| Abstract: | ABSTRACT Modeling in epidemiology has followed many different strategies and philosophies. Artificial neural networks (ANNs) comprise a family of highly flexible and adaptive models that have shown promise for application to modeling disease phenomena in general and plant disease forecasting in particular. ANN modeling requires the availability of representative, robust input data and exhaustive testing of model aptness and optimization; meanwhile, ANNs sacrifice much of the biological insight often derived through other model forms. On the other hand, ANNs may extract previously undetected and possibly complex relationships, which can increase prediction accuracy over mainstream statistical methods, usually in an incremental manner. |
