Skip links

On-Demand Webinar: To Learn How Payers Can Automate Prior Auth at Scale with AI. View Now.

Industry blog

The Data Analyst Will See You Now

NPR recently wrote an article about a project NYU medical school students are required to do entitled, “Health Care By the Numbers.” In the project, students are given access to a massive data set with more than 5 million patient records. They are asked to analyze the records to draw conclusions about care quality. The projects are not only interesting, they have produced some great insights: one student measured the rate at which prices for a hip replacement varied in different parts of the state compared to the rate at which prices for a Burger King hamburger varied; another group looked at the rate of C-sections in districts across the state – they are also useful. As one of the professors said, “With literally millions of records, these in-class student projects often involved more patients than the published literature.”

What happens when physicians gain access to powerful healthcare data sets?

This class is forward-looking because in the future, more and more physicians will gain access to powerful health care data sets, through companies like our own. In the best-case scenario, these data will enable real-time data-driven feedback on clinical decisions. When faced with a 40-year old patient with diabetes who has nightly fevers, a physician may be able to draw on the database to see that similar patients with these symptoms previously turned out to have one of three different diseases. It would be up to the physician to examine the data, ask tough questions of it, cross-reference its conclusions to first-hand observations of the patient, and decide on a course of treatment. Physicians could also use the data proactively, for predictive care that enables earlier interventions and better outcomes.

But there are also many ways that offering this data to physicians could lead to poor results. If physicians are unable to use the software system to acquire the correct data or unable to understand the data, it could actually lead to worse care. This isn’t a far-fetched fear; the rollout of electronic health records (EHRs) in many systems bears it out. In a recent article entitled Transitional Chaos or Enduring Harm, in the New England Journal of Medicine, Dr. Lisa Rosenbaum describes the extent of physician fear and confusion over EHRs, the dominant existing healthcare software systems. She writes, “There’s the critical care doctor who, unable to identify new information in daily notes, has begun printing them out and holding two superimposed pages up to the light to see what’s changed.” She also tells of an 18-year old who is given a near-fatal overdose of antibiotics after their doctor and pharmacist ignored several alerts in the EHR.

Where does the burden of using healthcare data lie, with software providers or physicians?

Certainly part of the burden to make the future look like our “best-case scenario” is on data software and data analytics providers to make a more user-friendly product. But part of the burden is also on medical schools to train a new generation of doctors to be able to use the massive data sets that will be available to them. I have a friend at the University of Pennsylvania’s medical school who told me about a recent talk he went to by Dr. Bob Wachter, a professor of medicine at the University of California, San Francisco, and the author of, The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine’s Computer Age. Dr. Wachter told the students, he said, that in the future every doctor will have to be a data scientist.

I don’t know if I’d go that far – we have professional data scientists at Apixio who would disagree that the typical qualification for their job, an advanced degree in computer science, is required for medicine. Moreover, strong healthcare data products have cognitive computing platforms that accomplish a lot of the real “data science” work (like data extraction and mining). But Wachter has an important point. Data analysis will be an essential tool for future physicians. Data isn’t just the domain of accountants and finance geeks any more – it’s for all scientific people, and that includes doctors.

🍪 This website uses cookies to improve your web experience.