ANÁLISIS COMPARATIVO DE MÉTODOS ESTADÍSTICOS PARA LA PREDICCIÓN DEL RIESGO DE SÍNDROME DE DOWN
Keywords:
Down Syndrome, Maternal age, Human chorionic gonadotrophin (HCG), Alpha-fetoprotein (AFP), Logistic Regression and Discriminant AnalysisAbstract
The discovery of an association between Down’s syndrome (SD) pregnancies and lower levels of maternal serum of alfa_fetoproteína (AFP), (Merkatz et al., 1984), and later, in 1987 when Bogart has shown that elevated maternal serum human chorionic gonadotrophin (HCG) levers are more effective than low AFP levels for the detection of Down’s syndrome. Carry out to a different method was study and formulated to predict the Down’s syndrome. At that time the AFP and HCG results were combined
with maternal age. These are three elements usually analyzed by a great number of authors to detecting early this chromosome abnormality. In this paper we will show the results of three statistical methods (Crossley’s method, Logistic Regression and Logistic Discriminant Analysis). We have a sample of 25 Down’s syndrome and 25635 controls. We will show the superiority of the Logistic
Regression method with a higher detection rate (88%), as well as to their simple implementation.
