How a healthcare data scientist can aid in value-based care
As hospitals begin to make the transition to value-based care, data scientists can help analyze patient trends that will lead to improved outcomes and better quality of care.
In 2015, Congress made a big change in the way healthcare providers are reimbursed. Instead of the previous fee-for-service model that paid providers for each service performed, reimbursement would now be provided based on the quality of care provided -- a concept known as value-based care.
But the transition has been slow, and providers who want to achieve value-based care may not always know where to start. That's where a healthcare data scientist can help. A data scientist analyzes and interprets large amounts of data so that providers can identify trends and potentially improve patient outcomes. Not surprisingly, the healthcare industry adds an interesting twist to this traditional analytics role.
One of the main capabilities a data scientist should have is an understanding of the problem that needs solving, said Jordan Mauer, executive vice president of marketing and membership engagement at NovuHealth, a consumer engagement company based in Minneapolis that uses performance analytics and behavior science to create rewards programs. Not only does the challenge involve deriving insights from the data, but it also means determining if those insights are practical and can be used to help a hospital, Mauer said, adding, "It's so critical in a very complex healthcare ecosystem."
Jordan Mauer
Mauer said it's also important that a healthcare data scientist understand what's pushing value-based care and ask contextual questions prior to looking at data, such as the following:
How does a provider get reimbursed?
What is the patient's role in that process, and is the patient economically stimulated?
What is a health plan's position on quality or gaps in care, and how does it affect how providers are reimbursed?
Readmissions are fertile territory
Mauer said data scientists could analyze conditions that have a high rate of readmission, which providers have the most and a patient's past behavior to determine the risk for readmission. That data could then be used to apply resources to encourage a change in a patient's routine or behavior that would improve their health and hopefully avoid a readmission.
You'll run into situations in healthcare where you can show [clinicians] the data, [and] being able to tell it in a story is very valuable.
Josh O'Rourkesenior software developer, Xtend Healthcare
Patient history can be a factor in determining readmission, Frost & Sullivan analyst Victor Camlek agreed, but the patient's therapeutic regimen plays a role as well. A healthcare data scientist could use data collected from the patient after discharge to determine the likelihood of readmittance. Camlek gave an example of a 75-year-old patient with coronary issues who has just been sent home from the hospital. "That hospital doesn't want to have a 30-day readmission, so they need to track and monitor that patient based on adherence to the [treatment] plan and other patients in that category that have had similar situations," he explained.
A data scientist would analyze that patient's case based on similar cases to figure out the likelihood of a readmission, Camlek added. From there, the hospital could apply data from remote patient monitoring to future cases to shift the focus to population health management.
Informatics shines brightly
Camlek said a data scientist would likely be someone with an advanced degree, such as a master's or Ph.D., who understands statistical analysis and modeling and can turn raw data into recommendations that clinicians can act on. Beyond technical skills, a data scientist in a hospital needs to understand healthcare.
"That's where I start to think in terms of the medical informatics degree -- people who have medical degrees and decided they didn't want to be involved in medicine but want to be involved in some other aspect of healthcare administration," Camlek noted. "Someone who has medical informatics would understand the science, data monitoring, data modeling, data statistics, analysis, predictive analytics, stratified information, plus they would be able to speak the medical jargon, understand all the conditions and the words that are being said to them and communicate to the medical team."
Programming experience counts
A healthcare data scientist should also possess some level of software development skills, said Josh O'Rourke, a senior software developer at Xtend Healthcare, a revenue cycle management company in Hendersonville, Tenn. "Not like a senior-level developer, but at least be able to program in a language like Python or R," O'Rourke said.
In addition, a data scientist needs to communicate findings in a way that makes sense to physicians. "You'll run into situations in healthcare where you can show [clinicians] the data, but they won't believe you," O'Rourke said. "So you need to be able to overcome that obstacle whenever you encounter it. Being able to tell it in a story is very valuable."
Nurses as data scientists
For hospitals that don't have the financial resources to hire a full-time healthcare data scientist, a nurse informaticist can fulfill that role. Joyce Sensmeier, vice president of informatics at the Healthcare Information and Management Systems Society, said every nurse informaticist starts with a nursing degree and builds on that knowledge with informatics and IT training.
Sensmeier also outlined the skills a nurse informaticist uses, which are similar to the skills a healthcare data scientist needs. "Every nurse has to manage the day-to-day care for her patients and make sure that things get done in the right time frame," she explained. "All of that requires a lot of documentation and data capture and an understanding of what the data means, so I think that kind of oversight or analysis ... is something that will go from a regular nurse role to an informatics nurse role and beyond."
Joyce Sensmeier
A nurse informaticist would also be able to look at the impact of a treatment plan or the care given to a patient to determine which factors improve outcomes, reduce the length of stay or lower readmission rates. "Now that there's data in the system, and informatics nurses are helping us get that right, ... you can analyze that structured data and see what makes a difference and contributes to the value and the outcomes," Sensmeier said.
As healthcare continues to move away from the current fee-for-service model to one based on performance, finding the right person to support a value-based care initiative will become even more critical. But before the data scientist can start analyzing the numbers, the provider or organization should clearly define what they hope to learn from the analytics.
Next Steps
The impact of data analytics on value-based reimbursement
Small medical practices will struggle with value-based healthcare
Analytics and informatics collide to manage healthcare data
Dig Deeper on Healthcare IT systems and applications