Even those outside the field of medicine recognize the complexity, breadth, and data-driven nature of its various disciplines.
Take pathology, for example—the science dedicated to diagnosing disease and injury through the analysis of body tissue. Pathologists face the significant challenge of analyzing human tissue to identify abnormalities while also reviewing patients' medical records. Understanding a patient's unique background and lifestyle is integral to making accurate diagnoses. These specialists must then convert their findings into understandable reports for clinicians who provide direct patient care. Additionally, they often need to follow up on results by ordering stains or preparing for tumor boards.
With advancements in modern technology, particularly generative AI, pathologists now have the chance to work more efficiently, spending less time on mechanical translations of information and more time thinking critically about their findings.
Enter Paige—a seven-year-old medical technology startup based in New York City—committed to revolutionizing cancer research, diagnosis, and treatment with its proprietary AI tools designed specifically for pathologists. Initially, Paige is rolling out a tool for research use within medical facilities rather than direct treatment.
This tool, named Alma, acts as an AI copilot, enabling pathologists to submit natural language questions directly through their work computers. Alma quickly retrieves relevant patient information from medical records, assists in report preparation, and facilitates follow-up actions.
As Dr. Juan Retamero, VP of Clinical Diagnostics at Paige, highlighted in a recent media call, “Most AI solutions focus on just one aspect—image analysis—but our models, including Alba, are designed to help with all three.”
What Paige’s new Alba AI copilot offers
Paige Alba integrates patient data from multiple sources, including Electronic Health Records (EHRs), Laboratory Information Systems (LIS), and Image Management Systems (IMS). By aggregating this crucial information—such as pathology reports, radiology findings, and historical medical data—Alba eliminates the need for pathologists to manually sift through disparate platforms, reducing administrative burdens and allowing them to concentrate on high-priority tasks.
The AI system summarizes patient history and vital data, delivering actionable insights almost instantly. It enhances decision-making through Paige’s clinical-grade AI tools, like Paige Omniscreen, which screens molecular biomarkers in tissue samples to identify potential cancerous areas. This system generates interim evaluations for expert review, streamlining diagnostic report creation. Physicians can review and approve these reports via voice command, improving workflow efficiency.
Retamero noted that “Alba combines visual analysis with natural language processing, helping pathologists by writing structured reports and pulling relevant clinical information from electronic health records or radiology systems.” This comprehensive approach not only aids in diagnosis but also reduces the time spent on repetitive administrative tasks.
Proprietary in-house AI models trained on millions of medical images
Alba builds on Paige’s extensive foundation in AI-powered cancer diagnostics. In August 2024, the company introduced its second-generation Virchow models—Virchow2 and Virchow2G—developed in collaboration with Microsoft and based on one of the largest clinical pathology datasets.
These models were trained on over three million pathology slides from more than 800 laboratories across 45 countries, encompassing a diverse range of demographics, including gender, race, ethnicity, and geography. This comprehensive dataset includes over 225,000 patients and 40 different tissue types, providing the AI with the depth needed to deliver insights into cancer across various pathology use cases, thereby enhancing diagnostic accuracy even in complex cases.
The scale of this data, combined with the models' 1.8 billion parameters, establishes Virchow2G as the largest AI model developed for pathology to date. Retamero emphasized that access to this extensive dataset, through collaboration with the Memorial Sloan Kettering Cancer Center in New York City, grants Paige a significant advantage in creating advanced AI tools capable of yielding meaningful clinical insights.
Alba's launch integrates these advanced model insights into real-time clinical practice, representing Paige's comprehensive approach to cancer care, ranging from diagnostics to research driven by AI.
Research-only for now
It is essential to note that while Alba has the potential to transform clinical workflows, it is currently designated for research use only (RUO) and is not yet approved for diagnostic procedures. However, its capacity to enhance diagnostic accuracy, especially in oncology, indicates a promising future for AI applications in clinical settings, paving the way for similar tools aimed at patient treatment. For now, it serves to assist pathologists in researching cancer features and enhancing their understanding of the disease.
Paige aims to push the boundaries of AI in healthcare. Its foundation models, such as Virchow2 and Virchow2G, have already illustrated the impact of large-scale AI on cancer diagnostics. Through continuous innovation, Paige strives towards a future where AI not only facilitates cancer detection but also improves personalized treatment plans.
According to Yousfi, the introduction of Alba is just the beginning. As AI technologies advance, Paige intends to further embed these capabilities into clinical practices, ensuring medical professionals have access to the most effective tools for diagnosing and treating cancer.