Building Bridges Between Data Silos: Mayo Clinic Platform’s Ambitious Endeavor To Enable A Learning Healthcare System
Building Bridges Between Data Silos: Mayo Clinic Platform’s Ambitious Endeavor To Enable A Learning Healthcare System Seth Joseph, Contributor
Platforms have long been disrupting industries, including healthcare, and are in an unmatched position to solve complex problems endemic to healthcare. For Dr. John Halamka, president of the Mayo Clinic Platform, the drive for platform disruption is personal, and he’s betting big on Mayo Clinic to lead the charge.
Research, scientific discovery and medical innovation throughout history has been driven by the sheer will and curiosity of pioneering scientists, often in spite of the small, incomplete and often messy data. In the mid 1800s, for example, as cholera ran rampant in London, it was only through the steadfastness and ingenuity of research scientist John Snow - overlaying mortality data on maps of the city - that contaminated wastewater was identified as the key source of the public health epidemics in England. This was a monumental discovery, using only a small set of data, with limited resources and no technology assistance.
Mayo Clinic Platform is trying to bridge the gap between healthcare data silosgetty
Fast forward to today. At least a third of all data generated globally is healthcare related. With this amount of information available, the idea of aggregating data for research purposes, in a way that can dramatically bring down costs and improve someone’s health or lifespan, is very exciting. Platforms that bring together healthcare data are one of the best suited means to achieving that end.
John Halamka, M.D., President, Mayo Clinic Platform John Halamka
During a recent conversation about what’s been happening at the Mayo Clinic Platform since its launch in 2019, John Halamka, MD, Mayo Clinic Platform President and longtime health IT champion, hit on a point that really resonated with me. “I want to leave this planet in a place where my daughter has the benefits of all the patients who came before her every time she has a treatment,” said Halamka.
As a father and health IT professional myself, Dr. Halamka’s passion and commitment to health tech advancement is both commendable and infectious, as is his vision for what the Mayo Clinic Platform can accomplish.
What is the Mayo Clinic Platform?
On its website, the Mayo Clinic Platform is described as being a “pathway” for health care innovations that can change how health systems and physicians provide care. More specifically, the Mayo Clinic Platform is a marketplace connecting its participating members to a suite of Mayo Clinic Network products, services and solutions, with the goal of improving clinical care and operational workflows, in part through advanced applications of AI and machine learning.
Gianrico Farrugia, MD is CEO of Mayo ClinicGianrico Farrugia
Gianrico Farrugia, M.D., president and chief executive officer of the Mayo Clinic, believes what Mayo Clinic is building should be a blueprint for other organizations looking to transform healthcare in a similar way. “Everybody will do this a little bit differently. We are just the vanguard. We’re providing the playbook,” said Farrugia.
Farrugia, who’s been with Mayo Clinic for over 29 years and transitioned into the role president and CEO in 2019, notes that there needs to be a shift away from pipeline business thinking to a platform mindset. “At Mayo Clinic, and through the Mayo Clinic Platform, we are introducing platform thinking and kind of nudging people away from the pipeline model,” which Farrugia says is still the default. “We need a shift in thinking in healthcare, and we need to explain and show how this works so people can see how transformative it can be.”
Taking what Halamka calls a “data behind glass” approach, the Mayo Clinic Platform enables advanced data analytics to be run on (and artificial intelligence algorithms trained on) de-identified data. The data comes from Mayo Clinic’s network and other platform partner organizations, as well as from the vast amount of clinical research and scientific literature that exists.
Because of the platform’s federated architecture, Mayo Clinic Platform partners can safely leverage deidentified data from a single, centralized place – behind the proverbial glass – while accounting for data privacy, security and access rights.
Maneesh Goyal is Chief Operating Officer of Mayo Clinic PlatformManeesh Goyal
“At the end of the day, we're trying to take all of the components of the entire healthcare value chain, which is incredibly fragmented and messy for solutions developers, and compress them,” describes Maneesh Goyal, chief operating officer of Mayo Clinic Platform, “because that's how platforms operate and that's how innovation happens.”
