Lawriston A. Wilson II
Thesis submitted to the Faculty of the Virginia Tech in partial fulfillment of the requirements for the degree of
William D. Hohenboken, Chair
Conrad D. Heatwole
W. E. Beal
August 16, 1996
An expert system to predict the fertility of beef heifers was developed using the A.I. Toolkit KAPPA-PC 2.3³. The knowledge base was developed from scientific literature and from a beef cattle reproduction expert. The expert system computes an evaluation age, age both at the start and end of the breeding season, and weight per day of age to classify a heifer as having either a ³LOW², ³GOOD², or ³EXCELLENT² likelihood of conception. The expert system summarizes the information that is entered into the computer and creates a text file of the summary. It also gives explanations for every prediction to help identify and alleviate any problem areas that may affect a heifer¹s ability to reproduce. The program requires an IBM compatible computer installed with Windows 3.1³ or greater. From simulated data for purebred or crossbred British cattle, there was 72% prediction agreement between the expert system and the expert. From analysis of historical data of Hereford-Angus crossbred cattle, heifers categorized as ³LOW² and ³GOOD² had significantly higher observed pregnancy rates than expected for each category. There was no significant difference between observed and expected pregnancy rates for heifers in the ³EXCELLENT² category. Pregnancy rates for post-weaning and pre-breeding evaluations for the ³LOW² heifers were found to be lower from the combined ³GOOD² and ³EXCELLENT² heifers at P=.03 and P=.06 respectively. Observed successful calving rates for heifers categorized as ³LOW², ³GOOD², and ³EXCELLENT² did not differ significantly from the expected calving rates for each category.
Send Suggestions or Comments to firstname.lastname@example.org