Public Health Sciences Division

Risk Prediction Symposium 2014

M1-A305/307, Wednesday, June 11, 2014

Predicting an individual's risk of an outcome is a goal in many areas of medicine today. Examples include future onset of cancer or cardiovascular events in healthy individuals, morbidity/mortality events in patients diagnosed with disease, and response to treatment in patents following treatment for disease. Risk predictions are sought to guide the use of treatment or preventative interventions, to identify high risk subjects who can be recruited into clinical trials of these interventions, or simply to inform individuals of their risk.  Risk prediction models are essentially calculators that output a risk estimate given an individual's data on factors such as demographics, clinical characteristics, genetics, and/or the results of imaging or biomarker analyses.  Given the growing interest in "personalized medicine" and the rapid development of high throughput technologies, the pursuit of risk prediction models, and of constituent risk predictors, is ever-expanding.  The clinical impact of these models is also multiplying, given their frequent dissemination over the internet.  Yet there has recently been considerable controversy in the field regarding the ideal evaluation of a risk prediction model.  

This symposium included talks illustrating the development, evaluation, and clinical use of risk prediction models; talks summarizing and interpreting the recent controversy regarding model evaluation; and discussions, led by an expert panel, of current "best practices." 

We are very grateful to our speakers for making their presentations available to the public. Click on the titles below to view or download.