 The paper Reviews Studies on Artificial Intelligence in Pathology, AIP, published from January 2017 to February 2022 and highlights the importance of data preparation methods for AIP's success. The paper discusses challenges in replicating high performance of AIP in clinical settings and presents strategies to enhance its clinical performance. Digital pathology is crucial for clinical-grade AIP and techniques such as data standardization and weekly supervised learning based on whole slide image, WSI, are effective ways to overcome obstacles of performance reproduction. The key to performance reproducibility lies in having representative data, adequate labeling, and consistency across multiple centers. This article was authored by Yuan Qingyang, Kai Sun, Yan Hua Gao, and others.