隨著現代醫療科技的進步，在診斷醫學中也發展新的診斷疾病的方法，讓病人得以盡早發現疾病，盡快接受治療。為檢測上述新的診斷方法相對於舊的診斷方法的準確性，需要決定招募多少受試者，所以樣本數的決定會是一項重要的步驟。如果受試者接受診斷的結果為罹病(+)或非罹病(-)的二元資料時，需要以敏感度以及特異度作為評量診斷試劑的標準。本文考量同一受試者接受兩種不同診斷方法所產生的成對資料，根據病人的敏感度與非病人的特異度評估診斷方法，針對新的診斷方法是否優越、非劣於舊的診斷方法，或新舊診斷方法具有等效性，建立適當的檢定。進一步求出上述檢定的非條件型I誤差率及檢定力，藉以計算病人與非病人之人數，之後經由疾病盛行率的調整，求出所需招募的受試者人數。最後在估計的受試者人數之下，求出比較敏感度與特異度檢定的非條件檢定力。;With the advancement of modern medical technology, new diagnostic methods have been developed in the field of medical diagnostics, enabling early detection of diseases and prompt initiation of treatments for patients. To evaluate the accuracy of these new diagnostic methods compared to the conventional ones, determining the appropriate sample size becomes a crucial step. When the diagnostic results for the participants are binary data, with positive and negative responses, sensitivity and specificity are used as the criteria to assess the diagnostic tests. In this study, we consider paired data generated from the same participants undergoing two different diagnostic methods. We evaluate the diagnostic methods based on the sensitivity for patients and specificity for non-patients. Our aim is to determine whether the new diagnostic method is superior, non-inferior, or equivalent to the conventional diagnostic method. We establish appropriate tests to calculate the type I error and power of these tests under an unconditional setting. The number of patients and that of non-patients are then estimated, and the total sample size is then obtained by taking into account the prevalence rate of the disease. Finally, under the estimated sample size, we compute the unconditional power of the test for the sensitivity and specificity.
