Skip to main content

Table 1 Summary of Results for MDR Models Fitting (a) Three separate genotypes for all polymorphisms and (b) Using the dichotomous grouping for the polymorphisms suggested by the literature (ACE DD vs. not DD, PAI-1 presence of 4G allele vs. 5G5G, tPA II vs. not II).

From: An application of conditional logistic regression and multifactor dimensionality reduction for detecting gene-gene Interactions on risk of myocardial infarction: The importance of model validation

# of Loci Loci Included in Best Model Cross-Validation Consistency Prediction Error
Fitting Three Separate Genotypes for All Polymorphisms:
2 ACE, PAI-1 86% 51%
3 ACE, PAI-1, t-PA 100% 51%
Using Dichotomous Groupings:
2 ACE, PAI-1 100% 46%
3 ACE, PAI-1, t-PA 100% 46%