|
Activity Number:
|
185
- Novel Methods for Clinical Trial Design and Characterizing Heterogeneity
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #313446
|
|
|
Title:
|
Estimate Health Disparity That Accounts for Interactions/Correlations Among Risk Factors
|
|
Author(s):
|
Chen-Pin Wang*
|
|
Companies:
|
UT Health Science Center San Antonio
|
|
Keywords:
|
health equity ;
health disparity ;
propensity score weighting;
doubly robust;
counterfactual;
machine learning
|
|
Abstract:
|
Health disparity is often attributed to multi-facet factors and the associated interactions and correlations. Existing statistical modeling of health disparity in coherent with the Institute of Medicine’s definition primarily focuses on assessing disparity associated with nonadjustable factors via balancing the marginal distributions of adjustable factors between groups. Unbiased estimation of disparity based on optimal balance of marginals of adjustable factors between groups is achievable under stringent assumptions. Estimates based on suboptimal balance of adjustable factors yet more realistic assumptions (about correlations between adjustable factors) may better project the disparity in practice. This talk compares inverse probability weighting and doubly robust estimates of disparity based on advanced weighting methods (CBPS and GBM) for balancing adjustable factors in combination with machine learning methods for outcome prediction under various scenarios of interactions/correlations between adjustable factors. Divergent estimates from these methods are found under skewed distributions of adjustable factors, and differential correlations of adjustable factors between groups.
|
Authors who are presenting talks have a * after their name.
