Abstract # 6238 Poster # 83:

Scheduled for Thursday, June 18, 2015 06:00 PM-08:00 PM: (Cascade AJBCD) Poster Presentation


MANY NEEDLES IN THE HAYSTACK: GETTING MORE FROM RECORD SYSTEMS AND COMPUTING TECHNOLOGY

D. Lawrence1,2, S. Kohama1 and B. Park3
1Oregon National Primate Research Center, Oregon Health & Science University, 505 NW 185th Avenue, Beaverton, Oregon 97006-3448, USA, 2Portland State University, 3Oregon Health & Science University
line
     We demonstrate a flexible and easy-to-use system for mining large data resources to define and evaluate predictive models. Developed in Mathematica, the application allows users to target specific data resources, of virtually any kind, for statistical analysis and computation of fitted models. To refine and demonstrate it, we queried decades of data from our institutional record system, specifically filtering for brain weight in rhesus macaques. We sought to test reports that adult weights are stable in maturity, where our larger dataset, notably covering advanced age, would provide additional statistical power; we also endeavored to establish age of brain maturity. After filtering outliers and underrepresented animals (Chinese and Indian-Chinese hybrids), piecewise regression on our final sample of 2093 males and 2713 females revealed that weight increased rapidly in infants (0.1942 g/day males to 106 days, 0.3161 g/day females to 61 days), slowed in juveniles (0.0122 g/day males 106 to 1690 days, 0.0112 g/day females 61 to 1342 days), was stable in adult females (1342 days and older), yet declined significantly (p < 0.05) in adult males (-0.00043 g/day 1690 days and older.) Hence this application’s ability to conveniently query large datasets indeed provided additional power to discern finer levels of sex-dependent change with age. Additional benefits include both flexibility to mine data across institutions and varied record systems, and portability to facilitate application sharing and collaborations.