When the Mayo Clinic Platform first launched in 2019, Halamka was bullish on the promise and potential of platform disruption in healthcare. “Platform business models have been a force of disruption in many sectors, and the rapid digitalization of health care is affording us an unprecedented opportunity to solve complex medical problems and improve lives of people on a global scale,” he said. More than three years later, he’s just as enthusiastic. “I think Mayo Clinic has the best chance, of any organization I’ve ever worked with, of actually catalyzing this fundamental change in the way that healthcare is delivered globally.”
A Three-Sided, Federated Platform
Digging deeper into its model, the Mayo Clinic Platform is a three-sided platform marketplace, featuring: data network partners (hospitals and health systems that make their data discoverable on the platform); solutions developers (companies leveraging Mayo Clinic Platform products and data to train and validate AI models); and healthcare service providers (organizations who want to procure solutions that have been vetted by Mayo Clinic Platform’s data). An additional set of partners (known internally at Mayo Clinic as “construction partners”) help to provide the underlying technology infrastructure to operate the Platform.
In July 2022, Mayo Clinic signed its first data network health system partner, Mercy, in what Halamka described as a “unique collaboration will eliminate the barriers to innovation in health care by bringing together data and human expertise through a new way of working together.”
Mercy has physician practices, outpatient facilities, urgent and acute care locations across Arkansas, Kansas, Missouri and Oklahoma. The organization touts more than 2,100 Mercy primary and specialty care physicians and 600 advanced practitioners in 300 locations.
Because both Mayo and Mercy were early adopters of integrated EHRs, the organizations have a vast amount of outcomes and clinical data for analysis and the AI treatment. The combination of Mayo Clinic Platform’s privacy-protected, cloud-based technology architecture, paired with AI and machine learning advancement, makes it easier to pinpoint disease earlier and identify best treatment options, reiterated the companies in the 2022 announcement.
While Mercy is Mayo Clinic Platform’s only data network partner to date, COO Maneesh Goyal says the company expects to have half a dozen or so data network partners in the mix by next year.
Joe Kelly is EVP, Transformation & Business Development Officer at MercyJoe Kelly
From Mercy's point of view, its work with Mayo Clinic Platform will enhance its ability to deliver personalized, predictive, and proactive care in near real time through the appropriate channel, says Joe Kelly, EVP of transformation and business development officer at Mercy. “Our intent is to create an environment where patient information remains secure, confidential, and completely deidentified, yet enables us to apply artificial intelligence and machine learning against one of the largest longitudinal clinical data sets we collectively believe exists in the world,” says Kelly.
In terms of network the ‘demand’ side of the Platform, Goyal notes there are 35 small- to midsize health systems that are currently part of the Mayo Clinic Care Network. Through both branded and unbranded affiliations, health systems pay a subscription fee to leverage Mayo Clinic Network. This fee gives subscribing provider organizations access to Mayo Clinic care protocols, Mayo Clinic clinicians for second opinions and consults, as well as access to tools developed at Mayo, including Mayo Clinic Platform.
As for why a data network partner would want to join the Mayo Clinic Platform, the idea is rooted in reciprocity, which is a critical part of a platform network strategy. As Goyal described, “The reason that any data network partner would join is because when they join, they get access to the entire data network.” Once the data partners are integrated, the organizations then have access to Mayo data, and vice versa, where Mayo will also have access to Mercy data.
Healthcare Platforms Aplenty
Healthcare platforms have been catching fire for years now and funding has exploded in kind, with digital health platform investments hitting nearly $12B in 2021, outpacing all digital health investments (platforms and non-platforms) just two years prior in 2019 ($8B).
What’s driven this incredible uptick and market interest is the fact that platforms are uniquely well-suited to fix healthcare’s biggest problems — from more effectively matching supply and demand for healthcare solutions and services, to providing different constituents a common technology infrastructure in order to improve care coordination. In short, platforms can lower significant transaction costs due to high fragmentation, and lessen information asymmetry by improving information flows among healthcare’s diverse stakeholders.
But platforms aren’t easy to build or scale. It takes time, experience, investment, and in-depth understanding of platform business models and network effects. It also requires trust.
Because Mayo Clinic has the brand recognition and trusted reputation working in its favor, the health system is well positioned to step in to play a trusted operator role. This doesn’t mean there aren’t others out there trying to tackle similar problems, however.
