Risk Prediction and How the HAS-BLED/CHA2DS2-VASC Assessments and Scores Interconnect- 2022 Quality Summit presentation (Non-CE)-Session

Video Transcription

Video Summary

In this video, Dr. Paul Barosi discusses major adverse event prediction in the L.A. Registry and how it is interconnected with risk scores and modeling. He explains the process of risk modeling, which is different from observational studies, and emphasizes the importance of careful consideration and collaboration in creating a risk model. Dr. Barosi presents the variables included in the risk model, such as age, sex, medical history, and physical exam findings. He also discusses the challenges of risk modeling, including the issue of multicollinearity and the need for model discrimination and calibration. The performance of the risk model is evaluated using the C statistic, calibration plots, and other metrics. The model shows less than ideal discrimination and calibration, indicating room for improvement. Dr. Barosi highlights that risk modeling is a complex task that requires expertise from various stakeholders, and acknowledges the contributions of the team, particularly Jim Freeman, Emily Ong, and Sarah Zimmerman. He concludes by addressing questions from the audience regarding variables, adverse events, device selection, and hospital volume. Overall, the video provides insights into the process and challenges of developing a risk model for the L.A. Registry.

Keywords

major adverse event prediction

L.A. Registry

risk modeling

variables

multicollinearity

model discrimination

calibration

risk model performance