CSP 2017 Online Program

Communicating with Clinicians About Models That Predict Risk Using Interactive Web Graphics (303464)

Keywords: interactive graphics, prediction, risk prediction, personalized medicine, D3, htmlwidgets

Models that predict the risk of a clinical event using patient information and biomarker data are often used to guide decisions about patient care. It is important for clinicians using these risk-prediction instruments to understand both model parameters and the accuracy of the predictions being made. In collaboration with clinicians, we have developed several interactive web-based graphics that aim to educate and communicate the impact of using risk-models to guide treatment. We show an example of these tools applied to a risk-model developed for men undergoing active surveillance for prostate cancer, and describe how a new user can generate these web-based presentations for their own risk assessment.