As a prospective and dynamic research paradigm, adaptive design allows sample size, endpoints, or recruitment strategies to be modified based on predefined rules during the course of a clinical trial, thereby improving power and ethical soundness while controlling the type I errors. However, the estimation of initial sample size, which is the core of the paradigm, still faces challenges such as methodological complexity, high dependence on specialized software, and limited localized application. Therefore, this paper systematically reviews the methodological progress in initial sample size estimation for adaptive designs, covering various variable types including continuous, binary, survival, and ordinal data. We summarize key parameters of commonly used strategies, particularly group sequential designs, such as alpha spending functions, information time, and sample size inflation factors, and compare the implementation approaches and applicability in mainstream statistical software, including PASS, SAS, Stata and R. This paper briefly describes the initial sample size estimation methods used for different variable types in adaptive designs so as to provide references for clinical research designers and implementers.