Team profile by members of the Data Science in Clinical Care project team

Clinical Decision Support (CDS) tools are data science innovations that promise to improve healthcare outcomes. For example, they can predict which patients are at risk of rapid deterioration and alert providers to act.

While the Food and Drug Administration provides guidance on the technical development of CDS tools, more needs to be done to ensure that they are well-suited for the clinical setting, connecting development and implementation so that the tools achieve their objectives in practice.

This project focused on developing a deeper understanding of how to connect development and implementation effectively and propose a formal process to facilitate the adoption of CDS tools in different clinical contexts. To achieve this goal, the team used a methodology called Participatory System Dynamics Modeling, which involves structured, facilitated workshops with staff to co-develop mental models of the problem of CDS adoption over time.

During the first half of the year, the team focused on understanding the problem. Team members conducted a review of literature on clinical decision support implementation and met with subject matter experts from the Duke Algorithm Based Clinical Decision Support (ABCDS) governance committee. The team developed and presented a poster about the methodology at the December 2023 AI Health Showcase.

During the second half of the year, team members focused on using system dynamics modeling to study readmission risk score adoption at Duke Health. One group worked on developing a taxonomy of clinical decision support tools to characterize the readmission risk score. 

Another group conducted a workshop with clinical staff to develop a visual diagram of facilitators and barriers to adoption of the readmission risk score. The workshop revealed awareness as a key factor affecting adoption and pathways that have affected staff awareness over time.

The team presented the taxonomy, workshop and findings at the 2024 Bass Connections Showcase. Deliverables will also include a report to health system partners.

This project will continue through 2024 to incorporate other CDS use cases into the CDS adoption model from group workshops with staff.

Engaging Multidisciplinary Teams to Develop a Model of CDS Adoption

Poster by Nina Sperber, Scott Rockart, Shatanshu Choudhary, Adam Johnson, Hannah Groos, Samantha Hamelsky, Afraaz Malick, Saanvi Pawa and Kriti Vasudevan

Engaging Multidisciplinary Teams to Develop a Model of CDS Adoption

Poster by Nina Sperber, Scott Rockart, Hannah Groos, Samantha Hamelsky, Saanvi Pawa, Kriti Vasudevan, Shatanshu Choudhary and Adam Johnson