on Friday 17th March with AQUITY Innovations, one of our key clients. This South African based nonprofit organization was established in 2010 to support government, health and social services institutions and communities as they strengthen their responses to the country’s health and social challenges, specifically TB.
During the sharing session, AQUITY’s Director for Health Programs Dr Sipho Nyathi explained that one of AQUITY’s key ambitions is to find innovative and scalable solutions for public health problems. Sadly, Sub-Saharan Africa has no shortage of public health problems: the region has the highest number of cases per capita of multidrug-resistant and extensively drug-resistant TB, according to an article published in The Lancet. In addition, if one compares the estimated burden of disease to diagnosed clients, there is a loss of approximately 100,000 people in South Africa - i.e. those with undiagnosed TB.
A two-pronged approach
SA subscribes to the World Health Organization’s goal of ending the TB pandemic by 2035. In order to reduce the prevalence of TB, more patients need to be identified. With this in mind, AQUITY reported on their two-pronged approach:
Public-private partnerships with private sector General Practitioners (GPs). AI can help predict TB burden at a subnational level by identifying hot spots, i.e. areas of relative disease. EPCON has used AI to increase active case finding in Pakistan (read about it here). The tracking and surveillance of TB cases is essential for disease control. AI-powered systems can identify high-risk populations, analyze TB transmission patterns and predict potential outbreaks, allowing health officials to take proactive measures to prevent the spread of disease.
Improving the quality of screening To find clusters of TB transmission , AQUITY made use of EPCON’s AI predictive model. With the help of EPCON’s dashboard, or geoportal, the all-important disease hotspots could be located. With the addition of more data to the system, specifically, TB positive patient identification, the system improves, and predictions become more accurate.
Contact management, where community workers go door to door, is very cumbersome. The EPCON model combines patient contextual data with program data to produce a digital twin model which could then map potential disease hotspots. Encouragingly, there was a fourfold increase in yield compared to community TB screening strategies, and 28% relative to a contact management strategy.
Lessons learnt
Primarily, digitization increases efficiency - more can be done with less. However, community workers and others involved with the project need to understand the process. Change management is therefore key.
Although AI applications are increasingly useful, their application in TB based public health interventions is still limited. More research is always needed, particularly to assess feasibility, model validation and cost effectiveness. Buy-in from policy makers is also essential.
Yet already, AI can be a game-changer for finding TB at the granular level in low and middle-income countries.