In clinical and observational studies, recurrent events are frequently encountered, including repeated episodes of hospitalizations, strokes, and others. Throughout the follow-up period, these recurrent events may be censored by a semi-competing risk event (e.g., death), resulting in dependent censoring. While joint modeling of recurrent events with the terminal event has garnered considerable attention, further advancements in analytical methods are necessary to meet real-world application needs. One commonly utilized approach is joint frailty modeling with a shared frailty factor for recurrent and terminal event processes. However, there are limitations, such as the strong assumption of conditional independence, which can be easily violated in practice. Also, handling multiple types of recurrent events, such as repeated occurrences of heart failure, myocardial infarction, stroke, or other cardiovascular diseases, is crucial but has not been thoroughly explored. In this paper, we will provide an overview of existing methods and discuss recently proposed ones to address these gaps, highlighting their advantages and disadvantages. Furthermore, evaluating patient prognosis and dynamically predicting the risk of death are clinically significant, and utilizing developed models considering historical recurrent events is expected to enhance medical decisions and improve healthcare outcomes. Finally, we will discuss current methodological gaps, present an illustrative example using a large-scale dataset and propose directions for future research.