6 Health AI Updates Revealed at The Check Up

At today's The Check Up, Google's annual health event, we delved into how we're harnessing the power of AI to boost health outcomes worldwide. From enhancing search capabilities to pioneering new medical technologies, here's a closer look at our latest advancements.

1. Enhancing Health Search Results

When it comes to health, people turn to Google Search and features like AI Overviews to get trustworthy information on everything from everyday ailments to rare diseases. Since we rolled out AI Overviews last year, we've noticed users are not only more satisfied with their search results but are also posing more detailed and complex queries. Thanks to recent updates in our Gemini models, we've been able to make AI Overviews even more relevant and comprehensive, especially for health topics, while maintaining a high standard of clinical accuracy.

We've also kept up our efforts with knowledge panels on common health issues, such as the flu or the common cold, guiding users to credible online sources. Now, with the help of AI and our top-notch quality and ranking systems, we've expanded these overviews to thousands more health topics. We're also rolling this out to more countries and in additional languages like Spanish, Portuguese, and Japanese, starting with mobile users.

People don't just want expert advice; they're keen to hear from others facing similar health challenges. That's why we've introduced a new feature called "What People Suggest" on Search. It uses AI to organize insights from online discussions into easy-to-digest themes, allowing users to quickly see what others are saying. For instance, someone with arthritis might be curious about how others manage to stay active. With this feature, they can easily find real-life tips and experiences from fellow arthritis sufferers, with links to delve deeper. "What People Suggest" is currently available on mobile devices in the U.S.

[ttpp][yyxx]

2. Medical Records APIs in Health Connect

Juggling health data across multiple apps can be a real headache. To ease this, we've launched our Medical Records APIs globally within Health Connect. These APIs allow apps to access and share medical record information—like allergies, medications, immunizations, and lab results—in the standard FHIR format. With this addition, Health Connect now supports over 50 data types, covering everything from activity and sleep to nutrition, vitals, and now, medical records. This makes it simpler to link your everyday health data with information from your healthcare providers.

In Health Connect, your data stays on your device, and you're the boss, deciding which apps can access your data and what information they can use.

3. Loss of Pulse Detection

We're excited to announce that last month, the U.S. Food and Drug Administration (FDA) gave us the green light for our Loss of Pulse Detection feature on the Pixel Watch 3. This groundbreaking feature can detect if your heart stops beating due to events like primary cardiac arrest, respiratory or circulatory failure, overdose, or poisoning. If you're unresponsive, it will automatically initiate a call to emergency services, which could be life-saving.

Initially announced in 2024 with availability in the EU, this feature is now accessible in 14 countries, and we'll start its rollout in the U.S. by the end of March.

4. AI Co-Scientist

Research drives our work at Google, whether it's through incremental advancements or groundbreaking innovations that reshape the industry. To aid biomedical researchers in developing novel hypotheses and research plans, we've introduced an AI co-scientist built on Gemini 2.0. This tool helps researchers sift through vast amounts of scientific literature and come up with high-quality, original hypotheses. For example, if researchers are studying the spread of a disease-causing microbe, they can describe their research goal in plain language, and the AI co-scientist will suggest testable hypotheses, along with a summary of relevant literature and a proposed experimental approach.

While this tool doesn't automate the scientific process, it's designed to spark new ideas and speed up research. We're already collaborating with partners like Imperial College London, Houston Methodist, and Stanford University, and we're eager to see how researchers worldwide will utilize this tool. It's still in its early stages, but the excitement is palpable, and we're gearing up for our trusted tester program.

5. TxGemma

The journey from drug concept to approval is long and costly, which is why we're collaborating with the research community to streamline this process. Today, we unveiled TxGemma, a suite of Gemma-based open models aimed at enhancing the efficiency of AI-driven drug discovery. TxGemma can process regular text as well as the structures of various therapeutic entities, such as small molecules, chemicals, and proteins. Researchers can query TxGemma to predict key properties of potential new therapies, like their safety and effectiveness.

We'll be making TxGemma available to the community later this month through Health AI Developer Foundations, allowing others to build upon and refine it.

6. Treatment Options for Pediatric Oncology

In partnership with the Princess Máxima Center for pediatric oncology in the Netherlands, we're developing an AI tool called Capricorn. Utilizing Gemini models, Capricorn helps physicians quickly identify personalized cancer treatments by analyzing vast public medical data and de-identified patient information.

Capricorn generates summaries of treatment options and relevant medical literature, enabling physicians to engage in more thorough discussions about the best possible outcomes for their young patients. With AI taking care of data analysis, doctors can focus more on what truly matters: providing care to their patients.

Together, these updates illustrate the transformative potential of AI in improving health outcomes globally, and we're just getting started.