Access to timely healthcare often depends on geography, and nowhere is this gap more severe than in the fight against tuberculosis (TB). TB is preventable and curable, yet it claims more than 1.23 million lives each year. A key bottleneck is radiology: in remote rural areas and mining communities, radiologists are scarce and imaging infrastructure is often days away. As a result, millions of active cases go undetected, sustaining transmission.
Google Research and Google for Health have worked to change this, moving from an AI tool for identifying presumptive TB on chest X-rays to prospective clinical trials, regulatory clearance with local partners, and open foundation models such as Med Gemma. The team is presenting this work at IDWeek 2026 in Washington, D.C., showing how foundational models could democratize TB screening at global scale.
The research began with a central question: can AI interpret chest radiographs (CXRs) with the accuracy of an expert radiologist? In 2019, a benchmark published in Radiology showed deep learning algorithms detecting active pulmonary TB at levels matching experienced radiologists. In August 2021, the Stop TB Partnership ranked the model among the top-performing TB screening algorithms globally, and the World Health Organization that year recommended computer-aided detection for TB screening and triage. A September 2024 prospective, multi-site validation in NEJM AI found that AI-assisted screening kept high sensitivity across diverse populations.
Publishing a paper is only the start. The "translation gap" between an algorithm and a trusted bedside tool involves making software reliable across different X-ray machines and offline settings, integrating it into clinical workflows, and meeting medical device rules. Google partnered with Nexus AI, a Pretoria-based medical technology company, which turned the algorithms into a certified clinical diagnostic tool. Nexus AI achieved CE MDR Class IIb certification under the EU Medical Device Regulation, ISO 13485 compliance, and an agreement with FIND to pursue World Health Organization Pre-qualification, with a U.S. FDA pathway ongoing. Nexus Intelligence is now deploying the technology where diagnostic access is most urgently needed.
Source: Google AI Blog · Summarized by HeadlinesBriefing