Reducing Clinical Variation in Medicine with Artificial Intelligence

Reducing Clinical Variation in Medicine with Artificial Intelligence

Let’s start with acknowledging that Medical Errors are now the 3rd leading cause of death in the USA, having surpassed COPD. In addition, studies have shown us that 30 percent of all medical care costs are unnecessary. While clinical variation is not the only reason for these two issues, it is a significant contributor. If eliminating medical errors and unnecessary medical care do not form the basis for a “Just Cause Initiative”, I don’t know what does.

As someone who has been skilled in medicine for 37 years, the last seven as a hospital CMIO, I can’t help but feel accountable for both these issues. The very fact that the majority folks try to deal with the movement of drugs from “volume to value", and lots of folks have a foot in each world and try to not fail! a number of us also are preparing for risk-sharing and if we are getting to do this, we must tackle the difficulty of Clinical Variation, or prepare to fail! during this article, I will be able to share how we, at Flagler Hospital, are reducing clinical variation to reinforce patient safety, improve our quality, while reducing the value of the care we offer.

Hospitals are trying to deal with this for a real while, usually with energetic clinical staff. Optimistic administrations have done the simplest they might, with chart reviews and data extractions. But these efforts are biased by the alternatives of what they decided to watch, and therefore the results are usually months old by the time the info is analyzed.

Over the past few years, we've had enormous advances within the hardware and software wont to store our patient data. Advances in AI have made it possible for us to seem in the least the info, allowing it to inform us what we've done well and where we've fallen short. It can now do that during a matter of minutes, not months or years.

At Flagler Hospital we've chosen to use AI software advanced by Ayasdi, which uses a particular branch of mathematics called Topology. This software allows us to seem in the least our patient data chronologically, creating a separately flagged event for each order, medication administration, lab result, and every one patient care tasks. In other words, everything that occurred to the patient. It then groups the patients into “treatment groups” that supported their similar care. Next, it shows us the Direct Variable Cost, length of stay, and co-morbid conditions, allowing us to look at and identify any statistical differences in these groups. 

We select the simplest treatment group, referred to as our “Goldilocks Group”.It demonstrates the simplest combination of lower direct variable cost, length of stay, readmission rate, and mortality. We use Ayasdi’s unsupervised learning algorithm to develop a CarePath from this treatment group. It tells us, “if you would like future patients to be during this group, here are the items you would like to try to, and therefore the timing and sequence with which to try to them”. We then edit the CarePath and deploy it, use it to form changes to corresponding admission and treatment order sets then monitor adherence of our physician to the CarePath.

While all of this is often music to our CFO’s ears, it's meaningless unless we will operationalize it. We, at Flagler, are fortunate. Seven years ago, we completed our “PIT Crew” (Physician IT Crew). It's comprised of 20 physicians, one or two from every department within the hospital, who got authority by the hospital board to form decisions concerning EMR workflow, order sets, and orders. They meet often and reassess all things “EMR” and may make decisions or recommendations to individual departments or the medical staff generally. This team has been vital to our EMR success and forms the “cornerstone” of our project to scale back clinical variation.

Two years ago, once I discovered Ayasdi, I brought an idea to my CEO who gave me the go-ahead for the Pilot. Hell Crew was brought in at the very beginning of the method, as were several Informatics staff members with excellent SQL skills. Here is how we've operationalized our use of Ayasdi:

1. We created 2,500 lines of SQL code to extract patient data from our EMR (Allscripts), our Analytics Platform (CPM), our surgical system. These queries were parameterized so we will easily move from one diagnosis to a different 

2. Once the info is extracted, one among our DBA’s uploads the info to Ayasdi

3. We complete several rounds of semantic and syntactic affirmation

4. We run the “unsupervised learning” algorithm of Ayasdi to get the treatment groups

5. We correlate the groups analytically using P-values and K-S test (Kolmogorov- Smirnov)

6. We present the groups to Hell Crew who review the info and choose on the “Goldilocks” treatment group

7. Ayasdi is then wont to generate the CarePath from the Goldilocks treatment group

8. A PIT crewman from the acceptable medicine and that I edit the CarePath

9. Admission and treatment order sets are changed to suits the CarePath

10. Education goes bent the medical staff

11. Order Sets and CarePath are deployed

12. Monitoring starts with the physician attachment report generated by Ayasdi

13. Quarterly reports go the Administration, the Hospital Board, the ACO and physician staff showing changes in Direct Variable Cost, Length of Stay, Readmissions, and Mortality

We concluded our first CarePath, associating Pneumonia patients from 1/1/14 to 6/30/18, in nine weeks. We completed our second, Sepsis, in a fortnight. We went on to finish COPD, coronary failure and are now on our 5th, Total Hip. We decide to roll out 18 CarePaths in 18 months.

We went to accept our first CarePath, Pneumonia in August of 2018. Our results portray a concession in Direct Variable Cost of $1,073 per patient, a discount long of Stay by 1.38 days and a discount in Mortality from 1.5 percent to 0.0 percent. With our efforts, we appear to get on the proper path to try to do our part in reducing clinical variation, also as reducing unnecessary costs and eliminating medical errors. I think that each one hospital, even sole community hospitals like ours, can do that well. Together we will improve the way healthcare is delivered for our patients, our communities and our nation!

Read Also

Day in the Life of COVID Reporting

Day in the Life of COVID Reporting

Amber Theel, System Director - Patient Safety and Quality, Adventist Health
Healthcare Quality: Who is in charge and why?

Healthcare Quality: Who is in charge and why?

Steve Tierney MD, Sr. Medical Director Quality Improvement, South central Foundation
In the Age of COVID: The Acceleration of Technology

In the Age of COVID: The Acceleration of Technology

Lynn Gibson, VP/Chief Technology Officer, Christus Health
Revamping Workflow in Pharma Industry with Technology

Revamping Workflow in Pharma Industry with Technology

Bozidar Jovicevic, VP, Global Head of Digital Therapeutics, Sanofi [EPA:SAN]
Progressive Innovation Improving Patient Outcomes

Progressive Innovation Improving Patient Outcomes

Emanuele Degortes, Head of Patient, Innovation and Access Policy, Vifor Pharma