HealthCare Analytics Needs To Go Beyond "Admiring The Problem"?

HealthCare Analytics Needs To Go Beyond "Admiring The Problem"

The world of healthcare—and, in fact, all—analytics is not monolithic but ranges from solutions that “admire problems” to those that “prescribe actions that create value.” The further healthcare institutions move to the right of this spectrum (i.e. using analytics that create valuable prescriptions rather than just describing them), the more they will experience the real benefits of Analytics.

The spectrum of Analytics

The Spectrum of Analytics

Descriptive Analytics

The most common forms of analytics are “Descriptive.” They merely  “admire the problem.” Imagine getting on a scale or using a thermometer. Before doing so, most people have a defined hypothesis they want to confirm about their weight or their temperature. The answer they receive from these tools either confirms or denies their hypothesis. But it does not tell them the underlying cause for the output nor does it give them a sense of how to move forward. 

 In the healthcare world, the parallel to this are the multitudes of dashboards and reports all kinds of committees and leaders receive from the EHR, Excel, Tableau and other “analytics.” For example in the OR, knowing your historical case volumes, times, turnovers, or delays merely underlines the issues you probably already know you have. This information alone is not sufficient to enact meaningful change.

Diagnostic Analytics

Diagnostic Analytics are better but still insufficient. Now imagine that once you look at your thermometer and see you have a high fever, perhaps you can surmise that “I have a fever because I was out in the rain for a couple of hours and got soaking wet.” That is certainly progress, but it still doesn’t help cure your fever. In a similar way, all the slicing, dicing, and deep diving that diagnostic dashboards provide are like explanations of yesterday’s weather. Wonderful for understanding where and when last night’s storm occurred, but still not helpful in deciding whether to take an umbrella with you when you go out today.

In the OR world, diagnostic analytics might help you understand that Dr. Jones’ cases always run late because he habitually requests too little time and doesn’t think in terms of “wheels in to wheels out.” However, they still don’t solve anything unless and until they can help Dr. Jones learn to request the right amount of time in the future and give him an easy means to do so.

Predictive Analytics

Predictive Analytics can help you start to plan. The real power of analytics starts to come into play when we gain the ability to forecast meaningful future events. Think about Google Maps and how it predicts that it’s going to take you 54 minutes to get home from the airport three days from now after you land at SFO at 5:00pm. That’s information that can help you plan the rest of your evening. How does Google Maps work such magic? By mining historical data from millions of trips drivers have taken over the years—by day of week, segment of road, weather conditions, whether it’s a public holiday or not, and a host of other factors. The app has no way of knowing exactly who will be on the road when you take your trip, nor can it rate their driving skills, but it can access a lot of data upon which it can build predictive models and your likely drive time from point A to point B. In a similar way, timestamp data can be mined to help an OR scheduler predict the surgeons/block owners who will not use their time well in the future or help an infusion center manager conclude, “I’m going to have a chaotic day next Wednesday between 10:00 and 2:00 and I need to plan for it.” Analytics really start to add value when they give you specific information about the future you can use to solve a potential problem.

Prescriptive Analytics Create Value 

The most useful type of analytics is one that drives high-value actions—for example when Waze tells you to take a different route to shorten the length of your journey. Or when FedEx and UPS use forecasting algorithms to predict the volume and mix of packages they will receive from all around the country and are able to put the right number of planes, trucks, and drivers in the right places at the right times to handle the stochastic demand. Examples in the healthcare world: a surgical department is able to alter its staffing patterns for a future day on which a surge in case volumes is predicted, or an infusion center is able to anticipate a chaotic Wednesday afternoon two weeks from now and shift a couple of patient appointments to better slots in order to “flatten the peak” and fit more patients in.

Hospitals need better tools, with better math and more predictive and prescriptive capability than EHR reports and dashboards can provide. Operational teams will need to go beyond describing or diagnosing problems to actually predicting what’s likely to happen and making action adjustments in anticipation—as illustrated by Waze, Uber surge pricing, and so many other real-world examples we all encounter in our day to day lives.

This article first appeared in Forbes


Alona Michelle Waldrop-Kelly

GCP QMS Specialist CQA-ASQ

3 年

I wish you had not used package delivery service example. The last tiny step is getting the right person's mail. I just walked a package to my neighbor yesterday, as I have done occasionally for 20 years. I would rather pick up at stores myself and I don't even bother ordering food delivery. 99.9% this step, and I would be in love. Like "Warning, this package doesn't belong here!" Maybe a barcoded front door, lol.

回复

要查看或添加评论,请登录

社区洞察

其他会员也浏览了