Google Expands AI Tuberculosis Screening Across Africa and Indonesia

Google Research has detailed its expanding use of AI for tuberculosis screening in a company blog post. Its work includes clinical chest X ray software, open models for local health systems and cough analysis through smartphone microphones.

Nexus AI converted Google's chest X ray algorithms into a CE certified diagnostic tool that screens images in under 45 seconds. The tool runs in about 40 healthcare facilities in South Africa, Sierra Leone and Côte d'Ivoire, with expansions underway in the Democratic Republic of the Congo, Zambia and Cameroon. So far more than 90 clinicians have evaluated over 25,000 chest radiographs. One deployment uses portable X ray equipment and offline AI four kilometers underground at South Africa's Mponeng Gold Mine.

Google also released MedSigLIP for medical image classification and MedGemma for combining radiographs with clinical information. Both can run on local infrastructure. Indonesia's Ministry of Health is fine tuning the models on domestic chest radiographs as part of a program intended to screen 50 million people annually while keeping patient data within the country.

A separate model called Health Acoustic Representations analyzes cough sounds recorded with smartphone microphones. Research groups in Zambia, Ghana and other parts of sub Saharan Africa are studying the model for acoustic tuberculosis triage.