Feature Story | 10-Nov-2022

By eliminating bias in AI models and offering access to deidentified data, Mayo Clinic Platform aims to transform health care

Mayo Clinic

ROCHESTER, Minn. — With the recent launch of a product to evaluate artificial intelligence (AI) models for accuracy and susceptibility to bias, Mayo Clinic Platform is accelerating adoption of data-driven innovation in clinical practice.

Mayo Clinic Platform_Validate confirms the efficacy and credibility of newly developed algorithms and ensures they are fit for their intended purpose. Validate is one of the first products in the industry that provides a bias, specificity and sensitivity report for AI models. It joins Mayo Clinic Platform_Discover, which offers AI developers access to curated, deidentified electronic health data in a secure, privacy-protected environment, along with the tools for discovery, analysis and training, to help solve some of patients' most pressing needs.

In nearly three years since it began, Mayo Clinic Platform has built an ecosystem that orchestrates multiple collaborations with health technology innovators, enabling discoveries across the health care sector.

"Our platform aims to improve patient care by fundamentally changing the nature of health care delivery," says John Halamka, M.D., president of Mayo Clinic Platform. "By addressing the countless deficiencies and inequities built into the current system, we are ushering in a new era of democratization of health care, where accessible, compassionate and personalized care is available to everyone, everywhere."

Mayo Clinic Platform_Validate

Much of the skepticism around AI and machine learning is focused on the poor quality of evidence supporting some algorithms. Biased algorithms can lead to incorrect diagnoses and other serious risks for patients.

With the new Mayo Clinic Platform_Validate product, algorithms are measured for model bias observed in categories from age to ethnicity. Similar to the way a nutrition label on foods and beverages details the products’ ingredients, Validate reports how an AI algorithm performs under different constraints, including racial, gender and socioeconomic demographics. Clinicians can trust that AI models were evaluated by an independent third-party source. This transparency will help reduce bias and achieve equity in health care.

Mayo Clinic Platform_Discover

Mayo Clinic Platform_Discover helps health care innovators solve some of the most complex clinical challenges by offering clinical insights from Mayo Clinic Platform's extensive deidentified electronic health data across a range of disease and therapeutic areas.

The deidentified data are in a secure environment that enables companies to develop algorithm models, but these data never leave the Mayo Clinic cloud computing environment.

Looking at the numbers offers a sense of the depth of real-world data AI developers can access with Discover:

  • 10 million patients, spanning across geographies and therapeutic areas
  • 644 million clinical notes from 5.3 million patients, collected over 40 years
  • 3 million echocardiograms, showing the heart muscle in action
  • 111 million electrocardiograms, or ECGs, showing electrical activity of the heart
  • 1.2 billion lab test results
  • 9 billion pathology reports
  • 595 million diagnoses and 771 million procedures

Discover also offers augmented-curation technology to analyze real-world data from health records to accelerate clinical decision support.

Just as an art or library collection is curated so artists' or authors' works are assembled, organized, exhibited and stored to allow access and interpretation, so too does data curation enable scientists to access secure and searchable indexes that make clear where and how data are generated and stored. Data are from diverse sources and can be integrated in repositories, authenticated and archived.

Companies such as BD (Becton, Dickinson and Company) and K Health are using Discover's data to improve patient care. BD, one of the largest global medical technology companies, is doing predictive modeling. K Health, an AI-powered clinical health care company, is using Discover to reduce the time between the diagnosis and treatment of hypertension.

As Discover continues to grow and evolve, a new family of applications is being offered to help customers filter, explore and analyze the data from multiple vantage points.

Continuing to grow

Mayo Clinic Platform’s products like Validate and Discover are core building blocks of one of the first comprehensive multisided platforms in health care, facilitating interactions among multiple producers and users.

"Coordination of care across labs, pharmacies, clinics, hospitals and home care settings requires platform thinking and a new attitude toward privacy-protected information sharing," says Maneesh Goyal, chief operating officer, Mayo Clinic Platform. "Using data-derived insights and knowledge, the platform model gives the opportunity to improve patient care beyond our walls."

The ecosystem has grown to include new data network partners and innovators tackling a range of health care problems.

  • Mayo Clinic and Mercy, based in St. Louis, recently launched a 10-year collaboration agreement, the first of its kind between two large health care systems to use the most current data science and years of deidentified patient outcomes to find diseases earlier and start patients on paths to better health more quickly. Mercy's and Mayo's different populations and geographic locations will improve accuracy, reduce model bias and create more diverse, and therefore stronger, treatment recommendations for patients.
  • Through the Mayo Clinic Platform_Accelerate program, early-stage health tech startups are able to develop and validate their AI-driven health care products, advancing innovative ideas from the proof-of-concept stage to clinical readiness.

"Using data, early prediction and detection of diseases can become a reality that transforms health care for people everywhere," Dr. Halamka says. "We feel Mayo Clinic Platform is uniquely positioned to make this ambitious goal a reality because we are founded on Mayo Clinic's dedication to patient-centered care, treatment of rare and complex cases, exceptional outcomes, and world-class research."

###

About Mayo Clinic Platform
Founded on Mayo Clinic’s dedication to patient-centered care, Mayo Clinic Platform. enables new knowledge, new solutions and new technologies through collaborations with health technology innovators to create a healthier world. To learn more, visit Mayo Clinic Platform.

About Mayo Clinic
Mayo Clinic is a nonprofit organization committed to innovation in clinical practice, education and research, and providing compassion, expertise and answers to everyone who needs healing. Visit the Mayo Clinic News Network for additional Mayo Clinic news.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.