RamaOnHealthcare November 4, 2022

Today, RamaOnHealthcare talks with Dr. John D. Halamka, President, Mayo Clinic Platform.

Dr. John D. Halamka, President, Mayo Clinic Platform

Dr. John D. Halamka, President, Mayo Clinic Platform

When Dr. John Halamka looks at the digital horizon, he sees a perfect storm for innovation. He says the forces of technology, policy, and cultural urgency are converging, creating a conducive environment to transform health care globally. Challenges remain, of course. Health inequity must be addressed by AI algorithms that are accurate, fair, and useful to all. Privacy and security of patient data must be safeguarded. By working with collaborators across industries and using data to enable wisdom, the Mayo Clinic Platform President envisions the ability to bring new cures and better outcomes to patients everywhere.

RamaOnHealthcare (ROH): Welcome, Dr. Halamka, to our thought leadership series. Can you tell us about Mayo Clinic Platform’s mission, game plan, and how you aim for human impact?

John Halamka (JH): The mission of Mayo Clinic Platform is to enable new knowledge, solutions, and technologies that improve patients’ lives. Our game plan is our roadmap — how we take steps to accomplish our mission. To achieve human impact, our roadmap leads us to other partners. We collaborate with other providers, pharmaceutical and medical device companies, health tech startups, and payers to drive innovation around diagnosis and treatment to transform health care and bring more cures to patients. 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.

ROH: There are ongoing media news and discussions, concerns on ethics and privacy. Please tell us your perspective that would educate our readers.

JH: Ethics: We know algorithms are only as good as their underlying data. So why don’t we publish statistics with each algorithm describing its fitness for purpose and evidence of how it was developed? We’re familiar with a nutrition or drug label on most U.S. foods, beverages, or drugs that tells you about the product’s ingredients. We envision a similar label for AI-enhanced algorithms that would list how the data set was derived and tested, what clinical studies were conducted to demonstrate it has value in routine patient care, and the methodology used to develop the model. Transparency is needed to reduce algorithmic bias and achieve equity in health care.

We are glad to be part of the Coalition for Health AI, working on an equitable AI framework with our colleagues in other academic medical centers, industry, and government. Together with colleagues at Berkeley, Duke, Johns Hopkins, Stanford, UCSF, Google, Microsoft, FDA, NIH, and others, we are working to create guidelines and ensure high-quality AI tools to address algorithmic bias.

We are glad to be part of the Coalition for Health AI, working on an equitable AI framework with our colleagues in other academic medical centers, industry, and government.

Privacy: We can’t overstate how foundational protecting patient privacy is. Healthcare providers have a legal and moral responsibility to keep patient data private. As we’ve shared previously, a new report from Cynerio and the Ponemon Institute says 56% of healthcare organizations have experienced one or more cyberattacks in the past two years that involved either Internet of Medical Things (IoMT) or Internet of Things (IoT) devices. More than half of the survey respondents say cyberattacks increased death rates. These devices include infusion pumps, glucose meters, pacemakers, and other vulnerable items. There’s a financial toll, too: 47% of organizations experiencing an attack were forced to pay ransom to retrieve stolen data. The report finds that 32% of the ransoms paid fall from $250,000 to $500,000. Only about 1 in five organizations say they had a mature system to address security issues proactively.

Mayo has a unique ‘data behind glass’ approach to algorithm development. We have a secure cloud computing environment that allows companies to build algorithm models they can use for innovation, but the data never leave the Mayo Clinic Platform-controlled environment.

ROH: Do you foresee a wide variety of healthcare professionals eventually having access to a platform/database to help determine diagnoses and treatments using AI/Machine Learning/Algorithms and proactively taking measures to maintain better and ensure individual health?

JH: The very purpose of a platform is to capture and facilitate interactions between users and producers. But innovative ideas are not limited to certain types of healthcare professionals. So yes, I see various people interacting with a platform database to advance diagnoses and treatments. It’s exciting to see that happening; clinicians, data scientists, biomedical researchers, and others are using platform models across many disciplines, such as cardiology, oncology, radiology, neuroscience, infectious diseases, and more.

The very purpose of a platform is to capture and facilitate interactions between users and producers. But innovative ideas are not limited to certain types of healthcare professionals. So yes, I see various people interacting with a platform database to advance diagnoses and treatments.

ROH: To what degree will technology (such as wearables and apps) help individuals impact their current and future health or improve compliance with treatment?

JH: Anecdotally, of course, we see this happening all around us already. Maybe you’re a runner, and you check your heart rate on your watch before, during, and after a run. Perhaps you are trying to add movement to your day, so you’re wearing a device to count your steps. You see the positive impact of your actions on your health.

We are seeing this individual interaction in clinical studies too. Mayo researchers recently reported on a study that showed electrocardiogram tracings done by an app downloaded on an Apple watch could be used to detect a weak heart pump. Left ventricular dysfunction afflicts 2% to 3% of people globally. It may have no symptoms or vague symptoms such as leg swelling. But in the study, an algorithm licensed to Anumana, a company we co-founded with our partner nference, used a consumer watch to detect what usually would require an expensive imaging test.

Participants securely transmitted 125,610 ECGs from 46 states and 11 countries over the six-month study period. The average app use was about two times a month. And overall participation was high – 92% of participants used the app more than once. Think about the ability to scale this to millions of people anywhere. When you know someone has a weak heart pump, lifesaving and symptom-preventing treatments are available.

