Thought Leadership > Rearchitecting Healthcare Data for More Cures

Rearchitecting Healthcare Data for More Cures

By Gianrico Farrugia, M.D. Photography by Matt Meyer

Artificial intelligence (AI) and healthcare AI have dominated headlines and boardroom discussions. I have been both a strong proponent for the use of safe, validated AI in healthcare and for the need of governance and continuous monitoring to ensure the result is better healthcare and more cures. It is increasingly apparent that we will not achieve this without significant changes globally to how we make healthcare data accessible in real time or near real time for both humans and AI agents. Every data point in healthcare should serve one purpose: improving outcomes and developing more cures for patients. As healthcare generates more data than at any point in our history and AI becomes more capable of reasoning across it, we must structure data so it is usable for discovery at scale, something not currently possible in most public or private healthcare systems.

I have said that “data + people = new knowledge,” and “new knowledge = more cures.” AI is now the exponent in these equations. It compounds the impact of data and people by helping us see signals earlier, validate them faster and move discoveries into practice at greater speed. But this can only happen when the foundation is the right one. Healthcare data must be ingested, enriched, vectorized, governed and connected on a platform. Only then can we use AI at scale to move from observation to insight, from insight to trial, and from trial to more cures for patients.

At Mayo Clinic, this is the purpose of Mayo Clinic Platform, which now includes 54 million de-identified patient lives across four continents. The goal is to help researchers and solution developers find signals across real-world clinical experience and test those insights quickly. Having the right data architecture on the right platform safely available to humans and agents is showing results.

One of the clearest examples is clinical trials, essential to developing more cures. A modern data architecture can make clinical trials faster, smarter and more targeted. For example, using Platform data, Joseph C. Ahn, M.D., and his team have simulated disease progression in alcohol-associated hepatitis against 10,000 synthetic patients and 1,400 actual patients, with nearly identical survival patterns. In another example, Cui Tao, Ph.D., and Nansu Zong, Ph.D., and their team fully emulated a randomized trial comparing warfarin and aspirin using Platform observational data. Preparing and validating the dataset took months, compared with 12 years for the original trial. Trial design, end points and statistical power should be highly informed by AI run on a modern platform data architecture before most, if not all, large clinical trials are initiated.

Every data point in healthcare should serve one purpose: improving outcomes and developing more cures for patients.
— Gianrico Farrugia, M.D.

Within orthopedic surgery, Matthew P. Abdel, M.D., and his team leveraged Platform clinical data to help predict surgical site infections before total knee replacement surgery. They built a virtual patient cohort and generated a synthetic control arm to develop an AI-driven predictive model that could identify high-risk patients before surgery and enable the care team to intervene early.

A similar approach is changing research for brain tumors. Our chair of Neurosurgery in Rochester and chief medical officer for Mayo Clinic Platform, Gelareh Zadeh, M.D., Ph.D., had previously spent five years building a cohort of a few thousand patients with brain tumors. Using Platform, the team could immediately begin working with a population approaching 4,000 patients with glioblastomas and 5,000 patients with meningiomas, with an average of 16 years of longitudinal data per patient, as well as 5 million clinical notes, 15,000 MRIs and thousands of structured data points.

This scale and depth of data create opportunities to uncover patterns that would otherwise remain hidden. For example, Dr. Zadeh’s lab is studying whether blood sugar levels and seizure medications are linked to better outcomes for brain tumor patients. Early analyses suggest that controlling glucose levels may help patients with brain cancer and possibly other cancers, and they also suggest that certain seizure medications — such as lamotrigine and levetiracetam — may be associated with longer survival in patients with malignant brain tumors. These findings are strong enough to influence patient care decisions and have led to two clinical trials.

Related Content

Persistence Pays Off

Neurosurgery leader's journey of discovery leads to Mayo Clinic.

We also need to link molecular biology to patient outcomes, and that is the purpose of our Bioresource and Research Data Atlas initiatives. Bioresource collects samples like blood, plasma and tissue and connects them to detailed patient information. Currently, more than 6.6 million samples from more than 219,000 patients have been brought together into one centralized digital system. Our Research Data Atlas then brings all this information together — biospecimens, patient data, lab results, imaging and more — into one place so researchers can work across genomics, spatial biology and cellular imaging in one environment.

These are just a few examples that demonstrate why rearchitecting healthcare data should be a global imperative requiring investments at a national level. It is one of the most important steps we can take to accelerate cures. AI is the exponent, but our data architecture is the base — and now we must remake it for a new era of inquiry.

This article was originally published on LinkedIn on June 30, 2026.