Statistica and Medicine: Q&A with Dr. Steven Melemis

In this post, I share highlights of a fascinating conversation with Dr. Steven M. Melemis, a long-time user of Dell Statistica and recognized expert in the field of addiction medicine. After earning a Ph.D. in statistics, Melemis went back to school for a medical degree and has merged his two academic pursuits to improve patient care and streamline emergency room operations.

Q: When did you first recognize the data/healthcare connection?

Given my background in statistics, I saw a long time ago that data drives almost all of our decisions. So if we can better analyze data, we can provide better patient care. The challenge in healthcare, however, is that many physicians suffer from a fear of statistics, especially now when the massive volumes of data can be intimidating.

Q: So, what’s the best way to help physicians conquer that fear?

While I enjoy analyzing complex data sets, I realize not everyone wants to dig that deeply. Still, you can learn a lot from simple data visualization. I tell people to look at the charts—they’ll tell you what you need to know. Statistica’s quick-start data mining recipes, analytic workflow templates and out-of-the-box analysis capabilities make it much easier to gain meaningful insights.

Q: When did you first start using predictive analytics software?

In my early academic career, I used statistical software that ran on mainframes. During my post-doctoral fellowship in the 1990s, however, I wanted to find something I could use on a PC. That’s when I first found Statistica—and I’ve been using it ever since.

What I like best about Statistica is it works across the entire spectrum of predictive analytics. While it features sophisticated tools, it’s also suitable for people who just want to draw a few graphs and really look at their data in different ways. When data is more accessible, you can put capabilities in the hands of the people doing the actual work and empower them to see things they didn’t see before.

Q: How has predictive analytics transformed alcohol withdrawal treatment?

It has dramatically improved how patients are monitored in emergency rooms and rehab settings. For years, a classic tool called the Clinical Institute Withdrawal Assessment (CIWA), was used to assess whether patients needed treatment for alcohol withdrawal. The labor-intensive CIWA required a nurse to monitor a patient by measuring 10 different variables every hour. The problem: Some people didn’t get the entire test because there simply wasn’t enough time or resources, and occasionally, patients fell through the cracks.

I knew there was valuable insight buried in the data, so I digitized a small core of CIWA measurements for 100 patients and fed it through Statistica—and one variable stood out! Determining if a person needs withdrawal treatment can be accurately predicted 75 percent of the time by simply determining if the person is sweating or not. Adding one more variable—perceptual disturbances—and the predictive accuracy jumped to 90 percent.

I took that information to the hospitals in Toronto, where I live and work, and also shared it with leading rehab programs. The bottom line: Treating patients for alcohol withdrawal has become faster and more efficient. All from a simple yet powerful data analysis performed by Statistica.

Q: How has predictive analytics changed your approach to addiction recovery?

My approach is best summed up in an article I wrote recently for the Yale Journal of Biology and Medicine where I outline relapse prevention and the Five Rules of Recovery. Predictive analytics have shown me that people can greatly improve their chances of recovery if the follow a few simple guidelines.

By empowering physicians of all kinds to take advantage of valuable data analytics, we can make major strides in personalizing healthcare and treatment plans while lowering healthcare costs and elevating patient care delivery.

And, the best part is that you don’t need to have a Ph.D. in statistics. All you need is an easy- to-use tool like Statistica to simplify big data analysis and improve critical decision making.

About the Author: John Whittaker