Prognosis of the sexually-precocious girl's luteinizing hormone peak value with the neural network and ultrasonic

Abstract

It aims at technologically forecasting the serum luteinizing hormone(LH) peak value by means of the artificial neural network combined with the ultrasound in the examination of exciting the gonadotropin releasing hormone(GnRH). In the process, 71 girls of the sexual precocity are selected to take the conventional ultrasonic testing on the uterus and ovary. And then, the uterus size, the ovary size and the inner diameter of the biggest ovarian follicle in the 61 of those selected girls are set to be the input variable while the LH peak value the output variable. And BP neural network is in formation, and another 10 girls are used as testing targets. As a result, the linear regression is used as a method to calculate the real value and the BP network forecasting value, showing that the correlation coefficient of the linear regression is 0.9485 and the slope is 0.9280. In conclusion, the LH peak value in the examination of GnRH can be predicted by using the ultrasound combined with the BP neural network.