A statistical approach for discrete pattern recognition: A case study in endocrine physiopathology

摘要

A logical model is used to describe the control system of the hypothalamo-pituitary-thyroid axis in human endocrinology. The problem is equivalent to the determination of unknown parameters of the model given discrete observations. A solution is proposed which combines the logical model with a statistical one. As it appears that quantification of the data is uncertain, special emphasis is given to the modelling of the observer error process and to its effect on the determination of the unknown parameters.