EPCON was excited that his CTO Thys Potgieter could present at this year’s the McGill Summer Institutes in Global Health programme.
Dynamic TB surveillance platform
Thys Potgieter emphasises that the objective of its AI-based surveillance platform involves predicting the likelihood of finding people with TB in different population groups and geographical regions. When this occurs, health risks are quantified at population and individual level, insights are then communicated to health partners and policy makers, and evidence-based decisions for intervention are implemented.
AI Engine
The EPCON AI engine relies on contextual and real world data sources collected, in part, by real-time collectors. Whatsapp bots, SMS and other technology aid custom data collection. Contextual data includes population level data, land use, accessibility to healthcare, distance to road networks, etc.
A digital representation is then assembled, leading to accurate predictions and insights regarding TB cases, which are reported via easy-to-use digital dashboards.
What insights can the platform provide?
EPCON’s AI-based dynamic TB surveillance platform imparts a wide variety of very useful data, including:
Predicted TB positivity rate
Predicted absolute number of people with TB
Predicted number needed to screen
Predicted TB cases per 1,000 screened
Missing TB cases, i.e. undiagnosed cases
Recommended screening locations, with a priority ranking
Importantly, the model can predict more extensive hotspots than those found by notification data alone. Also, any additional clusters found from the model’s predictions can indicate areas of potential missing cases.
Further areas of research
Matthys Potgieter raised promising avenues for further research. For example, are areas with low case notifications actually low TB burden areas, or is it merely a lack of resources that leads to low notifications? Also, what is the effect of using predictive modelling recommendations on the yield of active case-finding interventions, as well as case notifications?
Beyond TB, the AI-based surveillance platform can be used for other diseases or conditions such as influenza, COVID-19, malnutrition and HIV. Using existing population, environmental and health context data, the model can easily expand to other areas to bring to light gaps in the healthcare system, and support network optimizations in function of risk, capacity and desired yield.
The Advanced TB Diagnostics course at McGill’s Global Health programme clarifies the desire for better TB diagnosis tools. EPCON’s AI-based dynamic TB surveillance platform is helping to meet that need.