Response to Fitzgerald & Barenholtz: There is still much work to be done for digital classifiers.
Response to Fitzgerald & Barenholtz: There is still much work to be done for digital classifiers.
复制标题
对菲茨杰拉德的回应
DOI：
10.1111/add.15788
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Afshar,Majid
中科院分区：
文献类型：
--
作者：
Karnik,NiranjanS;Thompson,HaleM;Afshar,Majid
We thank Drs. Fitzgerald and Barenholtz for their thoughtful comments [1]. We appreciate the points that our colleagues are making about sensitivity and the importance of having accurate reference standards. There are significant challenges in developing these methods in real-world settings like major hospital systems. Ultimately, our goal here is not to demonstrate the superior performance of our classifier over other tools but instead to improve screening rates in the hospital setting. If we continue to rely on manual screening, it is our belief that large numbers of patients will continue to be missed due to resource limitations of staff and time that generally restrict how extensively these can be done. Our approach has the advantage of being a fully automated screening process. In this scenario, we would aim to show non-inferiority of our digital classifier approach to existing methods. Finally, we concur with our colleagues that a larger, multisite trial of this digital approach against current manual screening is merited and we hope to see this come to fruition in the near term.
