Every Sunday night John Halamka, MD, flies out of Boston to Rochester, Minn. He stays in a 600-square foot apartment that's about a five-minute walk from Mayo Clinic where he is working as the health system's inaugural president of Mayo Clinic Platform. On Thursday night, Dr. Halamka flies back to Boston where he spends Friday, Saturday and much of Sunday on his farm.
Mayo Clinic tapped Dr. Halamka to serve as president of Mayo Clinic Clinical Data Analytics Platform in December. He has been on the job since Jan. 1. And though he's only three weeks in, Dr. Halamka has already established the first partnership of the platform.
The Clinical Data Analytics Platform partnered with Nference, a startup developing AI-enabled biomedical software in which Mayo Clinic Ventures recently invested. Integrating Nference's existing AI technology into the Clinical Data Analytics Platform will enable researchers to better identify drug targets and biomarkers, thus advancing the development of drugs to prevent, treat and cure complex diseases.
Prior to joining Mayo Clinic Platform, Dr. Halamka served as the CIO at Boston-based Beth Israel Deaconess Medical Center. When asked why he decided to transition to the Midwest, Dr. Halamka said it was a once in a lifetime opportunity.
"If you look at the job description for this particular position at Mayo Clinic, it says candidates must be a doctor, must be an engineer, must have 20 years of digital transformation experience, must have international experience and extensive oral and communication experience," Dr. Halamka told Becker's Hospital Review. "Looking at that, it is the phenotype of about three people in the U.S. I wouldn't normally consider leaving, but when I looked at this opportunity, which was going out and changing the world of healthcare by scaling digital innovations with significant resources and a great team, it was something too good to turn down."
Below, Dr. Halamka describes his plans for Mayo Clinic Platform in 2020 as well as untapped areas for hospitals to leverage data.
Editor's note: Responses have been lightly edited for clarity and length.
Question: In your first year as chief, what goals do you have for the Clinical Data Analytics Platform?
Dr. John Halamka: As you might have read, there are three businesses that I have been asked to develop. One is how we share data ethically. Two is how we are going to build acute care in the home. Three is how we deal with data that might come from IoT and wearables.
On the Clinical Data and Analytics Platform, the idea is that we have a standard process for de-identifying the data and keeping it secure in a close container and inviting partners in to use their algorithms. Nference is the first partner. But you can imagine in this first year we will sign up some new partners. Some might be focused on payer analytics or others that might be focused on patient-facing tools. What I hope is to have five partnerships in clinical data and analytics for this year and do it in a way that everyone considers private and secure.
Q: You were quoted saying you wanted to create an "innovation factory." What does that mean to you?
JH: Here's a challenge: what happens often when you want to do a collaboration or a partnership it takes a year just to get through the legal issues or the IT issues. So, what if you could create a set of standard technologies, such as APIs that make selected deidentified data available' standard policies with templates and legal agreements; a team of people who are totally focused on this collaboration, could you do collaborations in weeks instead of months to years?
I have been on the job for three weeks, and we've already signed a major collaboration that brings Nference, which is doing machine learning, and Mayo and Google together to accomplish interesting pharma discovery work. In this effort, the data stays private, but shares the experience of 20 years of healthcare. I think if this example can be carried to others, we will be able to be much more agile in bringing on new collaborations and partners than in the past.
Q: How would you define a "platform"? And how is it being used in this case?
JH: Back in the day when you would sublet an apartment you would have to go down to your neighborhood newspaper and take out a "want" ad. It was very transactional, expensive and inefficient. Today, we have this thing called Airbnb, allowing millions of people to interact in a low-cost, frictionless environment. Think of a platform in healthcare as somewhat similar. That is if you can come up with the right technologies that de-identify the data and protect privacy. Then inviting innovators in to develop new insights from that data is a bit like Airbnb booking. It's a frictionless, low-cost transaction that is highly repeatable.
Q: Where are unseen opportunities for hospitals and health systems to use advanced data analytics?
JH: The challenge is that in many healthcare organizations there is decades of data, but it may be siloed inside an EHR and hard to use for insights. What Mayo recognized five years ago was that we should take data out of the EHR and provide a longitudinal healthcare data mart, call the Universal Data Platform, which has administrative data, patient reported outcomes, potentially genomics and other things not in the EHR.
What we are going to find is that we are in the post EHR era. We will have to recognize that the EHR is fine for what it does, but we need to surround it with other tools, especially in the analytics world.
Q: How are you prioritizing data privacy?
JH: Everything I am doing is on Google Cloud. But it's important to understand that Google has no access to that data. We might collaborate on certain projects under institutional review board-approved research or use their engineering talent, but that's all controlled by Mayo Clinic.
When you walk into Mayo Clinic the first thing you see is the fundamental, primary value: the patient always comes first. It's not a debate. In any discussion of any project, the patient always comes first. There are 11 layers of privacy protection in the architecture that we've created for this data analytics platform that ensures the data is not only de-identified but is certified as de-identified by external experts. It is never sent to anyone. Instead, qualified experts are brought in to run their tools on the de-identified data, but it's never outside of Mayo control.
Q: Any final thoughts?
JH: I look at 2020 and think why is 2020 a year for the perfect storm? As mentioned, we are now done with EHR implementations; the technologies that we have at our disposal, such as mobile and cloud and machine learning and APIs, are good enough. We now have an opportunity to take healthcare to the next level. We aren't just trying to deal with fulfilling our regulatory requirements. When you think of it, the themes of 2010s were EHR implementation, ICD-10, and the ACA. We are now at a point where if we have the institutional will and sense of urgency, we can do great things with the existing technology and the foundation we've laid.