Abstract: Due to the trend of rapid growth of birth defects in China， it is impractical to study all defect cases occurring in all cities. To balance the cost and the efficiency of research， it is essential to pick up the most important cities and the most typical birth defects before further investigations. A method called BDRank is proposed， to rank both birth defects and cities in which birth defects occur. Unlike PageRank， BDRank deals with multi-type systems which are composed of more than one type of objects. Further more， BDRank also takes advantage of knowledge from prior work of researchers. Experiments are conducted on real birth defect data in China from 1987 to 1989. The experimental results show an improvement in ranking accuracy over the naive method used by some birth defect researchers.

Key words: link structure, birth defect, birth defect monitoring system, undirected graph

摘要： 因为中国出生缺陷发生呈快速增长的趋势，研究人员没有条件研究发生在所有城市的所有出生缺陷案例。为了平衡研究的成本与效率，有必要选择“重要”的城市 和“重要”的出生缺陷作为研究对象。介绍了BDRank，用于对出生缺陷和出生缺陷发生的城市排序。与PageRank不同，BDRank能够处理由多种对象组成的系统。更重要的是，BDRank能够有效使用研究人员的先验知识。实验使用了中国1987-1989年的出生缺陷数据，实验表明：和出生缺陷研究人员使用的现有方法相比，BDRank具有更好的排序精度。

关键词: 连接结构, 出生缺陷, 出生缺陷监测系统, 无向图