過去在研究地中海果蠅產卵量與壽命之間的關係，通常是依照總產卵量大小對果蠅進行分組，來觀察產卵數與壽命之間的相關性，本研究是以果蠅產卵量因子分成接近連續的方式，將1000隻雌性地中海果蠅產卵量由小至多排序，以累積樣本的方式來觀察產卵量對壽命的影響有什麼樣的變化。在探討時間相依共變數與存活時間之間的關係時，一般較常使用的半母數方法是以部分概似法來估計參數，前提是必須具有完整的共變數資訊及沒有測量誤差，才能夠達到準確地估計。為了減少使用部分概似法對參數的偏誤，可以使用聯合模型來做配適，以Cox比例風險模型與加速失敗模型的風險函數作為架構，主要目的是比較部分概似法與聯合模型對共變數與時間有關之下，其參數估計值的變化與影響，以動態的曲線來探討總產卵量的多寡是否影響參數的估計以及變數與存活時間之間的關係，並尋找果蠅的樣本會造成每日產卵量對存活的效應在Cox模型和AFT模型中有矛盾的結論。;In the literature, Mediterranean fruit fly is usually classified into a few discrete groups based on the total numbers of eggs laid to observe the relationship of the numbers of eggs and the life expectancy. In our thesis, we adopt semiparametric method for approximately continuous groups according to the total eggs laid, We are interested in the change of impact of daily egg laying when sample size increasing. The semiparametric methods applied including Cox partial likelihood method and full likelihood joint model method. For partial likelihood approach, the complete covariate information and zero measuring error are required, so that we can estimate accurately. If we consider the time-dependent covariates measured with random errors, joint model then is employed with survival described by Cox proportional hazards model or accelerated failure time model. Dynamic curve of regression coefficients as a function of total egg-laying are plotted to identify the change of impact of daily egg-laying on survival. Moreover, from the dynamic curves, we identify specific samples may have contradict conclusions using Cox model or AFT model.
