Achieving Individualized Precision Prevention (IPP) through Scalable Infrastructure Employing the USPSTF Recommendations in Computable Form

The U.S. Preventive Services Task Force (USPSTF) provides evidence-based recommendations to improve health. As the number and complexity of the recommendations increase, it becomes more challenging for primary care providers to personalize and prioritize preventive services for their patients. Thus, capabilities to tailor and prioritize preventive services for specific individuals are needed. The idea of a learning health system (LHS) has gained increasing recognition since the appearance of a seminal report from the Institute of Medicine: health systems that effectively apply data and evidence to improve patient outcomes and care are called learning health systems. A LHS “learns” through a cyclical process, including assembly and analysis of data relevant to their important problem, which leads to discovery of new knowledge from the data. The learning cycle is completed by direct application of that knowledge to change practice. Changed practice generates new data, driving the next iteration of the learning cycle, which leads to improvement over successive iterations.

Researchers at the University of Michigan applied LHS methodology with the goal of advancing individualized precision prevention (IPP) for grade A and B USPSTF recommendations, or those recommended for all eligible patients. The Knowledge Grid (KGrid), a broadly applicable platform created by the research team, was used to support the knowledge-to-practice aspect of the LHS. KGrid includes knowledge objects, a digital library to hold and manage learning objects, and an activator to deploy the objects as digital services. Enhancing the infrastructure, the Substitutable Medical Applications & Reusable Technologies (SMART) initiative supports an application ecosystem to extend the capabilities of electronic health records (EHRs). The study used KGrid, with support from SMART, to build and test a new application capable of automatically personalizing and prioritizing preventive measures to achieve IPP.

The specific aims of the study were as follows:

• Use KGrid to design and develop computable knowledge objects for USPSTF A & B recommendations, along with an executive object capable of applying computable risk and benefit models, forming a core knowledge object collection for individualized precision prevention (KOs-4-IPP). 
• Design, develop, and test a shareable SMART IPP APPLICATION that can be integrated with various EHRs and which draws on KOs-4-IPP services from KGrid to provide IPP information. 
• Conduct a study with primary care providers as subjects to assess the potential utility of IPP information. 

Using realistic patient scenarios, when testing concordance between the IPP algorithm’s ranking of preventive service recommendations and primary care provider rankings of those same services, researchers found an intermediate level of concordance. This finding suggests that there may be a role for IPP algorithms to help guide selection and enactment of preventive services in practice. Researchers developed shareable computable USPSTF knowledge objects, a shareable web app, and then used a special ranking questionnaire with 12 realistic patient scenarios that culminated in data collection and analysis that led to initial insights into how to effectively use these resources to individualize preventive services.