Mayo Clinic has spent the last 10 months focused on data and analytics initiatives with a long-term goal in mind: to make healthcare data more accessible and meaningful for clinicians.
The healthcare organization is turning to AI to make that happen. A significant part of the initiative involves a strategic partnership with Google; a year ago, Mayo Clinic announced that Google Cloud would be its primary technology platform. Since then, Mayo Clinic has been building its Mayo Clinic Cloud to store healthcare data securely; it also launched the Mayo Clinic Platform, a clinical data analytics platform where health IT developers can use de-identified data to build AI algorithms.
Mayo Clinic CIO Cris Ross said the Google Cloud partnership was critical to its AI success. The relationship enables the healthcare organization to bring healthcare data into a modern analytics infrastructure.
“A lot of people have said to me over the years … wouldn’t it be fantastic if you could use the power of the search and other tools of Google in order to get a deeper and more meaningful insight into patient care,” Ross said during a HIMSS webinar. “That was the root of what we were trying to do when we came together with Google.”
Ross said getting healthcare data to the point where it can be useful for analytics is key, which makes building the foundation to support AI a crucial element for healthcare organizations.
Laying the foundation
When priming a healthcare organization for AI and machine learning initiatives, there are three main stages, said Ilia Tulchinsky, engineering director at Google Cloud Healthcare & Life Sciences. They are data integration, data cleansing and data analysis.
- Integrating data
Tulchinsky said it’s important to connect healthcare data from different systems, such as medical imaging systems and EHRs, and be able to ingest that data into the cloud to make it accessible in one location. Getting data into the cloud is a necessary step for the success of an AI program, he said.
James Buntrock, vice chairman of IT at Mayo Clinic, said before this stage, Mayo Clinic focused on identity and security measures, as well as data privacy, monitoring and access, as a foundational element to moving healthcare data to the cloud. Then the healthcare organization turned to its Epic EHR first for data ingestion.
- Harmonizing data
“We sanitize it, ensure quality control and map it to common schemas that lend themselves well for the next step, which is analytics and AI,” Tulchinsky said.
If health data from different sources is going to be analyzed together, quality checks and common industry stands such as the Fast Healthcare Interoperability Resources (FHIR) data standard to create a common data language are critical, according to Tulchinsky.
Buntrock said Mayo Clinic’s data is being stored in a FHIR-store built by Google as the organization pursues the “longitudinal patient record,” an evolving, complete account of the patient that can continuously be added to.
- Analyzing data
Tulchinsky said best analytics practices are critical when building AI and ML systems. The first he recommended is making sure the problem is well-defined and that the metrics of success are well understood.
One of the most important steps with AI in healthcare is how it’s integrated into a clinical workflow. Providing the right data at the right time in the right way to a busy clinician is paramount to the overall success of the AI initiative, Tulchinsky said.