Predicting Hospital Length of Stay with Neural Networks

Critical care providers are faced with resource shortages and must find ways to effectively plan their resource utilization. Neural networks provide a new method for evaluating trauma patient (and other medical patient) level of illness and accurately predicting a patient’s length of stay at the critical care facility. Backpropagation, radial-basis-function, and fuzzy ARTMAP neural networks are implemented to determine the applicability of neural networks for predicting either injury severity or length of stay (or both). Neural networks perform well on this medical domain problem. The backpropagation networks achieved the best performance for predicting a patient’s length of stay, but the fuzzy ARTMAP produced superior performance in evaluating patient’s level of injury (especially for the more severely injured patients). Thus a combination of backpropagation and fuzzy ARTMAP neural networks is recommended to produce the optimal combined (injury severity and length of stay) results.