AI & Data Exchange 2026: NIH’s Susan Gregurick on Overcoming Data Silos with AI Analytics

Why It Matters

By unlocking fragmented health data, NIH’s AI strategy promises faster public‑health responses, more precise research insights, and streamlined grant administration, positioning the agency as a leader in data‑driven biomedical innovation.

Key Takeaways

• NIH holds about 440 petabytes of data across three cloud providers.
• Bridge2AI builds AI‑ready datasets, beginning with Type 2 diabetes in AI/AN populations.
• AI clusters NIH’s 20,000 grant applications, improving reviewer assignment and conflict checks.
• Cloud partners AWS, Google, Azure provide sandbox environments for AI research.
• AI extracts pathology report data to link COVID infections with cancer progression.

Pulse Analysis

The National Institutes of Health is confronting a classic bottleneck in biomedical research: fragmented, siloed data. With roughly 440 petabytes spread across Amazon Web Services, Google Cloud, and Microsoft Azure, the sheer volume and variety of clinical records, wearables, and survey inputs have outpaced traditional analytics. By deploying large‑language models and specialized extraction tools, NIH can now pull actionable signals from pathology reports, electronic health records, and even real‑time public‑health feeds, dramatically shortening the time from data capture to insight.

A cornerstone of this transformation is the Bridge2AI initiative, which curates AI‑ready, gold‑standard datasets for the research community. Its inaugural focus on Type 2 diabetes prevalence among American Indian and Alaska Native populations not only fills a critical knowledge gap but also demonstrates how high‑quality, standardized data can accelerate studies on health disparities. These datasets are openly shared, enabling external scientists to train and validate models without the overhead of data cleaning, thereby fostering a collaborative ecosystem that leverages AI for equitable health outcomes.

Beyond research, AI is reshaping NIH’s internal operations. Automated clustering of the 20,000 grant applications received each year streamlines study‑section assignments and flags potential reviewer conflicts, freeing staff to focus on strategic evaluation. Partnerships with cloud providers supply sandbox environments where investigators can prototype algorithms at scale, while AI‑driven dashboards promise near‑real‑time surveillance of disease trends, such as lupus diagnoses across the network. Together, these advances position NIH to act swiftly on emerging health threats and to sustain a data‑driven learning health system for the nation.