Activity Number:
|
123
- Binary and Ordinal Outcome Regression
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Biometrics Section
|
Abstract #327242
|
|
Title:
|
Methods for Estimating Points Based Risk Score for Binary Clinical Outcome
|
Author(s):
|
Alok Dwivedi* and Muditha Perera and Durgesh Kumar Dwivedi and Anit Parihar and Sada Nand Dwivedi and Rakesh Shukla
|
Companies:
|
Texas Tech University Health Sciences Center El Paso and Texas Tech University Health Sciences Center El Paso and University of Texas Southwestern Medical Center and King George's Medical University and All India Institute of Medical Sciences and University of Cincinnati
|
Keywords:
|
Risk scores ;
Prediction models;
Points based score;
Multivariable analysis;
Receiver operating characteristics ;
Generalized additive model
|
Abstract:
|
Clinical risk predictive tools are often used in evidence based decision making for screening, diagnosis, and assessing patient risk for an adverse outcome. The points based risk scoring methods facilitate a rapid assessment of patient risk without complex calculations using computers or electronic devices. Several approaches have been proposed to estimate a points based risk score associated with a multivariable predictive model. However, these approaches use subjective categorizations of risk factors and do not reduce complexity for patients to evaluate their own risk. In absence of optimal thresholds for risk factors, we have proposed simple points based risk scoring methods for computing risk of an adverse outcome which can be used for shared decision making and developing decision aids. The proposed methods exploit receiver operating characteristics curve and generalized additive model analyses for determining multiple categories of risk factors and use appropriate multivariable models to weight the points allocated to each variable. The applications of proposed approaches are illustrated for predicting malignant vertebral lesions and prostate cancer using multiparametric MRI.
|
Authors who are presenting talks have a * after their name.