Reducing 30-day Readmissions on Rocky Top -- Testimonials Title
Tennessee. It is a beautiful state, with wonderful people challenged by pressing health issues. Ranked 45th in overall health, the state ranks 47th in obesity, 49th in physical activity, and 46th in smoking. In an October 2014 release, Kaiser Health News reported that the majority of Tennessee hospitals were fined my Medicare for readmission rates.
This is a tough population to serve, and this central Tennessee medical center sees some of the most complex and challenging medical cases within the state. Top this off with new mandates, penalties, and value-based purchasing requirements, and they face what feels like an insurmountable task. It’s not just enough that they have to deal with readmission penalties. With the population they serve, the medical center has to work hard to stop complications from chronic conditions and patient deterioration.
The provider’s Chief Quality Officer wanted to reduce readmissions as part of an overall drive within the organization to improve quality and patient outcomes. Her idea was simple: give the medical center the ability to predict which patients were at risk of a readmission to prevent patient deterioration and improve health.
Jvion’s clinical predictive solution combines clinical rules, advanced statistical modeling, and deep machine learning to deliver the most accurate patient-level risk predictions available. Leveraging the latest data science approach based on the development of clinically relevant data pods, the solution's predictive engine uses the data that a provider has on hand to hone in on patients at risk of patient deterioration, illness, readmissions, nosocomial events, and other target diseases. These predictions, along with contributing risks, interventions, and engagement scores, can be rendered directly through the existing electronic health record system and can be delivered within weeks of project initiation.
In less than three months, the medical center has seen a 10% drop in readmissions. In fact, month-one model performance delivered three times the accuracy of the system’s existing LACE scoring tool. And this accuracy is expected to increase exponentially as Jvion's solution consumes more data and continues to “learn” the provider’s patient population.
Future plans include expansion of the number of predictive use cases that are live within the provider’s environment including hospital acquired conditions, patient deterioration, and community health.