SCF realized that equitable workload distribution was critical. SCF needed a method that could predict what the upcoming future of expected work would be for its panels of customer-owners, and it needed to be regularly updated so panels could be open or closed as they became full. After selecting criteria, SCF used a machine learning tool with a goal to have a weekly update on the level of work for each of SCF’s Integrated Care Teams.
The method is titled the Nuka Model, which combines several prediction models into a hybrid model to predict customer-owner primary care utilization over the next six-month time frame. A team’s Relative Value of work was the unit decided upon, and the definition was focused on the work a medical home team would perform with a panel of customer-owners. Specific models for adults and children, and for those with and without any recent care utilization were developed.
The Nuka Model works by calculating Relative Value Units (RVUs), which are the relative level of time, skill, training, and intensity it takes to provide a given service. The more time or intensity a service takes to provide, the higher its RVU. So for example, a preventive visit with a younger customer-owner would have a lower RVU than a preventive visit with an older customer-owner.
There are a large number of variables that go into predicting RVU, including:
Many other variables are also used.
The Nuka Model has been effective at predicting utilization. At the customer-owner level, it predicts 52% of the variation in primary care utilization, compared to only 38% for the predictive model previously used by SCF. And at the panel level, it predicts 89% of the variation in primary care utilization compared to 51% for the previous predictive model. SCF has found the Nuka Model to be a valuable tool for workload balance in its primary care clinics.
For more information about the Nuka Model for panel balancing, or any other aspect of SCF’s Nuka System of Care, feel free to contact the SCF Learning Institute.