Launched in 2021, Truveta is a large-scale data platform created from a consortium of 14 health systems — a number which has since grown to 25. Truveta uses machine learning to structure clinical data into an FHIR model for faster insights, and the platform updates daily with new data. All member organizations have an equity stake in the company, which has been
As described by Terry Myerson, Truveta CEO and former Microsoft executive, Truveta was developed in response to the Covid-19 pandemic and the need for a comprehensive dataset to help healthcare researchers, scientists, academia and others do their jobs.
Terry Myerson is CEO of TruvetaTerry Myerson
“Truveta was sparked by Covid and the experience of 2020; you had a pandemic, no one knew how to treat it, and no one knew where to ask questions or where to look for answers,” says Myerson. “If you have a dataset that is representative of the whole country, now you can go to one place for answers.”
Truveta also regularly publishes research on the impact of Covid-19 across populations, leveraging data on its network to uncover new findings. A more recent study from Truveta Research found racial and ethnic disparities in Covid hospitalizations at varying times during the pandemic, including an increased length of stay for Black patients compared to white patients. Results of the latest study were published in Frontiers in Public Health in November 2022.
Mayo Clinic Platform’s Startup Acceleration Program
Mayo Clinic Platform_Accelerate is a newer part of the business that launched in 2022. The program aims to attract solution developers to the platform, in part by helping early-stage health tech AI startups become market-ready in a number of months. Accepted startups get to work with Mayo data leaders on evaluating AI model requirements, checking for AI fairness and bias in models, and educating leaders on the federal compliance process, among other benefits. Program participants also gain access to Mayo’s platform data for model validation and study development.
“These companies were able to take already exceptional products and services and accelerate them with access to rich deidentified data sets and expert advice,” said Jamie Sundsbak, partner relationship manager of the program, of its first four members.
Impressively, one member of the Accelerate program’s founding cohort, Seer Medical, credits its FDA clearance and resulting U.S. launch of its epilepsy diagnostic and monitoring technology to its collaboration with the Mayo Clinic.
As an Accelerate program participant, the Melbourne, Australia-based Seer Medical could leverage Mayo’s expertise and data platform to refine its AI models with an eye towards compliance and improving the status quo. Currently, it takes on average about five years for a patient to get to a good treatment plan for epileptic patients, explained Seer CEO, Dean Freestone. The company wants to reduce that down to a few weeks.
The FDA’s decision to approve Seer Medical’s technology was based on the successful results of a stateside clinical trial that was completed at the Mayo Clinic, explained the company.
Challenges And Opportunities of the Platform Model
Mayo Clinic Platform and Truveta are both multisided platforms trying to create value in healthcare by providing aggregated access to healthcare data, and both are rel by iant on virtuous network effects to succeed. And while the approaches may be different (e.g., ownership models, monetization strategies, participants, etc.), each company is ultimately looking to develop scalable, more efficient ways to facilitate safe and secure access to data in order to enable rapid healthcare innovation, research and discovery, and each is using platform business concepts to try and accomplish this goal.
But achieving virtuous network effects is tough, especially in a highly-regulated and complex industry like healthcare. One of the only examples of platform business success in healthcare, Surescripts may now be known as the nation’s e-prescribing network, but it didn’t become the dominant network by accident. It made significant upfront investment in infrastructure, participant recruitment, building trust, and proving out the maturity of an end-to-end ecosystem of partners.
Truveta is seeking to overcome the cold start problem in part by opening up equity ownership to health systems in order to attract them as participants in the platform. And, unlike Mayo, Truveta has been upfront in communicating that the company sells and monetizes the patient data on its platform, with Myerson explaining, “Health systems want to diversify their revenue streams. Data could be a part of the way health systems run their operations in the future.”
Given Truveta’s 25 health system partners, its approach seems to be resonating in the market.
On the other hand, Mayo Clinic Platform may simply be designing its platform for the long game. The organization’s partnership with nference (a ‘construction partner) is one way Mayo Clinic Platform differentiates itself, as nference provides de-identification technology that is key to the “data behind glass” approach.
“We believe very strongly that everyone should keep and hold on to their own data,” says Halamka, where the blinded data on Mayo’s network stays within the federated network vs. being sent out to and owned by others. “While aggregating data is generally a good idea, organizations have to be really careful to respect patient preferences and do it in a way that protects the data and privacy preferences.”
