Google details AI health research achievements including 90 papers and $60M funding

Google teams have published 90 peer-reviewed research papers on understanding and addressing diseases over the last five years, accumulating more than 90,000 citations. These efforts span fundamental cellular discovery to laboratory analysis, aiming to make disease more detectable, treatable, and preventable.

Decoding Disease Mechanics

Google’s early-stage biology models serve as collaborative tools for scientists to reveal molecular and genetic mechanics.

• Drug Discovery: Researchers at the Institute for Global Health, Barcelona, used AlphaFold to screen 30,000 compounds for antiparasitic activity against Chagas disease, identifying two FDA-approved drugs for repurposing.
• Genetic Variants: The University of Exeter used AlphaGenome Atlas to analyze genomic data from over 54,000 individuals, pinpointing variants linked to aging (PLA2G7) and cellular oxygen sensing (EGLN1).
• Brain Mapping: In partnership with Harvard’s Lichtman Lab, Google reconstructed a human brain fragment at nanoscale resolution, making data available via Neuroglancer, alongside maps of fly, zebrafish, and songbird brains.

Accelerating Medical Solutions

New tools have replaced slow, iterative processes with systems capable of rapid, accurate data processing.

• Rapid Diagnosis: Collaborating with Stanford University School of Medicine, Google used DeepVariant and DeepConsensus to achieve genetic diagnosis via whole genome sequencing in less than eight hours, setting a Guinness World Record.
• Custom Enzymes: Google introduced AlphaProtein Novo, a machine learning pipeline that designs new-to-nature enzymes from first principles, successfully creating enzymes to synthesize drug motifs and degrade toxic plastics.
• Public Health Forecasting: Using Google Earth AI models, researchers from WHO AFRO and the DRC’s INRB identified Ebola exposure risks faster, replacing weeks of manual coding to help deploy mobile laboratories.

Supporting Medical Providers

Machine learning models are helping providers detect diseases more accurately and reduce administrative burdens.

• Cancer Detection: A mammography screening system evaluated across 125,000 women caught 25% of historically missed interval cancers and showed potential to reduce specialist reading workloads by up to 40%.
• Leukemia Mutations: DeepSomatic, applied by Children’s Mercy Hospital, confirmed known variants and detected 10 previously missed mutations in childhood leukemia cases.
• Flu Forecasting: A Google flu forecasting model was ranked #1 by the CDC for predicting flu-related hospital admissions during the 2025–2026 season, outperforming 38 other systems.

Funding and Accessibility

Google.org has provided over $60 million in 2026 to organizations applying advanced models to health challenges. This includes support for the Infectious Disease Institute at Makerere University to forecast malaria drug resistance and the University of Pennsylvania to design novel cellular therapies.