With the exponential expansion of medical databases, the number of records in a single file has reached hundreds of millions, and the number of variables has increased significantly (such as the MIMIC database and the 4,484 variables in the ELSA survey). Traditional SQL queries and distributed storage systems are difficult to meet the needs of ordinary researchers due to their high technical barriers and high maintenance costs. This study proposes an innovative "slicing + dictionary" data processing strategy: by designing a multi-level slicing scheme for clinical dimensions, event dimensions, and mixed dimensions, massive datasets are decomposed into manageable subsets; at the same time, a standardized dictionary system with an encoding-description-location-attribute structure is constructed to achieve precise mapping between clinical concepts and data locations. Experimental verification shows that this method significantly reduces computational resource consumption, improves query efficiency, and supports flexible customization, enabling ordinary computing devices to load and process large-scale data on demand, which opens up new avenues for the popularization of medical big data utilization.