จดตดทเหมาะสมสำหรบการวเคราะหเสนโคง Receiver Operating Characteristic (ROC) ในการพฒนาเครองมอนวตกรรมทางสขภาพ: กรณตวอยางโดยใชโปรแกรม STATA

This is the authors' abstract. We don't add key points for this paper.

Receiver Operating Characteristic (ROC) curve analysis is a method that is used quite widely in Diagnosis of disease, but for use in the development of health innovations it is rarely seen, and its possible application in this area remains unclear. This article presents the principles and indicators that are used for assessing the accuracy of diagnostic tests, and considers three optimal cut-off point methods, including - Euclidian’s index, Youden’s index and Weighted Number Needed to Misdiagnose (Weighted NNM). In addition, this work assesses the overall discriminant accuracy of the tests, for their potential application in health innovation - assessing each method individually and comparatively, using the area below the ROC curve (AUC). Examples were demonstrated using commands in STATA software programme. The results and related statistics were presented and interpreted as well. Finally, the implications of using ROC analysis were identified, which cover both the advantages and limitations of the ROC analysis.