LONDON, July 8 — DeepMind, the artificial intelligence research lab acquired by Google in 2014, has become a central player in the accelerating field of AI-driven medicine, according to the public record of its research and regulatory filings. The timeline of DeepMind’s work in health science is anchored by its development of AlphaFold. In 2021, the system predicted protein structures for over 350,000 proteins from the human proteome.
That single achievement, documented in peer-reviewed literature, gave researchers a structural map for a significant portion of known human proteins at the time, a leap in molecular biology that pharmaceutical companies and academic labs are still integrating into drug discovery pipelines. Google Health, a separate division of Alphabet Inc., published a 2020 study in Nature on breast cancer detection using AI models trained on mammography images.
That paper, which drew on datasets from the UK and the US, contributed to a growing body of evidence that machine-learning systems could match or exceed human radiologists in specific screening tasks. The study did not lead to immediate clinical deployment, but it established a benchmark for subsequent work in the field. The regulatory environment for these technologies has been taking shape in parallel.
The U.S. Food and Drug Administration, operating under the 21st Century Cures Act passed in 2016, established a framework for approving AI-based medical devices. That law clarified that certain clinical decision-support software would be regulated as devices, setting the stage for a wave of submissions from technology firms.
In 2023, the FDA released draft guidance specifically addressing AI and machine-learning-based software as a medical device, or SaMD. That guidance, still in draft form as of this writing, signals the agency’s intent to handle the iterative learning nature of AI systems—software that changes its own parameters as it processes more data—differently from traditional fixed-code medical devices. The broader competitive landscape includes several well-funded efforts.
IBM Watson Health launched in 2015 with an ambitious plan to apply natural-language processing and machine learning to oncology and drug discovery. The initiative faced well-documented challenges, including difficulty integrating with hospital record systems and questions about the accuracy of its treatment recommendations. By the early 2020s, IBM had scaled back the unit and sold parts of its health data and analytics assets.
The record shows that the Watson Health experience has become a cautionary case study for the industry, underscoring the gap between AI’s laboratory performance and its real-world clinical utility. Startups have moved into that gap.
PathAI, based in Boston, focuses on AI-powered pathology tools for cancer diagnosis. Zebra Medical Vision, an Israeli company, has developed algorithms for automated analysis of medical imaging, including CT scans and X-rays. Both firms have secured regulatory clearances in the US and Europe for specific applications.
Neither has disclosed revenue figures that would indicate broad hospital adoption, but their regulatory wins suggest a path toward clinical use. The FDA’s evolving stance on AI devices is the key variable for all these players.
Under the 21st Century Cures Act framework, the agency has cleared hundreds of AI-based medical algorithms, most of them for radiology. The 2023 draft guidance on SaMD addressed how manufacturers can submit modifications to AI software that learns and changes over time—a process the FDA calls a “predetermined change control plan.” If finalized, that framework would allow companies like DeepMind and Google Health to update their models without filing a new premarket application for every iteration, a regulatory shift that could accelerate deployment timelines.
DeepMind’s corporate structure within Alphabet gives it access to vast computational resources and the Google Health research apparatus, but the lab has historically operated with a degree of independence on research direction. The AlphaFold protein database, made freely available to the scientific community, has been downloaded by researchers in numerous countries. That open-access model contrasts with the proprietary approach of many commercial AI health startups, and it has positioned DeepMind as a contributor to basic science rather than solely a developer of clinical products.
What to watch next is the translation of these research successes into approved medical devices. The FDA’s final guidance on AI/ML-based SaMD, expected later this year or in the near future, will set the rules of the road.
Companies that have already built relationships with the agency—through the pre-submission process and the SaMD pilot programs—will be best positioned to bring their algorithms to clinic floors. DeepMind’s parent company, Alphabet, has the regulatory experience of its Verily life sciences unit and the clinical partnerships of Google Health to draw on. Whether the lab can replicate its AlphaFold success in the more regulated, slower-moving world of patient care remains an open question, one that the FDA’s coming rulemaking will help answer.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Key Players in AI Medicine: Google Health, DeepMind, and FDA”, “description”: “LONDON, July 8 — DeepMind, the artificial intelligence research lab acquired by Google in 2014, has become a central player in the accelerating field of…”, “datePublished”: “2026-07-08T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Ariel Antonio”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
