MDCalc today launched CrossCover, an independent safety layer designed to sit between clinicians and the clinical AI tools they use. As generative AI moves rapidly into clinical practice, physicians increasingly need ways to understand not just what the AI recommends, but whether those statements align appropriately with clinical guidelines, how patient and physician privacy is protected, and where an answer deserves a closer look. CrossCover is designed to provide that additional layer of scrutiny while keeping physicians in control of the final clinical decision.
For more than 20 years, clinicians have used MDCalc to bring evidence-based clinical decision tools to the bedside. CrossCover extends that approach to generative AI. Just as new rules of the road and city planning developments provided the core infrastructure that allowed cars to explore the globe more safely, CrossCover enables medical professionals to navigate clinical AI with greater confidence. CrossCover adds an independent safety layer to participating clinical AI tools, with safeguards designed to protect sensitive information, check responses against trusted medical evidence, and surface potential issues for physicians to review.
“In medicine, ‘cross-cover’ means stepping in to help care for another clinician’s patients,” explains MDCalc Co-Founder and CEO Dr. Joseph Habboushe. “The role signals trust, shared responsibility, and continuity of care among physicians. That is exactly how we hope clinicians will view CrossCover, as a reliable partner that can extend their capabilities while supporting the provision of excellent, well-informed care.”
CrossCover is designed to work across the clinical AI ecosystem rather than replace the tools physicians already use. Participating AI partners generate the underlying response, while CrossCover independently applies its safeguards and checks around the interaction.
CrossCover is built on proprietary technology developed by MDCalc, including patent-pending features designed to better protect physicians and patients from inadvertent HIPAA violations, help avoid biased responses from LLMs by hiding user identity, and keep physician judgment at the forefront. Its CrossCheck feature, currently in beta, compares AI-generated responses with authoritative medical references and highlights potential discrepancies for physician review.
CrossCover launches in collaboration with its first two partners AMBOSS, the leading Medical Intelligence Platform, and Vera Health, the leading medical search engine that finds, grades and cites the best available evidence in seconds, used by more than 800,000 clinicians. Clinicians can access participating LLM partners directly through CrossCover, with additional partners also expected to join the platform. The American College of Physicians (ACP), a medical society and publisher of the Annals of Internal Medicine, and ECRI, which hosts the former national guidelines clearinghouse, also join CrossCover as content providers, delivering authoritative sources of medical data that will appear alongside AI-generated information using the platform’s CrossCheck (Beta) feature.
CrossCover launches with four primary features:
- PHI Scrubbing™: While best practice dictates never putting patient data into an LLM, human error can happen. PHI Scrubbing detects potential identifying patient information and recommends redaction before a query is sent to an AI model.
- CrossCheck™ (Beta): Independently compares AI-generated clinical information with authoritative published medical references – starting with ECRI and ACP – and highlights potential discrepancies and areas that warrant closer physician review.
- Tracker Blocking: Limits unnecessary information about the physician — such as identity, location and other contextual signals — from reaching partner AI models, thereby reducing the potential for those factors to influence a response.
- MDCalc Tools: When a clinical question calls for a validated clinical calculator, CrossCover surfaces the appropriate MDCalc tool directly in the workflow, bringing deterministic clinical calculations into the generative AI experience.
“In July, the independent Stanford-Harvard benchmark study called NOHARM ranked AMBOSS number one for safety and completeness for the second time. This really reflects over a decade of work by the 100+ physicians who built AMBOSS,“ said AMBOSS CEO and Co-Founder, Dr. Madjid Salimi. “Clinical excellence is at the core of everything we do and joining CrossCover as a founding partner extends that same culture of safety and rigor.”
“Every clinical AI says it’s accurate. Clinicians deserve one that can prove it,” said Vera Health co-founder Maxime Allouch. “Vera has been the world’s most downloaded app for clinicians every month since June, and that comes with a responsibility: showing the source and strength of the research behind every answer, even when that evidence is thin. We built Vera to be checked, which made joining CrossCover as a founding partner an easy decision. In clinical AI, trust is earned in the open.”
“We designed Vera to mirror how clinicians think,” said Vera Health co-founder Taieb Bennani. “Before it writes a word, it weighs every study the way a panel of expert clinicians would: by its design, level of evidence, size and recency. And it shows. In our latest evaluation, Vera was judged more accurate than every AI system it was compared with, including the latest frontier models.”
“We’re not building CrossCover because we want physicians using less AI. We think they’re going to use much more of it,” said Dr. Graham Walker, MDCalc’s co-founder, Head of AI, and author of The Physicians’ Charter for Responsible AI. “Medicine has always adopted powerful new technologies by building safeguards around them. Clinical AI should be no different. The goal is to help physicians move faster with AI while keeping evidence, transparency and clinical judgment at the center.”
Physicians
CrossCover is assembling its founding class of Charter Physicians — practicing clinicians across specialties who will receive early access and help shape how CrossCover works, what it flags, and how it fits into clinical practice.
To become a CrossCover Charter Physician, visit www.crosscover.ai/signup/
Partners
CrossCover is building an ecosystem of clinical AI companies committed to giving physicians greater transparency, safeguards, and control as they use AI in patient care. Partners can integrate with CrossCover to bring independent checks and safeguards around their clinical AI experience.
To explore becoming a CrossCover partner, contact feedback@crosscover.ai
View source version on businesswire.com: https://www.businesswire.com/news/home/20261111753002/en/
Media gallery
