Even after rigorous experimental design and implementation, data missing in clinical studies cannot be avoided. The absence of data may lead to adverse effects such as reduced test efficacy, increased complexity of data analysis, and bias. Making reasonable assumptions about the missing mechanism of data is an important prerequisite for choosing processing methods. Little and Rubin divided the missing data into missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR) according to the relationship with known and unknown variables. The real deletion tends to be a mixture of 2 or more of the above mechanisms. Researchers can fill in the data whose main missing mechanism is MAR, that is, the missing variables are not related to their own unobserved values, but are related to other known variables.