By John Musenze
Researchers have unveiled a new suite of artificial intelligence (AI) technologies that could significantly reshape how tuberculosis is detected, monitored, and controlled across high-burden settings.
Among the most promising developments is CAD4TB+, an integrated AI platform designed to streamline screening, diagnosis, and surveillance in both urban and hard-to-reach communities.
The innovation, launched by Delft Imaging and EPCON at the Union World Conference on Lung Health in Copenhagen, Denmark (18–21 November), brings together multiple tools under a single dashboard, offering real-time analytics, hotspot prediction, and improved case-finding.
Experts say the system could strengthen national TB programs by closing long-standing gaps in monitoring, workforce shortages, and diagnostic delays, especially in places where access to X-ray machines and trained radiologists remains limited.
According to the latest World Health Organization (WHO) Global Tuberculosis Report, an estimated 10.7 million people fell ill with TB in 2024, with 1.25 million deaths recorded. Low- and middle-income countries (LMICs) carry the overwhelming burden, with eight out of every ten people who develop TB living in resource-limited settings where late diagnosis, weak health systems, and inadequate diagnostic tools continue to slow progress.
Imaging and EPCON presented the platform as an AI-powered system capable not only of detecting TB but also of monitoring, predicting, and mapping its spread in real time. While AI tools have been used in TB screening before, this is the first time screening, surveillance, hotspot prediction, and national-level data analysis have been consolidated into a single ecosystem.
For countries struggling with fragmented diagnostic pathways and delayed reporting, CAD4TB+ offers a potentially transformative approach.
For decades, early TB detection has been hindered by the limitations of conventional laboratory testing and shortages of trained personnel, particularly in rural and peri-urban areas.
In sub-Saharan Africa, WHO estimates show mortality rates remain six to eight times higher than in high-income countries. The 2024 WHO report further revealed that around 2.4 million people with TB went undiagnosed globally, most of them in LMICs. These missed cases continue to fuel transmission, particularly in crowded urban settlements, mining regions, refugee camps, and other high-risk environments where access to care is inconsistent.
According to scientists, CAD4TB+ aims to bridge these gaps by merging Delft Imaging’s well-established AI chest X-ray technology with EPCON’s surveillance and epidemiological modelling system. Each chest X-ray analysed by the AI becomes an immediate data point feeding into a national dashboard, offering health programs a clearer, faster picture of where cases are rising or where screening remains inadequate.
One of the platform’s most striking features is its predictive analytics module, which can forecast where TB hotspots are likely to emerge. EPCON’s modelling technology has already undergone successful testing in Nigeria under the USAID-supported TBLON-3 project, where hotspots identified through data modelling recorded TB positivity yields up to 103 per cent higher than areas selected through traditional methods.
Guido Geerts, President and CEO of Delft Imaging, said the ambition behind CAD4TB+ is to ensure early and efficient detection, especially in vulnerable and remote areas.
“With CAD4TB+, every screening becomes an insight that helps close the detection gap and accelerate the path towards TB elimination,” he said.
EPCON’s CEO, Caroline Van Cauwelaert, added that the platform was designed to bridge the longstanding disconnect between diagnosis and national-level decision-making. By transforming each X-ray into a meaningful public health signal, she said, countries can deploy resources more accurately and quickly.
Beyond the core technology, the conference also highlighted the need to expand screening to communities that have historically been left behind. MinXray, a leader in portable digital radiography, showcased ultra-portable X-ray systems that can be deployed in remote, nomadic, and conflict-affected regions where fixed radiology facilities do not exist.
In an effort to address the stigma surrounding tuberculosis, MinXray also collaborated with artist and TB survivor Paulina Siniatkina to create Breathe In, a digital artwork composed of 64 chest X-rays stitched together to resemble a breathing figure. Siniatkina said she hoped the artwork would encourage more open conversations about TB.
“The stigma surrounding TB often stops people from talking about it,” she said. “Through art, I want to keep the conversation going and shine a light on the urgent need for relief.”
The broader mood at the Union conference oscillated between urgency and cautious optimism. Dr Cassandra Kelly-Cirino, Executive Director of The Union, warned that global funding cuts and geopolitical conflicts threaten to reverse hard-won gains.
“We must squeeze every drop of potential from this conference to avert further derailment,” she said.
Meanwhile, Guy Marks, President of the International Union Against Tuberculosis and Lung Disease, noted that the innovations showcased reflect the extraordinary potential of AI to transform TB care if countries receive adequate support for adoption and scale-up.
For LMICs, where TB control efforts are constrained by late diagnoses, limited laboratory infrastructure, and fragmented reporting systems, CAD4TB+ offers a shift towards proactive disease management. It combines speed, intelligence, and predictive capability in a way that has not been available before. Instead of reacting to outbreaks, countries may soon have the tools to anticipate them.
TB continues to kill more people each year than any other infectious disease, especially among the world’s poorest communities. New tools are urgently needed. CAD4TB+ offers a glimpse into a future where detecting TB in remote villages, busy slums, or displaced communities can be done quickly, cheaply, and with far greater precision.
