Controlling for population variances in health and exposure risk using randomized matrix based mathematical modeling

  • Authors:

  • Brian M. Gurbaxani;Troy D. Querec;Elizabeth R. Unger

  • Affiliations:

  • Chronic Viral Diseases Branch, Division of High Consequence Pathogens and Pathology, Centers for Disease Control and Prevention, Atlanta, GA;Chronic Viral Diseases Branch, Division of High Consequence Pathogens and Pathology, Centers for Disease Control and Prevention, Atlanta, GA;Chronic Viral Diseases Branch, Division of High Consequence Pathogens and Pathology, Centers for Disease Control and Prevention, Atlanta, GA

  • Venue:
  • SBP'13 Proceedings of the 6th international conference on Social Computing, Behavioral-Cultural Modeling and Prediction
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

In a previous work, we analyzed the co-occurrence of HPV types in 6 large studies with cervicovaginal samples, representing 32,000 women, to ascertain if associations exist among HPV types and to guide policies on HPV vaccination and vaccine development. The data showed that more women either were uninfected by HPV or had multiple concurrent infections than could be explained by independent assortment, which could result from variance in health and exposure risk factors. Modeling exposure and immune competence proved unstable, so we used a randomized matrix based approach that obviated the need to understand the underlying risk factors. We randomized our source data while preserving increasing levels of fidelity to the original data structures to discover the type associations for HPV infection. We offer that this could be a generally useful technique for studying any type of association in biosocial science, e.g. between demographic, socioeconomic, or other variables.