Commonly used statistical analysis methods often require that the data be independent, that is, two pieces of data will not affect each other, but there are often non-independent data in clinical studies, such as multiple routine blood measurement records during hospitalization, two hospitalization records of one child, and several records before and after may be correlated. If the traditional method is used, only one record of one case can be kept, which will lose more information; Selective deletion of a record may cause bias. In order to avoid data loss and bias, several commonly used analysis methods for non-independent data are described below.