Mayo Clinic

The vast majority of pathology practices remain tethered to analog processes, hindering access to critical diagnostic data that could expand diagnostics, treatments, and speed development of new therapies. Mayo Clinic sought to modernize its pathology practice and unlock the value of its extensive slide archive for AI-driven diagnostics.

Mayo Clinic formed Mayo Clinic Digital Pathology, digitizing its pathology practice by scanning historical and prospective pathology slides. The platform leverages 20 million digital slide images linked to 10 million patient records (including treatments, medications, imaging, clinical notes, and genomic data), partnering with NVIDIA (using NVIDIA Clara infrastructure) and Aignostics to build and deploy AI foundation models for precision medicine.

In less than two months, Mayo Clinic and Aignostics developed a leading foundation model built on 1.2 million deidentified slides from Mayo Clinic and Charité – Universitätsmedizin Berlin. Current efforts are deploying new solutions enabled by this model, with future plans to train on 5 million slides to further accelerate disease detection and precision medicine.

• Digitizing a large, diverse archive of pathology slides and linking them to rich clinical records creates the data foundation needed for powerful diagnostic AI models.
• Strategic partnerships combining clinical data expertise (Mayo Clinic), domain-specific AI capabilities (Aignostics), and computing infrastructure (NVIDIA) can compress model development timelines dramatically.
• Foundation models trained on multi-institutional data can generalize better and accelerate downstream AI product development across pathology.