Artificial Intelligence at Mayo Clinic

Mayo Clinic has integrated over 200 artificial intelligence projects across its healthcare system, moving beyond experimental pilots to full clinical deployment. These initiatives leverage the organization's vast historical patient data to address both diagnostic challenges in cardiology and administrative bottlenecks in medical record management. By embedding AI into existing workflows, the institution aims to improve patient outcomes for conditions like heart failure while reducing the significant documentation burden on physicians.

Mayo Clinic is leveraging its massive scale—including nearly 85,000 employees and a $9 billion capital investment—to deploy more than 200 AI projects across its medical network. A primary focus is the early detection of asymptomatic left ventricular dysfunction, a precursor to heart failure that affects millions of Americans and costs the U.S. health system an estimated $32 billion annually. Mayo researchers trained a neural network on approximately 98,000 paired ECG and echocardiogram records to identify electrical patterns indicative of a weakened heart pump. This AI tool, which reads data from routine ECGs without requiring additional clinical steps, was validated in the EAGLE trial involving 22,641 patients across 45 medical centers and has since received regulatory clearance through licensing for the Anumana ECG-AI LEF model.

Beyond diagnostics, Mayo Clinic is addressing the administrative strain caused by the tens of millions of pages of medical records it receives annually, often from patients seeking third or fourth opinions. In collaboration with Scale AI, the institution developed "Record Time," a tool built on the Scale Generative AI Platform that ingests and organizes fragmented external patient records. The system generates chronological summaries and makes documents searchable, allowing physicians to quickly surface critical information buried in high-volume documentation. This application is currently one of roughly 150 AI models actively running within Mayo’s HIPAA-compliant environment, illustrating a strategic shift toward solving well-defined administrative bottlenecks rather than relying on broad diagnostic claims.

The broader implications of Mayo’s AI strategy highlight a preference for high-yield investments that utilize existing data streams rather than introducing new, complex procedures. By focusing on tools that integrate into established workflows—such as an AI-enabled digital stethoscope that flagged twice as many cases of peripartum cardiomyopathy as standard care in a Nigerian study—the organization is overcoming common barriers to AI adoption in healthcare. The collaboration with Scale AI also extends to safety event detection, such as identifying wrong-site surgeries or falls hidden within routine reporting noise. These efforts reflect a mature AI ecosystem where randomized trial evidence and operational integration take precedence over experimental novelty.