This is the age of advanced analytics in healthcare: population management, predictive analytics, personalized medicine, and clinical decision support. Laboratory analytics are a dominant — in fact, disruptive — solution well-suited for many health systems.
Laboratory-based predictive analytics
Laboratory-based predictive analytics require an automated application to pull lab test results for an individual patient. And then there’s a prediction engine that has been trained using something like a Bayesian system, neural network, or any of a number of different machine learning applications. This would essentially give a probability — a risk score — for that particular patient, and that patient would experience a particular event. Let’s call this patient Mr. Jones.
Mr. Jones is admitted to the hospital and has admissions labs. Using lab-based predictive analytics on the admissions test results would tell you, for example, that Mr. Jones has a 20% chance of being readmitted. And that score would be delivered to Mr. Jones’ clinicians through, perhaps, an EHR or mobile device. The hospital would then use that information to modify the way in which they’re managing Mr. Jones.
These readmission predictions are broadly used right now. Where the lab is particularly important and more valuable than other types of readmissions predictive analytics is in that you can get to disease-specific readmission applications, which these health systems have been wanting. That’s really where you need lab data as opposed to other sources. A lab application will most likely be your best way to get a predictive tool for a specific disease.
Where the lab is particularly important and more valuable than other types of readmissions predictive analytics is in that you can get to disease-specific readmission applications.
The opportunity for health systems
If we think about their needs and capabilities with respect to advanced analytics and value-based care, health systems can be broken down into three segments: large health systems, a big mid-size sector, and small systems or even hospitals.
Large health systems
Many of the large systems, even if they’re not ACOs per se, need ACO-level analytics. What I mean by that is that they need, increasingly, full-scale population predictive analytics. They also need an enterprise-wide coordination solution. So, they really do need to invest to coordinate across inpatient, outpatient, post-acute — everything. Large academic centers, like Partners in Boston, probably also want to distribute their own clinical knowledge as intellectual property (IP). Many of them probably have the capability to integrate their EMR. They may have analytics groups, and they certainly have substantial budgets — in the billions — to put this together.
Mid-size health systems
Industry analysts have verified that it’s different for mid-size health systems. They certainly need advanced analytics, but instead of it being a full-scale solution, they need a “program-based solution” for particular areas, like diabetes and heart failure. They certainly are going to be ACOs as well, but it’s a more targeted type of situation — similarly so for care coordination. These are systems that probably have a lot of bundling going on, so it’s condition-centered as well. But they don’t have the analysts in-house to build these tools for them. In terms of EMR integration, they’re unlikely to have a full enterprise data warehouse in most cases. They have some integration, but not a lot. And, most importantly, they don’t have billions of dollars to spend on IT analytics. Certainly they don’t have the billions of dollars that the large systems do.
This mid-sector, as well as the smaller systems and hospitals, can’t afford a huge enterprise data warehouse that is trying to integrate all sources of data. It takes three or four years to build one, and even if they had the time, they don’t have the capabilities or the resources to do so. Most importantly, the return on investment doesn’t make sense for them. So, an advanced analytics program that is primarily lab-driven — not to say it can’t have any other types of data; it certainly could — is a perfect disruptive solution for this group.
The return on investment doesn’t make sense for them. So an advanced analytics program that is primarily lab-driven is a perfect disruptive solution.
Small health systems and hospitals
The small systems, of course, are even more limited in their capabilities and integration. Their analytics needs are really more targeted, lab-based “point solutions” and professional services.
Laboratory data provides a disruptive solution
A “disruptive solution” is a term that was coined by Harvard Business School professor Clayton Christensen. Essentially, it’s a product or offering that comes into the market and provides a simpler solution that fits a user’s simpler needs, and does so at a lower price point. The user’s needs are simpler in that they are not as great as the ones the product already existing in the market was designed to meet.
Pathology and laboratory data has a tremendous role to play, and the time really is now. The advanced analytics wave is really taking off at this moment in time. And there’s a whole sector of healthcare with needs that aren’t being met by large enterprise data solutions. I believe lab has the power and the tools, and is very well positioned to serve the advanced analytics sector. It’s an exciting time to be in lab and pathology medicine.
For more of Dr. Herriman’s insights into the new era of analytical tools, download “Advanced Laboratory Analytics — A Disruptive Solution for Health Systems“.