But we still have a long way to go for all healthcare providers to have access to data that reveals patient correlations that aren’t recommended or obvious. Mining, aggregating and applying big data will play a dominant role in making that goal a reality.
Let’s Talk
The effort to use personalized medicine to identify missed opportunities in patient care began primarily as a discussion between CMIOs in large hospital systems and a few physician champions in certain specialties, such as cardiology or orthopedics. This “manual” process identified omissions in patient care as well as high-risk patients, and then turned that information into workflow protocols that would involve calling patients or checking appointment books to look for trends. Even though highly skilled clinicians were involved, it was still a very subjective way to find and close gaps.
Data Extraction from EHRs Brought Objectivity to the Process
Since electronic health records became the norm around 2009, physicians were expected to put information into specific fields in the EHR. In some cases, a physician’s compensation was tied to how well they adhered to this system. This dramatically changed the way hospital leadership views personalized medicine. Instead of identifying gaps in care in a conversation within a practice, a dialogue with doctors or a committee, hospitals and medical practices began employing a much more scientific approach. Using technology tools to extract data out of the EHR is not only much simpler but also more objective.
In instances where only one hospital served a region, and everyone used the same system, data was easily extracted, and reports were built to close the gaps. For example, when a report showed that patients diagnosed with multiple ailments hadn’t had an A1C test within the last six months, or their last A1C results were high, a protocol was built around that data to alert the care team to test that patient’s A1C and perhaps prevent or properly treat diabetes.
While such methods to close gaps in care are very powerful, they are not ideal. First, the data obtained for these patients was often incomplete. It didn’t give a holistic view of the patients as it contained only what the hospital or health system could access. In the previous example, a patient might have gone elsewhere for laboratory testing or care and found not to be at high risk for diabetes. It’s amazing to see that we were able to make data actionable, but the actions we were taking against that data as an industry were insufficient or inconsequential.
Data Harmonization Fuels Better Results
Today, applying those same principles to tackling gaps in care at the data layer, data can be harmonized from all the different entities where patients might be treated or tested. For example, using connectivity and secure data extraction capabilities, an accountable care organization (ACO) that connects to 30 different EHR systems across more than 100 practices is finding that the actionable data they extract is much more complete. If they identify a patient as high risk for a disease, they are more confident that designation is true, because they have pulled data about that patient from so many systems.
Progress Has Been Made – But Do We Know What We’re Missing?
It’s great to see that in the past five years or so, we’ve taken the concept and the workflow for personalized medicine and moved it to something that’s very data-driven. Outcomes started to change when the interoperability piece of technology became more prevalent. Now we’re seeing personalized medicine grow in leaps and bounds, because the identification of gaps in care is much more achievable through connected data.
However, we’re certainly not done. Yes, we’ve identified some gaps in care, and created personalized protocols for specific populations. But we still haven’t developed a way to use non-standardized, denominating rule sets beyond those recommended by the National Quality Forum, which is very limiting to the potential of personalized medicine. The information you don’t have is often just as important as the information you already have. Machine learning algorithms that look at big data sets, allow us to find additional at-risk patient populations we’ve never even considered…once we have enough data to do so.
Clinical Data Repository: A Single Source of Truth
We’ve had healthcare technology communication systems for a long time, and they’ve mostly looked like health information exchanges. They basically just move one piece of data – an admission, a discharge, a transfer – from one place to another. Why not have a clinical data lake, that’s comprised of data from many different sources, to effect change in how medicine is delivered? Personalized medicine will be realized looking at all of a patient’s lab results, not just those from a recent hospitalization or a single doctor’s order.
It’s time for the healthcare business to catch up to other industries, like retail and finance, in creating and using massive, diverse data lakes. Our industry now has the competitive landscape and financial drivers to accomplish this. Amazon can tell customers what they want even before they even know that they want it, because Amazon applies machine learning algorithms to a lot of data. The same possibilities exist, and are starting to be realized, in our industry. Data repositories will enable the advanced technologies that are fueling the growth in these other sectors. Once the data is all in one place, technology will do the heavy lifting. A computer’s ability to recognize patterns is something doctors can use to save lives. Radiology reports, for example, have been the subject of some fascinating machine learning studies. Highly sophisticated software has been able to process large collections of digital images and recognize cancerous areas that radiologists did not see. This technology is not intended to replace highly skilled radiologists, but rather to assist them in making the most accurate diagnoses – and enhance the outcome of a personalized medicine approach. Humans will always be cared for by humans, but data will be a vital tool to make that care even more complete.
The bottom line: any company in healthcare, be it a vendor, a third-party supplier, a hospital system or medical practice group, which invests in discovering and using as much data as possible will facilitate what can truly be called personalized medicine.
________________________________________
Cliff Cavanaugh is a founder and Chief Technology Officer of Healthjump, whose mission is to make health records available and useful to all members of the care continuum.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
