Page 84 - SAMRC AnnualReport 2025-26

AI in Medical and Health Research
Imaging of the lung tissue obtained by surgery
from TB patients, indicating localization of
neutrophils with tissue destruction (far right).
Lung tissue from cancer patients (left) is used
as control, while non-necrotic lung tissue
(less damaged) shows less accumulation of
neutrophils.
Images showing AI as a utility to analyse
severity of lung pathology in the spectrum
of TB disease. Top image shows an AI model
utilized to grade lung damage. Second panel
depicts TB lung images generated through
AI, showing distinct pathology across lesions
of different disease severity. Bottom panel
shows the use of AI to grade lung pathological
changes over time post TB treatment.
82 SA M R C A N N U A L R EP O R T 2 0 2 5 / 2 0 2 6
CTR researchers are increasingly integrating
artificial intelligence (AI) and data-driven
technologies to strengthen TB research,
innovation, and evidence-based decision-making.
Advanced bioinformatics and computational
approaches underpin the Centre’s work in
pathogen biology, enabling the analysis of large-
scale omics and clinical datasets to identify genetic
determinants of drug resistance and tolerance,
understand pathogen evolution, and investigate
host-pathogen interactions. These efforts are
supported by whole genome sequencing, targeted
next-generation sequencing, digital PCR, and
robust data harmonisation pipelines that generate
AI-ready datasets tailored to African populations.
The Centre is also expanding its use of machine
learning to analyse genomic and pharmacogenomic
data, including the identification of TB susceptibility
markers, optimisation of polygenic risk scores, and
characterisation of pharmacogenetic variation
relevant to precision medicine. In parallel, CTR
applies AI in imaging and diagnostics through AI-
assisted PET-CT analysis for improved assessment
of pulmonary and subclinical TB, and through the
EDCTP-funded AddiCAD project, which combines
AI-based chest X-ray interpretation with rapid
biomarker testing to enhance TB detection in
decentralised and resource-limited settings.
AI further supports translational research and
innovation at CTR. Machine learning models are
being developed to predict antibiotic activity,
prioritise novel drug candidates, characterise
Mycobacterium tuberculosis phenotypes, quantify
bacterial growth in culture, and inform treatment
decisions for drug-resistant TB. In addition, AI
tools are increasingly embedded in bioinformatics
pipelines, laboratory automation, data analysis,
literature synthesis, and manuscript preparation,
aligned with institutional governance frameworks
for responsible and ethical AI use in health research.