Google says its open MedGemma models are screening patients in remote clinics and busy hospitals
Google says its open-weight MedGemma models, built to understand medical text and images, have been downloaded more than 10 million times and are being adapted for local health needs. Examples include an offline triage app in rural Uganda, a cervical-cancer screening app in Zambia used for more than 3,500 women, and an eye-screening system in India used for more than 50,000 patients. Google also says AIIMS Delhi is piloting dermatology and outpatient triage tools, and Indonesia's health ministry is building a tuberculosis detector aimed at screening 50 million people a year, with models run locally to keep patient data on site.