Activity Number:
|
358
- Contributed Poster Presentations: Section on Statistics in Epidemiology
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, July 30, 2019 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #307102
|
|
Title:
|
On the Performance of Various Risk-Scoring Approaches
|
Author(s):
|
Yared Gurmu* and Jeong-Gun Park and Francesco Nordio and Kyungah Im and Jing Qian
|
Companies:
|
and Brigham and Women's Hospital and Brigham and Women's Hospital and Brigham and Women's Hospital and University of Massachusetts Amherst
|
Keywords:
|
risk scores ;
rounding;
missing data;
net reclassification;
C-index
|
Abstract:
|
Clinical risk scores have garnered widespread attention and rapid uptake as a tool for risk stratification in the biomedical field. This presentation will examine the statistical properties of various pragmatic approaches that are employed to further simplify implementation of risk scores in clinical practice. The three broad themes explored include: the impact of rounding regression coefficients, the portability of risk-score weights across models (i.e. weights developed from a logistic regression model applied to a Cox Proportional model) and optimal missing data approaches when a component of the risk score is not available. Using extensive simulation, the impact of each these approaches on bias, variance, and model performance metrics such as Harrell’s C-index and Net Reclassification index will be presented. Application data from the field of Cardiology will be used to illustrate the performance of these various approaches in practice.
|
Authors who are presenting talks have a * after their name.