联合应用CART及BP-ANN法建立中国西部汉族女性乳腺癌环境-基因风险评估模型的研究

建立乳腺癌风险评估模型，筛选高危人群，再开展有针对性的行为干预及乳腺筛查的策略被认为具有较高的卫生经济学效益。目前国内开展的风险评估模型研究存在以下局限性：乳腺癌危险因素危险度量化不准确（研究样本代表性不足及因素危险度结论不一致），遗传易感性与环境因素的交互作用对乳腺癌发病风险的贡献尚不明确，以及所用的建模方法难以解决复杂疾病的多病因关联问题。本研究将开展大样本的病例对照研究（病例对照800对)，收集研究对象环境危险因素（生殖生育、膳食、不良行为方式等）信息，检测基于流行病学研究及GWAS证据的乳腺癌易感基因多个SNPs。采用分类与回归树（CART）法探索基因-基因，环境-基因交互作用；应用误差逆传播人工神经网络（BP-ANN）模型建立基于环境危险因素的汉族女性乳腺癌患病风险评估的基础模型；并初步探索建立环境-基因风险评估的校正模型。研究结果为今后建立中国汉族女性乳腺癌发病预测模型提供依据

Behavior intervention and screening in the high-risk population of breast cancer selected by Risk Evaluation Model has been recognized as cost-effective and reasonable. In China, the studies on risk evaluation of breast cancer have limitations as follow: Firstly, the attribute risk of risk factors for breast cancer has not been estimated accurately for lack of represent studies and the inconsistent effect evaluation values of risk factors.Secondly,the contributions of environment-gene or gene-gene interactions to breast cancer have not been definite.Thirdly,the classical logistic regression model may not appropriate to explore the complex associations of multiple risk factors for breast cancer.For above reasons,we will implement a case-control study with large sample (800 pairs) to survey the environmental risk factors for breast cancer (e.g. reproductive factors, diet, unsatisfied behaviors, etc.) and examine Single Nucleotide Polymorphisms locating at susceptible genes of breast cancer, which were selected by epidemiological studies and genome-wide association study. Classification and Regression Tree (CART) will be used to explore gene-gene and gene-environment interactions; back propagation artificial neural network (BP-ANN) will be employed to establish breast cancer risk evaluation model for Han Nationality women based on environmental risk factors (basic model) and on environment-gene factors (adjusted model). The results of this study will provided a reliable basis for developing breast cancer prediction model for Han Nationality women.