Executive Network
Click Here, Learn More
Nathan Patrick Taylor
Director of Data Science and Analytics, Symphony Post Acute Network
Supernova Award Category:
Data to Decisions
The Company:
Symphony Post Acute Network is one of the most innovative providers of post acute care in the United States. They are a proud, family-focused operation that takes a proactive approach to delivering quality outcomes for guests in their care. Symphony currently has locations in Illinois, Wisconsin, Arizona, and Indiana.
The Problem:
Symphony wanted to use AI and machine learning to improve care for its 80,000+ patients a year recovering from procedures like knee surgery or receiving dialysis treatment. They understood that deep inside a patient’s medical history could be indicators and data points that can predict if a patient is particularly at risk for readmission or certain medical conditions (i.e. more likely to suffer a dangerous fall), and therefore require extra precautions. Being able to predict these items allows Symphony to take corresponding preventative measures and improve patient outcomes while reducing costs. Unfortunately, Nathan and his team didn’t have the resources to deploy machine learning meaningfully or at scale -- traditional modeling methods were time consuming and prevented him from tackling as many projects as he wanted. The company hired two more data scientists, but they were expensive and not getting the return that was expected. He knew if his team could not only build models faster but also deliver predictions that were more accurate, they could improve patient quality of care and reduce expenses.
The Solution:
After years of being slowed down by manual processes, Nathan turned to DataRobot and its automated machine learning platform. DataRobot’s platform automated many of the previously manual and time-consuming processes, which greatly upleveled his team. Each member of the data science team was now doing the job of 6 or 7 people, and Nathan was able to build more models and get more accurate predictions to prevent hospital patient readmissions and improve quality of care.
The Results:
Implementation was the first challenge: most people have a hard time understanding math, so Nathan needed to learn how to communicate the results of his predictive models to the doctors and nurses at Symphony’s facilities and what that meant for them, without using data science terminology.
The second challenge was technology: some of Symphony’s doctors and nurses refused to let a “robot” make decisions for them. But once Nathan encouraged nurses to dig deeper into patient’s data to see why the model was making the predictions – and once predictions proved to be extremely accurate – skeptics were convinced and converted.
The third challenge was that as soon as everyone recognized the power of automated machine learning, Nathan’s team was inundated with requests, as nurses and doctors believed DataRobot was “a magic wand” that could deliver all sorts of predictions.
DataRobot is now a core part of Symphony’s analytics strategy, allowing Nathan to drive a far more ambitious predictive analytics agenda internally, delivering more accurate results that drive cost savings and generate revenue.
Metrics:
DataRobot has driven significant impact for Symphony. A few key metrics include:
- Reduction in readmission rate: According to Nathan, “If you can [reduce readmission rates] by 1%, you’re doing really well.” With DataRobot, Symphony was able to improve readmission rates from 21% to 18%. Each readmission costs the company $13,500.
- Substantial increase in productivity: The old process and machine learning lifecycle for building one model used to take one person approximately 62 hours. With DataRobot, they were able to bring that down to 10 hours. The team also experienced 6x increase in individual productivity and efficiency, and a 3x increase in number of models built.
- Advancement of near real-time predictions: 240 doctors and nurses across Symphony’s network get predictions from models built in DataRobot, productionalized in Symphony’s PowerBI dashboards. Results funneled back into dashboards every four hours, and those predictions are actioned on for up to 80,000 patients per year.
The Technology:
DataRobot’s automated machine learning platform
Disruptive Factor:
Nathan was constantly up against challenges internally as he rolled out more robust and better predictions. Some of Symphony’s doctors and nurses said “I will not let a robot make decisions for me.” A nurse once challenged Nathan’s model and pointed out that a patient who had been deemed a “readmit risk” could not possibly be, and that the model was wrong. Nathan encouraged her to dig deeper into the patient’s data – based on “prediction explanations” that DataRobot had highlighted – and once the nurse did that, she discovered several risk factors she wasn’t aware of, such as a history of mental illness. The nurse was instantly converted, and as more “haters” became champions, the DataRobot technology caught on and changed people’s minds.
Once that happened, stakeholders all across Symphony – including both executives above Nathan and end-users like doctors and nurses – demanded more predictions from Nathan.
Shining Moment:
Nathan was named a “Data Science Superhero” by DataRobot in 2017.
