Length of stay of an inpatient reflects the severity of illness as well as the practice patterns of the hospital. Predicting the length
of stay will provide a better perception of the different resources consumed in a healthcare system. Neural network trained using
back propagation has been discerned as a successful prediction model in healthcare systems 1
In this paper, a robust stochastic
optimization technique called Particle Swarm Optimization (PSO) is compared with back propagation for training. The algorithms were evaluated based on error convergence, sensitivity, specificity, positive precision value and accuracy and corresponding results are presented.