毒物研究中每日可服劑量(Allowable Daily Intakes；簡稱ADI)之決定是一項重要的課題，因為服用過多的劑量將會有毒性反應產生。本文研究如何就多組符合Cox比例風險模式之右設限資料，根據機率衡量及限制平均壽命衡量建立額外風險，並且在一定的安全閥值下，估計毒物標竿劑量及其信賴下限，做為ADI的估計值。之後，藉由模擬研究探討額外風險要求下機率衡量與限制平均壽命衡量建構的標竿劑量信賴下界之覆蓋機率及其與真正標竿劑量的偏誤。最後，以一實例說明本文所提方法之應用。 In toxicity study, how to determine ADI (Allowable Daily Intakes) is an important issue because of taking the dose over acceptable region may cause an abnormal adverse. According to safe threshold values demand, we use two statistical methods to estimate ADI - one is probability measurement, and the other is restricted mean life time measurement under Cox proportional hazard model for right-censored survival data. We further conduct a simulation study to investigate the coverage probability of the lower confidence limit for benchmark dose (BMD) and bias under these two measurements. Finally, the use of those procedures is illustrated with a right-censored survival data
