The incorporation of artificial intelligence deep learning algorithms into clinical diagnostic aids has gained substantial interest in the past 5 years; a 2019 study reported that a deep learning algorithm outperformed dermatologists in a head-to-head classification of dermoscopic images of melanoma.
However, most diagnostic tests remain unsuitable for use due to inadequate performance in real-world, low prevalence populations, such as primary care or the general community. Therefore, these algorithms need rigorous, prospective validation among the populations who are intended to use them, to determine whether they lead to earlier detection and improve patient safety and quality of care, while minimising overinvestigation and overdiagnosis.
Only once validated should these algorithms be incorporated into smartphone apps for patients or clinical decision support for primary care health-care practitioners.
在過去的5年中，將人工智能深度學習算法整合到臨床診斷工具中引起了人們的極大興趣。 2019年的一項研究報告說，在皮膚鏡下黑色素瘤圖像的頭對頭分類中，深度學習算法的表現優於皮膚科醫生。但是，由於在現實世界，低患病人群（例如初級保健或普通社區）中性能不足，大多數診斷測試仍然不適合使用。因此，這些算法需要在打算使用它們的人群中進行嚴格的前瞻性驗證，以確定它們是否能夠導致較早發現並提高患者的安全性和護理質量，同時最大程度地減少過度調查和過度診斷。 只有將這些算法驗證一次後，才能將其整合到智能手機應用中，以為患者提供醫療服務或為初級保健從業人員提供臨床決策支持。
【MORE】