Halamka also points to the breadth and depth of data that Mayo Clinic Platform is gathering as a differentiator, making the data contribution process easier for data network partners and in turn attracting solution developers.This differentiation is especially important in the early days of a platform, before “the flywheel effect” is reached. Mayo is also letting participants white label various platform solutions to help drive interest, participation, and generate network effects.
Different approaches to creating data networks need not inhibit collaboration, however. There is room for more than one platform solution to these problems, and partnerships can be a very powerful way to amplify the collective work happening in different networks.
At Mayo Clinic Platform, “People can partner with us any way they want,” invites Halamka. “If they just want our experience or playbook to run it themselves, we’re open to that. If they want to contribute data, that’s fine. We believe this is an important mission to fulfill.”
A Few Big Questions For Mayo Clinic (And Truveta) To Consider
While Truveta is upfront about its efforts to build a data platform that can eventually generate enough business to deliver a meaningful revenue (and perhaps dividend) share back to its health system customers and investors, Mayo Clinic seems to be more focused on achieving the vision that Farrugia and Halamka have laid out than on achieving any financial return. At some point, however, even a not-for-profit organization must consider how a major, CEO-led initiative will pay for itself.
It’s clear that, while both organizations place an emphasis on establishing and maintaining trust in how they operate and share data, other challenges that platform operators commonly face are likely to impact them. Here are a few critical questions that Mayo Clinic Platform’s and Truveta’s leadership should be asking themselves as they attempt to build enough momentum to ensure their platforms can become self-sustaining.
- Who are the most likely long term sources of revenue for a data platform? Clearly, both Mayo Clinic Platform and Truveta expect that there will be demand for responsible aggregators of healthcare data. Truveta’s early named clients (Pfizer and Boston Scientific) suggest it views biopharma and medical devices as likely paying customers, while Mayo Clinic has focused more on bringing artificial intelligence companies onto its platform. Figuring out where demand is most likely to come from, in order to start generating meaningful revenue, will be important to building a path to sustainability.
- Have we set the right expectations about it takes to reach critical mass? Many commonly think of platforms and marketplaces as ‘overnight success stories’ due to the network effect. The reality is that most network effect-driven platforms take five to seven years to reach a critical mass at which point network effects become self-sustaining. Managing investor and board expectations, and showing progress in the interim, can be the difference between ultimate success and failure.
- What services do we need to offer in order to ‘unlock’ demand? Mayo Clinic Platform products are largely geared toward data discovery and de-identification. It also acts as a convener, but is currently not involved in the actual commercialization of solution developers’ products, nor their subsequent relationships with customers. Determining the most compelling and valuable parts of the value chain, and creating solutions to help reduce friction for both solution developers, health systems, and other sources of demand, will be important for both Mayo Clinic Platform and Truveta.
Finally, it’s important to note that, in operating the platform under its own brand, Mayo Clinic Platform may face challenges in recruiting other health system partners that Truveta (as an independent entity whose shareholders are its customers) may not. Given that the two entities will ‘compete’ in part on the breadth and depth of their data, Truveta’s early momentum may be difficult to overcome.
Healthcare Platforms And Leaving A Lasting Impact
Over 150 years ago, English physician John Snow utilized limited, messy, incomplete data to derive insight that identified the source of a cholera outbreak.
Today, we have largely digitized health records for every American. Yet that data is not yet widely shared or accessible in a trusted, privacy-preserving manner. Fixing this will not an easy mission, but it’s imperative in today’s fragmented healthcare system. Absent trusted platforms like that of Mayo Clinic and Truveta, science and medicine will continue operating in silos, which is not only inefficient, but leaves far too much opportunity for discovery and innovation to slip through medicine’s fingers.
“I’ve been working for 40 years to get to this point,” says Halamka, who is excited and optimistic about seeing the technology he always envisioned become real. “I don’t want to leave this to the next generation to solve.”
Coming out of a pandemic that reinforced how fragmented the healthcare system is and how that adversely impacts care today and research for tomorrow, health data platforms have the power and the potential to transform healthcare worldwide. There will continue to be a vast number of organizations coming together to build capabilities that never existed before. As for who the leader of the pack will be, that remains to be seen.