Application of Statistical Methods in Epidemiology Studies

今年的工作涉及两个主要项目。首先，我们使用患者的甲基化和基因表达数据来开发预测前列腺癌复发风险的生物标志物面板。建议的小组比常用的临床生物标记物，如PSA和Gleason Score具有更好的性能。第二，与斯蒂芬妮·伦敦博士合作，我们开发了一组甲基化生物标记物，这些标记物与母亲怀孕期间的吸烟状况高度相关。这组生物标志物被用来计算吸烟得分，结果表明，吸烟得分比自我报告的吸烟状况更能预测出生体重。

The work this year involves two main projects. First, we used patients' methylation and gene expression data to develop panels of biomarkers that are predictive of prostate cancer recurrence risk. The proposed panel has better performance than the commonly used clinical biomarkers, such as PSA and Gleason score. Second, in collaboration with Dr. Stephanie London, we developed a panel of methylation biomarkers that are highly correlated to mother's smoking status during pregnancy. This panel of biomarkers is used to calculate a smoking score, which is showed to be more predictive of birth weight than the self-reported smoking status.