1Department of Scientific Computing, Florida State University, 2US Army Natick Soldier Research, Development and Engineering Center
Saturday All day, Clinch Concourse
Over the years a large sample of whole-head surface scans has been collected. These scans could provide valuable information on head-shape variation for the design of protective and therapeutic head gear. Unfortunately, hair often obscures the vault of the head, which makes it difficult to identify the vault surface with certainty. Thus, a model to predict the obscured data would be extremely valuable. Here, we examine the extent to which anthropometric variables and self-reported ancestry can predict cranial index as a first step toward developing more sophisticated models utilizing landmark coordinates, anthropometric variables, ancestry, and other information.
All data were taken from the United States Army’s Anthropometric Survey II for Hispanic and White males (N=61/ancestry). Linear models were developed to predict cranial index based on hand length and horizontal foot breadth. These variables were selected as the best predictors of cranial index from seven candidate measurements via a backward stepwise algorithm using AIC as a criterion. This model was compared to a comparable linear model that included ancestry.
There was no significant difference in cranial index between ancestries (p=0.1199), and, although both models had significant fits (p<0.005 for both), the addition of ancestry did not significantly improve the fit (p=0.529) and yielded a higher leave-one-out RMSE (3.795517 versus 3.821896). Although linear models based on anthropometric data can predict cranial index, ancestry data does not always improve the predictive power of such models, and therefore, ancestry may not play a large role in future models.
This work was funded, in part, by Cooperative Research Agreement W911QY-12-2-0004 between Florida State University and the U.S. Army Natick Soldier Research, Development, and Engineering Center