ROH: The ability to predict disease may be very beneficial in preventing it. Do you foresee information such as DNA, Lifestyle, Social Determinants of Health, and more to be included for research purposes?

JH: I see an enormous potential for using data, AI/ML, in early disease detection, prediction, and prevention. Genomic data, data from wearables, and social determinants of health are already used to help health care be more personalized, but also for population health management. The key is linking those data, deriving actionable insights, and creating scalable solutions. Mayo Clinic Platform aims to solve the problem of deploying and adopting those predictive clinical AI solutions.

ROH: Mayo Clinic has an impressive amount of data! Do you foresee diverse providers and services throughout healthcare collaborating to integrate data into practice to transform health for all effectively?

JH: We already see diverse providers working together to use data to transform healthcare globally. In June, Mayo and Sheba Medical Center in Tel Aviv, Israel, signed an agreement to make it easier to share health technology and support early-stage startup companies. In July, Mayo and Mercy, based in St. Louis, signed a 10-year collaboration 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.

I’ll stress what was said earlier: Algorithms are only as good as the underlying data. AI developers must use diverse data sets across organizations and geographies or test their models on various data sets. The data needs to be unbiased, fit for purpose, and labeled appropriately. Our Mayo Clinic Platform_Validate product can now offer AI developers a report that evaluates their algorithms for accuracy, efficacy, and susceptibility to bias.

Our Mayo Clinic Platform_Validate product can now offer AI developers a report that evaluates their algorithms for accuracy, efficacy, and susceptibility to bias.

By collaborating with Mercy, for example, we can offer aggregated, deidentified clinical data that combine their history of care delivery in diverse communities with Mayo’s expertise in highly complex care. Our varying populations and geographies improve accuracy, reduce model bias in the data sets, and, importantly, create more robust treatment recommendations for patients.

ROH: You mentioned a goal in 2020 to turn healthcare data into wisdom, which will result in less invasive care, more timely care, less costly care, and ultimately better wellness. What outcome data can you share with us a year and a half later?

JH: One example is work being done to detect pulmonary hypertension, which has nonspecific symptoms such as fatigue and shortness of breath but can be deadly. With access to 8 million deidentified electrocardiogram readings among Mayo Clinic Platform data, artificial intelligence developers trained an algorithm to recognize pulmonary hypertension signs. We collaborated with Janssen and Anumana; an AI-driven tech company co-founded by us and our partner nference. This early detection algorithm recently earned breakthrough device designation from the U.S. Food & Drug Administration.

Another positive development is Platform’s work with Mayo’s Division of Pulmonology and Critical Care Medicine to improve the diagnosis and management of patients with pulmonary nodules by incorporating AI into current processes. AI has been integrated into two commonly used clinical tools. As a result, both tools have better precision and reach, allowing providers to identify treatable patients more quickly and develop personalized treatments for them.

ROH: You provide encouraging remarks regarding data and healthcare! Where do you see the further positive impact for patients and physicians ahead?

JH: Patients will have more personalized and convenient care, which will increasingly be tailored to their needs. The models of care delivery will increasingly be hybrid. In partnership with Medically Home, Mayo Clinic offers patients hospital-level care delivered in the comfort of their homes. We are working with partners to make that option available to more people across the country.

Physicians will practice better medicine with less administrative burden because their knowledge and skills will be augmented by digital tools embedded in the workflow at their fingertips.

Physicians will practice better medicine with less administrative burden because their knowledge and skills will be augmented by digital tools embedded in the workflow at their fingertips.

ROH: Thanks for chatting with us today, Dr. Halamka. Where do you see the “healthcare, digital, data-driven transformation” market headed in the coming two years?

JH: I appreciate this opportunity. We know economic forces may be a little rocky in the next few quarters, but there’s much momentum with AI developers, healthcare providers, industry, and government. Ultimately, we’re all trying to improve care and quality of life for patients. Using data, we can scale innovations to help people far beyond our walls.

Using data, we can scale innovations to help people far beyond our walls.

About Dr. John Halamka

John Halamka, M.D., is president of Mayo Clinic Platform, a strategic initiative to improve health care through insights and knowledge derived from data. He leads a portfolio of platform businesses focused on transforming healthcare by leveraging artificial intelligence, connected healthcare devices, and a network of trusted partners. Trained in emergency medicine and medical informatics, Dr. Halamka has been developing and implementing healthcare information strategies and policy for over 25 years.

Previously, he was chief information officer at Beth Israel Deaconess Medical Center, where he served governments, academia, and industry worldwide. As the International Healthcare Innovation Professor at Harvard Medical School, Dr. Halamka helped the George W. Bush and Obama administrations and governments worldwide plan their healthcare information strategies. He has written a dozen books about technology-related issues. He was elected to the National Academy of Medicine in 2020.

 
Topics: AI (Artificial Intelligence), Health System / Hospital, Interview / Q&A, Provider, Technology, Trends
AHA podcast channel hits 100,000 plays this year
Interview with Ben Albert, CEO & Cofounder of Upfront Healthcare
CIO Podcast – Episode 45: The Impact of IT in Healthcare with Skip Rollins
Q&A: How Our Aging Population Will Change Health Care
Podcast: Cancer Screenings, Cases and Mortality Rates 12/1/22