South Africa
Enhancing TB Control through AI
and Genomic Innovation
Project Context
South Africa remains one of the seven highest TB-burden countries globally, with Port Elizabeth standing out as a major hotspot. A significant barrier to TB elimination is the large number of undiagnosed cases - many of which are asymptomatic. It is estimated that 57% of people with TB do not report symptoms yet continue to transmit the disease. To address this challenge, EPCON has led multiple complementary projects in South Africa, combining AI-driven hotspot prediction, mHealth interventions, and most recently, genomic surveillance to support targeted and effective TB control.
Project Objectives
-
TB Risk Prediction: Identify undiagnosed TB cases through AI-based mapping and guide targeted community screening
-
mHealth for DR-TB: Implement a mobile health solution to support treatment adherence among patients with drug-resistant TB
-
Genomic Innovation: Support the MAGMA-MICK project by visualizing the spread of resistant TB strains and linking genomic data with geospatial patterns
-
Connect diagnosed patients with care providers and support treatment follow-up
-
Map adherence trends and TB risk together to improve regional strategies
EPCON's Approach
EPCON developed high-resolution TB risk models based on contextual and epidemiological data to identify hotspots at sub-municipal level. In earlier phases, predictive models guided screening in Port Elizabeth, increasing diagnostic yield significantly.
Key Outcomes and Impact
-
AI-based screening led to a 3x increase in TB case detection compared to initial passive approaches:
-
Year 1 (pre-AI): 105 cases out of 3,808 presumptive TB patients (2.7%)
-
Year 2 (with AI): 1,640 TB cases out of 18,445 presumptive cases (8.9%)
-
This was 300% higher than targeted contact screening alone
-
-
mHealth adherence monitoring enabled tracking of DR-TB patients' adherence and well-being, with regional variation mapped alongside TB risk
-
In MAGMA-MICK, EPCON is visualizing genomic resistance data to understand and anticipate DR-TB risks better.
