In the current paper we mathematically try to support the medical operation decision made for the sake of patients suffering from gastric cancer. We involve the linear model of the simple neural perceptron to distinguish between two decision states determined as “operate” contra “do not operate”. The perceptron input signals are proposed to be given as codes of levels of the most decisive biological markers, considered in surgery decision making. To find the level intervals, tied to the codes, we introduce a procedure of level fuzzification by a family of parametric membership functions.