MedGemma is helping global healthcare providers deliver better care

MedGemma, an open-weight artificial intelligence platform, is assisting healthcare professionals in delivering superior care to patients worldwide. The technology supports essential tasks such as triage, diagnostic screenings, and clinical decision-making, all while maintaining the privacy and security of sensitive patient data.

Accessible via the Health AI Developer Foundations site, MedGemma offers developers the opportunity to build customized health tools tailored to specific local languages and health requirements. Through its utilization of Gemma models, MedGemma equips researchers and public health organizations with an adaptable foundation, enabling them to create specialized tools for triage, diagnostic screenings, and more.

Since its inception, MedGemma has been downloaded over 10 million times, empowering thousands of research adaptations worldwide. This versatility enables MedGemma to be deployed in various settings, such as frontline care, high-volume hospitals, and nationwide public health programs.

In rural Uganda, for instance, Crane AI is leveraging MedGemma to develop EaseHealth, a clinical decision support application that enables community health workers to evaluate symptoms, review guidance, and make informed triage decisions without an internet connection. Similarly, Dawa Health, based in Zambia, has employed MedGemma alongside MedSigLIP 1, a lightweight encoder for medical text and images, to create the DawaMom app.

This offline-capable tool has been used to screen more than 3,500 women for cervical cancer, with plans to expand the initiative across the country and neighboring regions.

In India, Visilant has implemented a smartphone-based imaging system, which has screened over 50,000 patients for cataracts and other eye conditions. By integrating MedGemma into its screening workflows, Visilant aims to enhance the detection of treatable eye diseases, preventing permanent vision loss.

Furthermore, MedGemma is instrumental in high-volume medical centers like AIIMS Delhi, where clinicians are piloting IndusDerma, an AI-assisted dermatology screening tool designed to cater to the unique healthcare needs and skin tones of the Indian population. IndusDerma facilitates early triage, structured clinical summaries, and decision-making at the primary care level, empowering non-dermatologists and accelerating time-to-care across a high-volume public health system.

AIIMS Delhi's goal is to scale these applications across their network of hospitals, which sees up to 15,000 daily outpatient visits. MedGemma can be deployed on-site or on any cloud server, allowing organizations to maintain complete control and ownership of patient data—a critical aspect for public sector health initiatives.

Ministries of health from various countries, including Indonesia, are utilizing MedGemma and MedSigLIP to address their respective countries' specific needs. In Indonesia, a tuberculosis detection model is being developed using MedGemma and MedSigLIP, trained on local chest X-ray data, with the aim of screening 50 million citizens annually to eradicate the disease.

MedGemma and MedSigLIP serve as foundational tools for global developers, enabling the creation of safe, privacy-preserving, locally adapted healthcare applications that cater to real-world health needs. However, it is essential to note that these models are not intended for direct clinical diagnosis, patient management, or treatment recommendations.

Instead, they are meant to facilitate the development of downstream healthcare applications involving medical text and images. Performance benchmarks provide a baseline, but developers must validate, adapt, and make meaningful modifications based on their specific use cases before deploying these models for clinical practice.