Cardiac surgery-associated Acute Kidney Injury (CSA-AKI) is one of the most common complications after cardiovascular Surgery, with an incidence rate as high as 30-50%. CSA-AKI has many hazards, including significantly increasing the hospital stay of patients after surgery and greatly increasing medical costs. Moreover, it significantly raises the short-term and long-term mortality (by 3 to 8 times) and the risk of occurrence or aggravation of chronic kidney disease (CKD) in patients after surgery. These risks increase with the aggravation of CSA-AKI. Therefore, early prevention and treatment throughout the perioperative period of CSA-AKI is crucial for improving prognosis. Early prediction, especially preoperative prediction, is the prerequisite for the early prevention and treatment of CSA-AKI throughout the course. The aim of this study is to establish a dynamic prediction model for the entire perioperative period of CSA-AKI, with the expectation of improving the prediction and diagnosis of CSA-AKI for the development and validation of a prediction model based on a multicenter retrospective cohort. Using retrospective multimodal data of perioperative cardiovascular surgeries from hospitals such as Beijing, Central China, Yunnan and Fuwai Hospital in Shenzhen, and based on artificial intelligence technologies (such as machine learning), a CSA-AKI risk prediction model suitable for the entire perioperative period dynamics of various surgical types was constructed.