Page 73 - SAMRC AnnualReport 2025-26

P E R FOR M A NC E I N FOR M A T I ON

Training initiatives during the reporting period included: automated interpretation of TUS images to support
triage and detection of radiological signs of TB and
› Training healthcare providers in the Eastern community-acquired pneumonia in children.
Cape Province to strengthen the diagnosis and
management of childhood TB and pneumonia. In addition, the deployment of a digital stethoscope
› Training healthcare workers in Pakistan on TB to record lung sounds alongside clinical assessments
diagnosis. aims to support the classification of lung sounds
for the assisted diagnosis of TB and other
› Training programmes on infant and paediatric respiratory conditions. Feedback from site-based
lung function testing across Africa. implementation has informed refinements to the
› Training in childhood pneumonia management technology, particularly in relation to positioning and
and non-invasive ventilation across several African noise-cancellation features, to ensure the collection
countries. of high-quality digital recordings for analysis.

New research collaborations were established
during the 2025/2026 period at both national and
international levels. These included partnerships
with the National Health Laboratory Service (NHLS),
the SAMRC, the Gates Foundation, the UK Medical
Research Council (MRC), the London School of
Hygiene & Tropical Medicine (LSHTM), the Uganda
Virus Research Institute, the University of South
Florida, Harvard University, Boston University, McGill
University, the University of Southampton, and new
research partners at the University of Cape Town. Measuring Pulse Wave Velocity at the 13 year
clinical visit – Drakenstein Child Health Study.
AI in Medical and Health Research
Three new research projects were launched during
the 2025/2026 period to leverage AI technologies to
improve the consistency and accuracy of diagnostic
interpretation for TB and other LRTIs.

The Eval Paed TB AID project was established
in collaboration with clinical sites in Pakistan to
evaluate AI-assisted interpretation of chest X-rays
for pulmonary TB. The study continues to advance
through iterative feedback and expert radiologists' 1 year Neurocognitive Assessment –
refinement of the QStudy platform. This process Drakenstein Child Health Study.
aims to ensure clear, consistent diagnostic
definitions; alignment of classification criteria;
and consensus on image quality, readability, and
suitability for inclusion in analyses.

BREATHE (Building Respiratory Evidence for AI
Tools and Health Equity) aims to improve point-
of-care diagnosis of paediatric TB and pneumonia
by contributing data to a multi-country database.
Site staff will be trained in thoracic ultrasound (TUS)
using point-of-care ultrasound technology. Imaging
data will be combined with complementary
clinical information, including symptoms, physical
signs, digital auscultation findings, cough sound
recordings, laboratory results, and final clinical Measuring Hand Grip Strength at the 13 year
clinical visit – Drakenstein Child Health Study.
diagnoses. These data will be used to develop
and assess prototype AI-enabled models for the

SA M R C A N N U A L R EP O R T 2 0 2 5 / 2 0 2 6 71