In the information era, data-driven decision-making methods have become mainstream. At the same time, the long-term accumulation of data has led to the growing presence of concept drift. To address the problem of concept drift in dynamic decision environments, a data-driven evidential reasoning decision-making method is proposed. First, aiming to handle the subtle nature of drift in historical decision data, the method integrates the idea of early drift detection and applies the cumulative sum control chart method (CUSUM) to identify potential drift within the data. Based on this, a dual-layer integration strategy is then designed to adapt concept drift by using evidential reasoning aggregation algorithm. The attribute weights are learned to be applied to aggregate local information to obtain local optimal decision results. Further, the contribution scores of local data are defined to aggregate global information to generate a final decision, which balances accuracy, adaptability, and interpretability of the decision results. Finally, the proposed method has been applied to solve a real problem of breast nodule auxiliary diagnosis from a tertiary hospital in Hefei, Anhui Province, which verifies the effectiveness and applicability of the proposed method.