During a May 9 webinar hosted by Executives for Health Innovation, James Buntrock, Mayo Clinic’s vice chair of information technology, described his organization’s cloud journey and relationship with Google Cloud.
Buntrock leads Mayo Clinic’s enterprise technology services, which covers enterprise platforms around data, API integration, identity management, as well as digital platforms and digital experience.
Through its websites and consumer applications, Mayo has been on a cloud journey for the last three years. But Mayo Clinic has actually invested in data warehousing since around 2005, he said, and has made incremental improvements to its data platform on premise. “At the peak of that, we were managing between two and three petabytes of data. With the cloud work that we've been doing with Google Cloud, we're up to 30 petabyte of data sets, including multimodal data, which is predominantly driven through imaging and digital pathology," he said, but also new types of data sets that historically they haven't been able to manage at scale, Buntrock said, “so it's giving us an opportunity that we just didn’t have before to drive some of these things like enhanced product innovation and improved diagnosis and treatment.”
With this foundation in place, Mayo Clinic is seeing more innovation as well as new types of digital tools, he said. “We recently refactored our patient applications, both for mobile and web portal. That's for the first time now running on Google Cloud. That's giving us a base to do more rapid innovation or changes,” Buntrock said. Another key area involves artificial intelligence. “Mayo is an academic medical center. We have a number of our “ologies” investing in data science to look for AI methods that could improve things around administrative burden to better prognostics or diagnostics,” he said.
As far as managing the cloud relationships, Buntrock said they have to keep a close eye on costs. There's a new term “cloud smart,” he said. It requires a different set of disciplines to manage costs. “There is a concept called FinOPs because the volume of the different charges that cloud collects is so high that you have to have new management techniques to manage costs and understand capacity optimization,” Buntrock said. “That is one that we're definitely paying attention to and I think that will continue to be a point that we'll be emphasizing in the next two years.”
Buntrock also touched on human capital requirements when moving to the cloud. “When we started our Google journey, we had like six people that had any kind of Google experience or cloud experience,” he said, “so that definitely takes time and investment in your workforce, a combination of upskilling and training but also recruiting, and I think that is not an overnight process. So that is a cloud technology constraint. You also are looking for people who not only know cloud technologies, but also know about clinical data and the content as well. As long as I have been doing enterprise data warehousing, understanding data to do ETL, to do mapping, to do semantic normalization is always a challenge. Getting people with the right skills to be able to analyze and make those changes is part of that human capital limitation.”
You have to recognize that this is a big investment, he said. “Don't underestimate the size of that foundational investment,” Buntrock said, but the return is not just in a year or two; the return is over a decade or longer. “We formalized our first larger data investments in the early 2000s. I mentioned that 2005 was the start of our enterprise data warehousing, but we've incrementally added in and it allowed us to jump into the cloud faster, because we had done a lot of data organization, a lot of data investment prior to that starting point.”
