Data and analytics are further integrating with clinical care, as AI technology begins to read imaging studies and health systems launch genomic initiatives.
Earlier this year, Rochester, Minn.-based Mayo Clinic partnered with a California genomics company to develop a genomic database that would help physicians and researchers identify health conditions in patients, and Sioux Falls, S.D.-based Sanford Health partnered with another genomics startup to expand the health system's existing program that integrates genetic medicine into preventative care.
A study published in the January 2020 edition of Nature also included a study that showed Google Health and DeepMind's AI technology could identify breast cancer more accurately than six U.S.-based radiologists. Both topics are familiar to Oscar Marroquin, MD, chief clinical analytics officer for UPMC Health Services Division. He believes there is great opportunity in AI for diagnostic as well as genomic purposes, and the health system launched its UPMC Genome Center in June 2018. It also has an Institute for Precision Medicine dedicated to biomedical research and personalized care.
Here, Dr. Marroquin discusses how data is revolutionizing clinical care and the most exciting applications in the future.
Question: Where do you see the biggest opportunity for data analytics in healthcare?
Dr. Oscar Marroquin: The biggest things happening today in data analytics for healthcare are the use of machine learning and artificial intelligence. I break up the innovation into two buckets; the first bucket is the work going on around AI and machine learning in the world of diagnostics, specifically using AI techniques for image recognition within radiology and dermatology as well as cardiology and ophthalmology. The big data for diagnostic purposes is going to enable and enhance the work clinicians do so they can make better diagnoses with those images.
In the second bucket, we are spending a lot of time and efforts in using big data techniques, analytics and AI to help us becoming a learning organization. Specifically, we are looking at how to use all of the techniques and data points we have in our system to become more introspective and apply our data better, whether that is for risk stratification, population health or other initiatives. We are looking at the macro level as well to leverage the data and learning we can gather currently based on the large number of patients we serve.
The ability to use our data and apply all different analytical techniques is what allows us to use those insights to become a little better and provide care more efficiently. In the second bucket, you see a tremendous amount of activity in the healthcare space when we talk about data and analytics.
Q: How do you feel about Google and DeepMind's collaboration on AI technology that can detect breast cancer as well or better than radiologists, according to recent studies?
OM: The headlines are always a little bit more sensationalized, conjecturing whether AI will replace doctors, and that's not a reality right now. Most of us in the field agree that AI in diagnostics affords us the opportunity to be more efficient in how clinicians are doing their work. There are tasks that machines can do reliably well, and we are going to take advantage of that to increase our workflow. At the same time, we can also improve our accuracy and efficiency of how we deliver care. I see what AI will bring on the diagnostic side as an enabler to what clinicians are doing and not necessarily as a replacement.
Q: With your background as a clinician, what are the most exciting applications for genomic data and personalized medicine?
OM: There is an explosion of availability of genomic and other omics data that will allow us to inch closer to personalized medicine. In order to truly deliver personalized and precision medicine, besides being able to generate and consume omics data, one has to have robust capabilities to deal with a large amount of phenotypic data, which includes data that lives in the EMRs, imaging data and data from other modalities that are becoming more common and viable. In the future I see the integration of all of these data sources as being a big area of focus for data analytics.