家族性病例對照資料之統計分析

Abstract

家族性病例對照資料研究方法近年來常使用於探討疾病與致病因子之關係。本論文回顧了分析家族資料的統計文獻方法：針對得病與否的反應變數，考慮了邏輯斯迴歸模型; 針對得病時間的反應變數，考慮了Cox等比風險模型。我們討論如何將建立在個別性資料上之研究方法推廣至家族性資料，並探討如何將適用於前瞻性資料的方法修改為分析病例對照資料所需做的假設與調整。此外，我們也透過模擬實驗來驗證推論過程中所需要之條件與比較參數估計之表現。

Familial case-control data are frequently used to study the relationship between disease and risk factors. In the thesis, we review literature for analyzing familial data. The logistic model is applied to model the probability of disease incidence. The Cox proportional hazards model is applied to model the age at onset of the disease. For each model, we discuss how to extend the method and model developed for individual data to familial data. In addition, we discuss the criteria and modification from prospective data to case-control data. We also propose simulation algorithms for generating case-control data and then, based on simulated data, examine parameter estimates and crucial properties of the inference procedure.