Hypothyroidism is a pathology that occurs in the system regulating the thyroid hormones and it is a consequence of insufficiency either in the production or in the action of these hormones. Most of the patients in hypothyroidism state must use hormonal replacement therapy for all their lives. In this work, a controller model for supplying therapeutic drugs to primary hypothyroidism patients, whose have been submitted to total removal of thyroid gland, is developed. Two multilayer perceptron neural networks predict the concentrations of the hormones TSH and T4 in the blood. These forecasts and the current concentrations are used as input to the controller. Based on a set of production rules, the controller defines the necessary drug dosage to maintain the concentrations inside regularity bands in the future. The performance of the controller must be such that abrupt changes do not occur on the time evolution of those hormones.
thyroid hormone; neural network; production rules
