AN AGENT BASED MODEL OF ACUTE BEE PARALYSIS VIRUS AND KASHMIR BEE VIRUS CO-INFECTION WITHIN A HONEYBEE COLONY

Abstract

The study examined the transmission dynamics of Acute Bee Paralysis Virus (ABPV) and
Kashmir Bee Virus (KBV) co-infection in a colony. The study is aim at determining the degree
of co-infection between the two honeybee viral diseases “ABPV” and “KBV” using machine
learning classification methods. The model is based on the presence of the Varroa mite in a
honeybee colony which is a vector of the diseases and as well transmit to a susceptible host
(honeybee) only upon adequate contact with an infected host. The study made used of
Honeybees Dataset generated from the NetLogo simulation environment. The experiment used
10-fold cross-validation to estimate accuracy of the classification of the co-infection between
the two diseases using machine learning approach. It splits the dataset into 10 parts, train on 9
and test on 1 and repeat for all combinations of train-test splits. The result obtained from the
study shows that, the co-infection between two diseases is very likely. Further, the study shows
that Classification and Regression Tree (CART) fitted most with classification accuracy of
90.38% for more than 100 mites and 82.88% for less than 100 mites on the honeybee dataset
for co-infection between the two viruses ABPV and KBV. Thus, the more the mites infesting
the honeybees, the more the accurate the model fit.