Last week, at the Lake Nona Impact forum for Advancing Global Health, I discussed the possibility that AI could significantly improve healthcare and advance science. Our recent AI breakthroughs provide an unprecedented opportunity to make healthcare more accessible, personalized and effective, and significantly accelerate scientific discovery. Below is the latest information on our progress and how we are working with our partners to collaborate with our global healthcare environment and our recently announced AI co-scientists.
AI makes accurate health information more accessible
Google is often the first place people turn to when they are looking for answers to health-related questions. That's why we strive to ensure that everyone has access to relevant high quality health information at the moment they need it. Using lenses, people can take photos to search for skin conditions that are visually similar to what is seen on their skin. YouTube steers AI tools with health creators and organizations like the Cleveland Clinic to make it easier to publish prestigious, high-quality content.
We launched MEDLM for care providers and searched for healthcare that could provide answers to medical questions. They are available on the Google Cloud Vertex AI platform, helping clinicians make more informed decisions and helping patients receive the exact care they need. Our research on medical facts ensures that health-related content generated by language models is as reliable and indeed grounded as possible.
Generated AI paves the way for personalized healthcare
Advances in multimodality and conversational AI can help us rethink patient care and rethink how we can personalize for everyone, with an emphasis on preventive healthcare.
From X-rays to digital health records, medicine is multimodal. Based on MEDLM research, we have developed the next generation model of healthcare from Gemini, Gemini's excellent multimodal and inference capabilities, and fine-tuned it to identified medical data. In a published study, Med-Gemini achieved 91.1% accuracy for US health check-style questions, demonstrating how the model effectively interprets 3D scans and answers complex clinical questions.
We are using Articulate Medical Intelligence Explorer (AMIE), a research AI system optimized for diagnostic inference and conversation, to explore how AI systems can serve as conversational diagnostic partners in clinical settings. It is designed to employ a “clinical history,” derive differential diagnosis and ask intelligent questions that help you handle discussions with empathy, including the domain of your associate specialist.
Mobile and wearable devices are another promising area where generating AI models can provide personalized insights for both healthcare and wellness using data such as step counts and heart rate. We designed a leading language model for Personal Health, another fine-tuned version of Gemini, which can interpret sensor data and generate insights and recommendations about individual sleep and fitness patterns.
AI is improving health outcomes around the world
Early diagnosis of illness is important to improve health outcomes. Over the past decade, we have developed AI models that can be used to utilize the imaging and diagnostic capabilities of AI to help detect diseases such as breast cancer, lung cancer, and diabetic retinopathy. Through partnerships, we bring these solutions into a clinical setting at scale, ensuring that more patients benefit from timely and accurate screening. This impact is particularly profound in countries with low-resource healthcare settings and fewer per capita specialists. Over the next decade, Indian and Thai health technology partners aim to provide 6 million diabetic retinopathy screenings to patients for free, and Apollo Radiology International will be built on an AI model to provide 3 million free screenings across India for tuberculosis, lung and breast cancer. In addition to various initiatives to address maternal health in Africa, we also develop ML models of cardiac imaging, which are used to predict fetal well-being and to explore the utility in a limited resource's health environment.
It also lays the technical foundation for wider access to healthcare. Our Health AI Developers Foundation includes open weight models and resources to enable developers to build healthcare AI models more efficiently. The Open Health Stack (OHS)-equipped solution is a set of open source tools that make it easy for developers to create next-generation digital health solutions for healthcare professionals, already deployed across Africa, South Asia and Southeast Asia regions, serving millions of patients.
AI is accelerating scientific discovery
Medicine is rooted in science. Based on the AI's ability to integrate information and perform complex inference tasks, we explore ways to enhance scientific and biomedical discoveries through research on AI co-scientists, a Gemini 2.0-based multi-agent AI system. AI co-scientists are designed to function as collaborative tools for scientists. It aims to uncover new, original knowledge and to help scientists develop new research hypotheses and proposals, and is tailored to specific research goals based on previous evidence. We have already demonstrated potential in areas such as drug reuse in acute myeloid leukemia, proposing hypotheses of novel therapeutic targets for liver fibrosis, and explaining the mechanisms of horizontal gene transfer that underlie antibacterial resistance resistance.
We continue to recognize the potential for AI to advance science and improve, personalize and democratize access to healthcare. The “magic cycle” in which we achieve research breakthroughs and transform them into real-world influences is accelerating and expanding in scope. He took responsibility for this opportunity, continued his research in collaboration with global partners, published over 50 papers sharing cutting-edge health research in 2024, and recently shared the 2025 Health Impact Report. Ultimately, we believe that AI will continue to support healthcare and science advances for the benefit of billions of people.
