Google says Nexus AI’s TB screening reaches about 40 healthcare facilities

AI-assisted tuberculosis screening is operating in approximately 40 healthcare facilities, Google Research said in an October 23 deployment update about its partner Nexus AI. The system addresses a shortage of radiologists in high-burden settings. The AI helps identify people who need follow-up testing; healthcare professionals still make the final clinical decisions.

Google puts the deployment’s footprint at six countries, naming South Africa, Sierra Leone and Côte d’Ivoire, with expansion underway in the Democratic Republic of the Congo, Zambia and Cameroon. More than 90 clinicians have evaluated over 25,000 chest X-rays using the tool. Google reports turnaround times falling from weeks to under 45 seconds.

A flag for follow-up, not the final diagnosis

The software examines chest X-rays and highlights findings that differ from normal lung appearances, including signs suggestive of tuberculosis. Its interface marks areas where further investigation is required. That makes it a screening and triage tool: it helps clinicians identify patients who should move to the next stage of testing.

Reading those images requires recognizing subtle patterns, such as fluid near the top of a lung or areas of damaged tissue. Google identifies scarce specialist expertise and distant imaging infrastructure as barriers for rural communities and underground miners. Nexus AI, a Pretoria-based medical technology company, translated Google’s chest X-ray algorithms into a certified clinical tool.

The work predates this deployment update. Google cites a 2019 benchmark comparing its algorithms with experienced radiologists and a 2021 external evaluation by the Stop TB Partnership. It also points to a prospective, multi-site validation published in NEJM AI in September 2024. Google says that study demonstrated high sensitivity—minimizing missed cases—across diverse patient populations, while also detecting incidental lung and heart abnormalities.

The work between an algorithm and a clinic

Deployment required more than image-reading performance. Google describes the practical tasks as making software reliable across different X-ray machines and offline settings, fitting it into clinical workflows and meeting medical-device rules. These are the steps Nexus AI took to turn the research algorithms into software that healthcare workers could use in patient care.

According to Google, Nexus AI achieved Class IIb certification under the European Union’s Medical Device Regulation and ISO 13485 certification for medical-device quality management. Other regulatory steps remain in progress: Nexus AI signed an agreement with FIND to pursue World Health Organization prequalification and is also pursuing US Food and Drug Administration clearance. Neither pathway is described as completed.

Four kilometers underground

At South Africa’s Mponeng Gold Mine, the deployment takes screening directly to workers four kilometers below ground. Deep-shaft miners face chronic exposure to silica dust and an elevated risk of lung disease. Nexus AI carried out the project with the Stop TB Partnership, Harmony Gold Mine and the South African Department of Health.

Google says the arrangement allows screening, triage and confirmatory testing to happen in a single visit, compressing a process that often takes weeks. Beyond the mine, Nexus AI has delivered more than 3,000 free screenings to under-resourced public clinics. The collaboration’s target is 10,000 sponsored screenings annually—a goal, rather than its current reported total.