A Nigerian case study: more than 6000 additional TB cases found

To curb the spread of TB, it is vital to locate TB hotspots. EPCON assisted with this goal in southwest Nigeria, with the result that, to date, an additional 6000 TB cases were discovered within 8000 population clusters. This USAID-funded project is managed by the Institute of Human Virology Nigeria (IHVN) in collaboration with the Society for Family Health (SFH).

EPCON's AI model

EPCON plays a pivotal role by modelling infectious diseases with its predictive model, based on Bayesian Inference. The EPCON AI model can forecast tuberculosis burden over a 3-year window within communities by combining digital representations of the real world, or digital twins, with population data and health records.

The importance of digital tools

EPCON’s model relies on two linked digital tools to operate effectively. These are:

  • The geoportal, a geospatial map which displays specific information about identified hotspots and community settings. This helps determine most appropriate intervention..

  • The dashboard, which generates a priority screening list of all areas and a recommended movement plan to guide community interventions while providing a snapshot of program activities and achievements.

One challenge encountered during the project was health workers’ reliance on paper-based methods for data entry. To overcome this challenge, WhatsApp was introduced as a low entry technology ideal for resource-constrained settings. By incorporating Chatbot functionality into WhatsApp, the technology can capture community TB screening data. The coordinates of the screened locations are accurately georeferenced and captured on weekly data sheets. Benefits included:

  • Accurate location data

  • Planning screening locations in areas of interest

  • Reporting key indicators of healthcare workers’ activities

Osun state: a success story

Before 2020, Osun State, the ninth smallest state in area and nineteenth most populous, with an estimated population of approximately 4.7 million, was registering only approximately 2,000 cases per quarter. It was recognized, however, that there were significantly more cases in the community. EPCON, through its AI modelling, targeted a number of TB hotspots. Implementation was based on all steps of TB cascade of care - i.e. the succession of medical services that drives TB diagnosis to treatment.

The pleasing outcome involved Osun State’s steady rise to become one of the states with the highest TB notifications in Nigeria. Among the cases in this state, 40% are now being found via this new approach and following model recommendations.

From scepticism to positive results

Within Osun State, the small, hard-to-reach community of Ile Ogbo was targeted. This area has very poor access to health care facilities because of difficult terrain as well as poor roads, and is only accessible by motorbikes. The EPCON model predicted this area as a hotspot which was initially viewed with some degree of scepticism. Gratifyingly, and lending much credibility to EPCON’s model, 28 cases out of 200 screened were reported. Without EPCON’s model recommendation, this community would have been a focus of infection for nearby populations.

This success demonstrates the benefits of predictive and AI-focused intervention planning yields optimal results for infectious disease control.