We investigate the potential of artificial neural networks in diagnosing thyroid diseases. The robustness of neural networks with regard to sampling variations is examined using a cross‐validation method. We illustrate the link between neural networks and traditional Bayesian classifiers....