近期，《柳叶刀数字健康》杂志发表了一项有趣的研究。该研究中，研究人员开发了一种名为FAHR-Face（健康识别人工智能基础）的人工智能模型。该模型使用超过4000万张面部图像进行训练，并针对两项关键任务进行了调整。这两个模型分别捕捉不同类型的信息：
首个人工智能模型通过分析面部特征来估算人的生理年龄。这可能与实际年龄有所不同，但有助于了解其整体健康状况和身体抵抗力。
第二个模型根据面部照片估算癌症患者的生存风险。眼下和鼻梁区域似乎对做出这些预测尤为重要。
结果的可靠性如何?
总体而言，研究人员获准使用约 43,000 名癌症患者的数据评估该人工智能模型。随后，他们对其中3820名患者进行了另一项测试。该模型将其中122名患者归类为高风险组。在这些患者中，52%在90天内死亡。
该模型将975名患者归类为最低风险组。同期，这些患者中仅有2%死亡。
意义
这种方法的一大优势在于其技术和经济门槛低。任何人都能在几秒钟内拍摄一张简单的面部照片。
然而，目前尚未确定明确的因果关系。人工智能可以识别统计模式，但无法自动解释导致这些模式的生物学过程。此外，这些发现是否适用于其他疾病患者也尚不明确。
结论
人工智能似乎能够从一张面部照片中提取与生物年龄和癌症患者死亡风险相关的信息。研究表明，即使考虑了临床预后因素，这种关联依然存在。然而，与现有的临床方法相比，面部照片究竟能提供多少额外的预测价值，目前尚无定论。
任何想尝试使用面部人工智能估算自身生理年龄的人都可以在此网站注册并参与这项“公开研究”。用户可以上传3到10张面部照片，可以选择一次性上传或每日上传。系统会为每张照片提供一个个人生理年龄评分，以及一个总分。隐私受到保护：照片不会被共享，并在一定时间后自动删除。其他信息将被匿名化，并提供给其他研究人员。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Recently, an interesting study was published in the journal " Digital Health of the Lancet ". In this study, researchers developed an artificial intelligence model called FAHR-Face (the basis of artificial intelligence for health recognition). The model was trained using over 40 million facial images and adapted to two key tasks. Each model captures different types of information:
The first personal intelligence model estimates a person ' s biological age by analysing facial characteristics. This may differ from the actual age, but it helps to understand their overall state of health and physical resistance.
The second model estimates the survival risks of cancer patients based on facial photographs. The immediate and nostrils appear to be particularly important for these projections.
How reliable are the results?
Overall, researchers were allowed to use data on approximately 43,000 cancer patients to assess the artificial intelligence model. Subsequently, they carried out another test for 3,820 of them. The model classifies 122 of these patients as high-risk groups. Of these patients, 52 per cent died within 90 days.
The model classified 975 patients as the lowest risk group. Only 2 per cent of these patients died during the same period.
Meaning
A major advantage of this approach is its low technical and economic threshold. Anyone can take a simple facial photograph in a few seconds.
However, no clear causal link has yet been established. Artificial intelligence can identify statistical patterns, but it cannot automatically explain the biological processes that lead to them. In addition, it is not clear whether these findings apply to patients with other diseases.
Conclusions
Artificial intelligence appears to be able to extract information related to biological age and the risk of death of cancer patients from a facial photograph. Studies have shown that this association exists even when clinical prognosis is taken into account. However, the extent to which facial photographs can provide additional predictive value compared to existing clinical methods is still uncertain.
Anyone who attempts to estimate his/her biological age using facial artificial intelligence can register on this site and participate in this “public research”. Users can upload 3 to 10 facial photographs, with the option of uploading them once or daily. The system provides an individual biological age rating for each photograph, as well as a total score. Privacy is protected: photographs are not shared and are automatically deleted after a certain period of time. Other information will be anonymous and made available to other researchers.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.