Google Research and Google for Health are advancing AI-driven tuberculosis (TB) screening, addressing critical gaps in global healthcare access. TB remains a deadly disease, claiming over 1.23 million lives annually, with a major bottleneck being the scarcity of radiologists and imaging infrastructure in high-burden settings. The team has developed AI-based tools to interpret chest radiographs (CXRs) with accuracy matching expert radiologists, validated through clinical trials and regulatory clearances.
The collaboration with Nexus AI, a Pretoria-based medical technology company, has translated these algorithms into certified clinical diagnostic tools. Nexus AI achieved CE MDR class IIb certification and ISO 13485 compliance, ensuring the tool's safety and reliability. The technology is now deployed in six countries, including South Africa, Sierra Leone, and Côte d'Ivoire, with over 25,000 chest radiographs evaluated, reducing turnaround times from weeks to under 45 seconds.
One notable deployment is at the Mponeng Gold Mine in South Africa, where ultra-portable digital X-ray kits and offline AI are used to screen miners on-site. This initiative has streamlined the screening, triage, and confirmatory testing process into a single visit, significantly improving efficiency. The project has also delivered over 3,000 free screenings to under-resourced public clinics, aiming for 10,000 sponsored screenings annually.
To scale impact globally, Google has open-sourced its chest X-ray embeddings and MedSigLIP, a lightweight vision-language model. This allows developers and health ministries to create tailored solutions for local needs. The Health AI Developer Foundations (HAI-DEF) and MedGemma, a family of open-weight multimodal models, have seen over 10 million downloads worldwide, supporting public health systems in building sovereign solutions that protect patient privacy.