在毒物學的研究中，每日可服劑量( allowable daily intakes；簡稱ADI)的決定是一個重要的課題，因為每日服用毒物大於此一安全劑量，則易於產生毒性反應。過去文獻針對在對數邏輯斯分布、對數常態分布或韋伯分布下的加速失敗時間模型，在最高可容忍毒性反應的標竿風險(benchmark risk；簡稱BMR)之下，估計標竿劑量(benchmark dose；簡稱BMD)，並且求出標竿劑量的下界(BMDL)作為ADI的參考劑量。本文考慮在具有廣義伽瑪分布誤差的加速失敗時間模型之下估計BMDL，並且模擬研究其覆蓋機率。 In toxicity study, how to determine the allowable daily intakes(ADI) is an important issue because of taking the toxicant over acceptable region may cause an abnormal reaction. In past study, we estimate the benchmark dose(BMD) and its lower bound(BMDL) for ADI reference based on benchmark risk(BMR) by the allowable toxic effect when the data are suitable for the accelerated failure time model with Log-logistic,Log-normal, and Weibull distribution. In this study, we estimate the BMDL under accelerated failure time model with error term which is distributed to a generalized gamma distribution. We also conduct a simulation study to investigate the coverage probability of the BMDL proposed.
