Lillian Dittrick, FSA, MAAA, a fellow with the Society of Actuaries, uses data analytics on patients at Unity Health to combat risks stemming from the treatment of diabetes. Using data analytics, Dittrick was able to identify patients with chronic conditions, manage costs and implement quality control measures, all while maximizing patient outcomes and minimizing financial risks.

In an exclusive interview with Clinical Innovation & Technology, Dittrick talks about her program, how it can be used in other areas of medicine and how technology has made these advancements possible.

Clinical Innovation & Technology: You’ve used data analytics to manage costs and increase quality control measures for patients who are overweight and facing diabetes. How do you think this program can be used in different areas of medicine?

Lillian Dittrick: It’s very transferable. I’ve focused on diabetic cases but we have queued up other chronic diseases. We are doing COPD next so the process of identifying chronic diseases through predictive modeling and using tools to find gaps in care—all of this can be applied to any chronic diseases. All you have to change is the underlying diagnosis that drives those diseases.

How do you think the treatment of chronic conditions, like diabetes, has changed over the years?

From a data analytics perspective, I think what has changed, especially when using predictive algorithms, is now you’re able to not only just look at past history utilization. You’re using this tool to predict what will happen in the future. We are able to get in front of diseases more quickly and prevent people with rising risks from tipping over into the high risk category. You’re managing all of the triple aim with your patients, so getting in front of the disease helps them achieve a better quality of life as well as managing costs.

How has the transition away from pay-per-service and to pay-per-performance affected caring for chronic conditions?

I believe care has improved because when you’re focusing on the value of the proposition, you look at the outcomes as well as the cost. This makes for a better all-around view of the patient and makes for better patient outcomes. You’re making sure you’re taking care of them and meeting them where they need to have care. Sometimes it may be recognizing that the best place for them to receive care is at home rather than in a hospital—and that leads to the value of proposition.

Where do you think the measures for improving patient experience, decrease costs and increase quality could be the most useful?

For the whole triple aim, it applies to no matter where the patient is in the stratification continuum; its just how you apply it. So for someone who is a low risk patient, you’re looking for things for them, like scheduling an appointment online or being able to text them. You’re figuring out how to meet them and their triple aim needs all the way up through the continuum to a high risk patient in any condition. You need to make sure you have good transitions and care for them, so nothing is falling through the gaps in their care, whether you’re in in a hospital or at home.

What was the most difficult part when trying to achieve the results of improving patient experience, decreasing cost and increasing quality of care?

When running a predictive model, it's always a challenge to bring together a lot of disparate data sources and then cleanse and normalize them to be able to use them in our model. You’re weeding out the false positives and getting the best results possible.

Explain how advancements in technology have helped you achieve these results

Data technology has come in as the biggest help. When dealing with very large amount of data, one of the things we are working on is a natural language processor. We have millions of records that sit within our electronic medical record (EMR) system that are in an unstructured format. There isn’t a specific field filled out where you can extract that information. It's people’s thoughts and notes that they’ve written. So the tremendous amount of valuable information in there that can help with predictive models, stratification and scoring. It hits so many levels and we need the advanced technology to be able to read that language and turn it into structured data so we can use it in our models.

What do you believe is the most underutilized method or device that could be implemented to help other medical facilities achieve these goals in their own systems?

The natural language processor, just because there’s so much data that’s not being used and holds a lot of valuable information. I think that you will see that technology become more and more prevalent, and not just in the healthcare industry. Working to better understand these very large data sources and how they can meet people’s needs no matter what industry you’re looking at.

What do you believe is your biggest achievement when you were conducting this program?

Through this whole process my team implemented a system wide stratification of over 1.2 million patients and we haven’t been able to do that before. We’ve taken that information and pushing it into our EMR system and onto a common care plan that is available to all clinicians in all system to see, no matter where you’re caring for a patient. It’s our front line information on what kind of health status the patient is in and to help them wrap the appropriate care around our patients.