The Extreme Mass-Ratio Inspiral (EMRI) refers to binary system with a mass ratio between 104 and 106, where the smaller object loses energy as it inspirals closer to a massive black hole, emitting gravitational waves. It is estimated that there are 105 cycles during the last year before plunge, providing rich information on the evolution of gravitational wave phases. The motion of the smaller object in the strong gravitational field of the massive black hole can reflect the surrounding spacetime structure. The massive black hole is typically located at the center of a galaxy, in which the galaxy environment leaves traces in the EMRI waveform, thus EMRI can be used to constrain the gravitational theories, no-hair theorem and so on. Multiple sources can provide constraints on the mass and spin distribution of massive black holes, contributing to the understanding of cosmic and galactic evolution. Due to this significance, EMRI has become an important target for space gravitational wave missions such as LISA, Taiji, and TianQin. Consequently, the EMRI data analysis has become a crucial task. However, due to the high dimensionality and complexity of the waveform, relevant methods are still under discussion. This article initially reviews the framework and discussions regarding EMRI data analysis conducted during the Mock LISA Data Challenge (MLDC). It subsequently provides a comprehensive overview of the challenges encountered and the corresponding improvements proposed based on the authors’ research. Finally, the article offers some clues and suggestions regarding potential advancements in EMRI data analysis methods.