Learning from other industries, health care is catching on to how consumers shop, compare options and make selections that best match their needs.
At home, you watch a movie. Then the system compares you with other people who watched the same movie and recommends other films based on similar connections. Or you go online and shop for clothes and then see ads for items in the same style and price range others like you also viewed.
Using the same kind of technology that allows online stores to serve up personalized ads or entertainment sites to recommend programs you might like, health care organizations are starting to determine — and more quickly meet — the needs of health care consumers.
A great way of demonstrating how health care is helping consumers make decisions with information personalized to meet their needs is a model called Next Best Action.
Next Best Action
This approach customizes recommendations for each individual to determine the treatment path others like them chose, identify what was successful and what was not.
OptumRx® uses Next Best Action to create a personalized experience for each individual. For example, when a call comes in to our call center, each call is treated as an opportunity to present personalized recommendations for an individual’s health and savings. This may include condition management, pharmacist medication consults, lower-cost drug alternatives, wellness programs and behavioral services.
Next Best Action is based on the principle that two individuals dealing with the same health issues may react differently to medications and treatment. The goal is to present health care consumers with a personalized course of care that they have a high likelihood of following.
For OptumRx, these predictions are so accurate that today members will accept the engagement option over 60% of the time.1
Graph technology forward
Graph technology is emerging as a relationship-building method to power the analytics behind next best actions. Graph technology is a way of storing data that allows the relationship between data points and the data points themselves to be treated as equally important. This allows for the recommendations you see on your streaming services.
Drawing from the data and relationships between data points, consumers are presented with personalized recommendations in seconds based on their unique health histories.
Simply put, graph technology means that no single piece of information lives in isolation. A graph database helps us see and understand connections between the data, which is perhaps nowhere more important than in health care. The complexity of health care is seen in the amount and diversity of data required to create a complete picture of the human condition.
How does graph technology fuel machine learning?
Machine learning, a subset of artificial intelligence, is an advanced analytic method used to identify next best actions. Machine learning models are designed to become more accurate at generating predictions as they are fed increasingly large volumes of information. To do so often requires drawing data from vast and disparate datasets. Graph technology translates disparate datasets into connected relationships and precise knowledge that machine learning can use and process faster.
When large amounts of data are stored and processed at once, there can be a time gap between data storage and data analysis. Graph technology powers the use of advanced analytics, such as machine learning, to deliver insights in minutes compared to what could take hours.
Querying relationships within a graph database is fast because they are perpetually stored within the database itself. Relationships can be intuitively visualized making them useful for heavily interconnected data. It also saves time by connecting different health care record systems and analyzing the data simultaneously.
Our most recent engines that use graph technology can do these calculations in less than a hundred milliseconds.
Consider the value for a health plan call center that might be utilizing three different databases — one for dental, one for vision, one for health. Without graph technology, it might take 30 minutes or more to gather information from all the databases and make sense of it. Why is this important? When it comes to our health, we want the best answers from our doctors and other health care providers expressed to us as simply and quickly as possible.
A story behind each data point
As an engineer, I think it‘s important to know that our work here at Optum is more than just about the technology. It is about understanding that behind each data point is a personal story. This is even more important when it comes to clinical decision support.
Our technical engineers and clinical experts work to develop graph technology that can be applied in health care for most any member or patient encounter. The ability to synthesize millions of data points in real time can be embedded into the electronic medical record systems that a clinic is already using, allowing providers access to highly precise clinical decision support.
I often think about my own family experience and how the appropriate use of graph technology and other tools can help. I remember a relative seeking help for chest pain who faced a number of invasive and expensive tests and multiple trips to different facilities.
At the conclusion, we found the underlying cause was primarily due to indigestion. With graph technology, an individual’s health history can be compared to the health histories of many other people like them. The graphical relationships can help to analyze what actions others took and understand the outcomes to better inform clinical recommendations.
One technology may not provide the complete answer. But enabling providers to make faster, more precise clinical decisions could have helped identify the low probability of cardiac problems and the high chance of indigestion problems. We may have been able to prevent tens of thousands of dollars in invasive tests along with several sleepless nights.
Not a good fit for some, ideal for others
Even though graph technology has many advantages, I want to point out that it may not be the best technology for all situations. For example, if you have transactional data and do not care about how it relates or connects to other transactions, then there are other options that should be explored.
But overall, the applications and benefits of graph technology in health care are clear. Health care is so complex, involving many different influences on clinical outcomes and costs. Graph technology thrives on gathering and analyzing huge volumes of messy, complex data.
For predictive models to work, you also need a lot of accessible data. So it is important to find a partner who can tap into diverse data sets across large populations. It is also equally important to select a partner who can ensure data is “explainable.” Information that is too complicated, or is not explained in human terms, will be ineffective in practice.
Here at Optum, we are not confined to one type of health care technology because of the depth of technical expertise and knowledge of the health care system we have. Of course, I am biased when it comes to graph technology.
I truly believe graph technology is ideal to better engage with consumers. With advanced approaches such as next best actions, we are able to build personalized relationships that truly matter to inform and shape a better health care experience for my family and yours.
Learn how Optum is advancing technology for human potential. Visit optum.com/technology.
About the author
Dan McCreary is a distinguished engineer at Optum as well as an accomplished author, speaker and evangelist for artificial intelligence and graph technologies. His background is in enterprise data architecture, enterprise data strategy and objective database architecture selection. Read his full profile in our people page. And if you’d like to work with Dan and our other engineers to help advance technology in health care, visit our careers page.