Providence, Microsoft develop new AI pathology model

WASHINGTON, UNITED STATES — Health system Providence, tech giant Microsoft, and the University of Washington recently launched Prov-GigaPath, an advanced AI-powered pathology model designed to enhance patient care and revolutionize cancer diagnostics.
This groundbreaking model captures global patterns across whole pathology slides, offering unprecedented insights into cancer subtypes and mutations.
Largest pretraining effort to date
Prov-GigaPath stands out due to its extensive pretraining on real-world data. The model was trained on 1.3 billion pathology image tiles from 171,189 digital whole slides provided by Providence, making it one of the largest foundational models for digital pathology.
This dataset is five to ten times larger than other established pretraining datasets, such as The Cancer Genome Atlas.
Superior performance and comprehensive analysis
Unlike previous models that analyze small portions of slides, Prov-GigaPath captures both local and global patterns across entire slides using the GigaPath vision transformer architecture. This comprehensive approach has led to superior performance, as evidenced by a study published in Nature.
The study benchmarked Prov-GigaPath against data from Providence and the Cancer Genome Atlas, covering nine cancer subtyping tasks and 17 pathomics tasks. Prov-GigaPath achieved top performance on 25 of 26 tasks, significantly outperforming the second-best method on 18 tasks.
Transformative impact on cancer research
Dr. Carlo Bifulco, Chief Medical Officer of Providence Genomics, highlighted the model’s potential to unlock new approaches to studying the tumor microenvironment.
“The rich data in pathology slides can, through AI tools like Prov-GigaPath, uncover novel relationships and insights that go beyond what the human eye can discern,” he said.
This capability could have significant downstream applications in cancer diagnostics and prognostics, aiding clinicians in treatment selection and offering broader biomedical impacts.
Overcoming previous challenges
Ari Robicsek, Providence’s Chief Analytics and Research Officer, emphasized the transformative nature of this work.
“This transformative work is the result of focused efforts to overcome three major challenges that have stymied previous computational pathology models from widely being applied in the clinical setting: shortage of real-world data, inability to incorporate whole-slide modeling, and lack of accessibility,” he stated.
With Prov-GigaPath now globally available, its holistic whole-slide modeling promises to unlock new approaches to cancer diagnostics and prognostics.
The model’s ability to predict actionable cancer driver genomic mutations and characterize the tumor microenvironment could significantly advance precision medicine and improve patient outcomes.
Microsoft’s healthcare partnerships
Microsoft has been very active in its initiatives within the healthcare industry. Just last month, the firm partnered with revenue cycle management firm Ensemble Health Partners to accelerate the development of AI capabilities for healthcare providers through Ensemble’s proprietary platform, EIQ.
Last March, outsourcing firm Cognizant also joined forces with Microsoft to integrate AI technologies into healthcare administration systems.
At the same time, Nuance Communications, a Microsoft company, entered a partnership with healthcare company MEDITECH to integrate Expanse into the Nuance Dragon Ambient eXperience Copilot (DAX Copilot) to automate clinical processes.