With the rapid development of Internet of Things(IoT), big data, and Artificial Intelligence(AI) technologies, massive amounts of data are being generated and utilized on an unprecedented scale. These data contain a large amount of sensitive information, and how to securely store sensitive data has become a realistic problem that needs to be solved. The existing data storage schemes usually focus on the direct protection of sensitive data, while ignoring the leakage risks associated with explicit and implicit associations between sensitive and non-sensitive data. The explicit and implicit relationships among data are deeply analyzed from the perspective of information entropy, and a method is proposed to quickly assess the explicit and implicit relationships and predict the leakage risk of sensitive data. By introducing the information Lift Ratio(LR) and the Probability of Information Control(PIC), the method can effectively identify the influence of non-sensitive data on the risk of sensitive data leakage. In the simulation experiments, the maximum single-attribute LR in the Statistical Property Dataset(SPD) is 0.308, and the joint-attribute LR can be up to 0.891, and the predicted value of the sensitive data leakage risk is significantly improved, up to 23.2%. The simulation results show that the method can effectively identify and cope with the security risks caused by explicit and implicit relationships, thus significantly improving the overall security level of sensitive data storage.