Most likely, the last time you purchased something online you were offered “recommendations” or saw what “others also liked.” The technology behind these common prompts uses predictive analytics. Your online merchant is leveraging data about your shopping and online searching behavior to predict their desired outcome — that you might buy something else!
As we enter health care’s digital information age, health informaticists are applying large database statistical methods to create powerful predictive models. These models allow health care organizations to take proactive steps to prevent unfortunate or avoidable outcomes.
Health data is now more accessible than ever before because much of it is digitized. Just like our online purchasing habits are being digitally analyzed, so too are health care events in electronic health information systems. The ability to develop predictive models will become more efficient as health data types (claims, EMR, personal device, etc.) grow in number, improve in integrity and are pulled together to better understand health behavior.
The fact that some health care professionals are skeptical about predictive analytics is understandable. Clinicians typically take actions that are based on what has worked for them in the past. But all too often, when clinicians have evidence-based information about the effectiveness of a treatment, we still tend to prescribe what is familiar to us rather than use the interventions toward which the evidence points.
But, just as we are comfortable trusting other aspects of life that are based on data-backed prediction — e.g., the weather, investing in the stock market, the lottery — we will begin to trust predictive modeling in health care. Of course, just like weather reports, stock outlooks and lottery chances, the predictive power of health care data is limited, but the data and the analytics are better than ever before. And their predictive power continues to increase.
It is unlikely that we will ever be able to exactly predict human behavior, just as online merchants cannot be certain that you will buy what they recommend. But, as predictive analytics create models that help identify preventable outcomes, health care organizations can (and many are beginning to) take proactive steps that, in the end, “triple-aim-it” — toward improving the patient experience, lowering costs and improving the health of populations.
For a deeper discussion on the power of predictive analytics in health care, download the Optum white paper, Predictive analytics: Poised to drive population health.
–Carl Johnson, MD, EdM, MSc.
About the Author:
Carl Johnson, MD, EdM, MSc. is a pediatrician trained at Boston Children’s Hospital. He completed a Medical Education fellowship at Harvard Medical School and was a faculty health services researcher at The Mount Sinai School of Medicine.
Before joining Optum Analytics he worked as a physician executive at Cerner Corporation. He is a graduate of the Mount Sinai School of Medicine in New York City and has held faculty positions at Harvard Medical School, University of California at San Francisco, The Ohio State University, and The Mount Sinai School of Medicine.
Dr. Johnson believes that healthcare can be transformed with the help of the right data. When he is not helping to transform healthcare, he can be found playing tennis, cooking, perfecting his French, taking photographs, reading historical fiction, listening to music, and watching Ohio State Football.