Editors: Raymond D. Aller, MD & Dennis Winsten
PathAI and Northwestern Medicine enter digital pathology collaborations
July 2025—PathAI and Chicago-based Northwestern Medicine have entered a multi-year venture in which Northwestern will implement PathAI’s AISight digital pathology image-management system. The two entities will also jointly undertake research initiatives and develop clinical innovation programs and artificial intelligence-powered diagnostic tools.
Northwestern Medicine announced that its 95 pathologists will use the AISight cloud-native platform to analyze pathology images, enabling remote case review across Northwestern Medicine’s 11 hospitals. “Integrating this technology into our workflow represents a significant first step forward toward implementing AI for clinical practice in our department,” Daniel J. Brat, MD, PhD, chair of pathology at Northwestern Medicine, told CAP TODAY magazine.
PathAI scientists and pathology researchers at Northwestern will also conduct studies to explore new applications for AI in digital pathology, such as for enhancing cancer detection, quantifying biomarkers, and discovering novel disease patterns. The collaborators too will develop algorithms, using real-world clinical data and feedback, and refine and pilot test AI tools in clinical settings at Northwestern Medicine.
AISight was available for research use only in the United States at CAP TODAY press time.
PathAI, 617-500-8457
Sanford microbiology lab first to use Roche automation solution
Sioux Falls, SD-based Sanford Health has announced that it is the first U.S. health system to use Roche Diagnostics’ new Molecular Process Optimization intelligent automation solution in a molecular microbiology reference laboratory.
Sanford is using the MPO solution in conjunction with Roche’s Cobas Prime preanalytical system and Cobas 6800 and 8800 analyzers for molecular testing. The MPO technology’s customizable algorithm is designed to improve turnaround time and reagent usage.
The technology will allow Sanford Laboratories to expand its range of test offerings while also improving testing capabilities for vulnerable populations, such as transplant patients, according to a press release from Sanford Health.
“The automation suite supports expanded hours of operation for molecular testing by reducing the manual steps involved,” Marijo Roiko, PhD, director of the Sanford infectious disease laboratory, told CAP TODAY magazine. “We can customize the MPO algorithm to perform testing after business hours, which is when many specimens are [delivered] by the courier. Also, by shifting some testing to after hours, we have the potential to free up some technologist time during the day shift to perform other molecular assays, such as those for transplant patients.”
Tribun Health secures FDA clearance for use of digital platform with two scanners
Paris-based Tribun Health has announced that its flagship Web-based image-management system, CaloPix, has received Food and Drug Administration 510(k) clearance for use with the Hamamatsu NanoZoomer S360MD and Leica Aperio GT450 DX whole slide imaging scanners.
“This is a major commercial and clinical milestone for us,” said Tribun Health CEO Jean-François Pomerol, in a company press announcement. “We’re entering the U.S. with a solution that has already proven its value in demanding, high-throughput clinical environments.”
OpenAI unveils open-source benchmark for evaluating large language models
OpenAI has introduced HealthBench, an open-source benchmark for measuring the performance and safety of artificial intelligence large language models in realistic health care-related scenarios.
“The 5,000 conversations in HealthBench simulate interactions between AI models and individual users or clinicians,” OpenAI posted on its website. “The task for a model is to provide the best possible response to the user’s last message. The conversations in HealthBench were produced via both synthetic generation and human adversarial testing. They were created to be realistic and similar to real-world use of large language models: They are multi-turn and multilingual, capture a range of layperson and health care provider personas, span a range of medical specialties and contexts, and were selected for difficulty.”
HealthBench is a rubric evaluation, in which model responses are graded against physician-generated rubric criteria specific to that conversation. Each criterion outlines what an ideal response should include, such as specific facts, or what it should avoid, such as unnecessarily technical jargon.
HealthBench was built in partnership with 262 physicians who have collectively practiced in 60 countries. These physicians collectively are proficient in 49 languages and have training in 26 medical specialties.
“Evaluations like HealthBench are part of our ongoing efforts to understand model behavior in high-impact settings and help ensure progress is directed toward real-world benefit,” OpenAI reported.
OpenAI is making the full evaluation suite and underlying data available in its GitHub repository (github.com/openai/simple-evals) in the hopes of generating community input and feedback.
Epic undertakes initiative to speed generative AI adoption
Epic has introduced a program, called Launchpad, to accelerate health care organizations’ implementation of the company’s Generative AI Suite starter kit workflows.
Epic clients can sign up to participate in a “Launchpad wave,” in which they collaborate with a group of stakeholders from other organizations and Epic AI experts to complete a four- to 10-week project, depending on the workflow selected, and go live with their workflow at scale. “To facilitate this rapid implementation, Epic experts lead a kickoff call and then host weekly group calls to review project plan progress, discuss feedback, and help resolve issues—if any arise—during your implementation,” according to product literature from the vendor. “Epic will continue to host calls to address feedback and optimize workflows for as long as the group continues to have meaningful discussions.”
Epic, 608-271-9000
Dr. Aller practices clinical informatics in Southern California. He can be reached at [email protected]. Dennis Winsten is founder of Dennis Winsten & Associates, Healthcare Systems Consultants. He can be reached at [email protected].