Probit regression in prediction analysis

Abstract

To avoid diagnostic surgery, the probability of nodal involvement of prostate cancer is modeled using the pobit link function to determine whether the lymph nodes of a patients are infected. X-ray status and level of acid phosphatase in the blood serum of patients are considered as preoperative explanatory variables with the number of patients having nodal involvement as response variable. Within the framework of the probit regression model, the level of nodal involvement is predicted and the probability of nodal involvement obtained.

KEYWORDS: Probit model, Nodal involvement, Standard cumulative normal distribution, Latent variable, Logistic regression.