Qiang Jiaqi, Hong Weixin, Sun Yuxin, Lyu Xiaohong, Pan Zhouxian, Wu Danning, Zhou Zhibo, Guo Xiaoyuan, Du Hanze, Yang Hongbo, Zhu Huijuan, Chen Shi, Pan Hui, Shen Zhen
Key Laboratory of Endocrinology of National Health Commission, Department of Endocrinology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100730, China.
State Key Laboratory of Multimodal Artificial Intelligence Systems, Beijing Engineering Research Center of Intelligent Systems and Technology, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China; School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China.
Int J Med Inform. 2025 Nov;203:105976. doi: 10.1016/j.ijmedinf.2025.105976. Epub 2025 May 20.
Artificial intelligence-based facial recognition (AI-FR) is promising in diagnosis of diseases with distinct facial features. Our team has retrospectively constructed an AI-FR system for Turner Syndrome (TS) based on 1295 facial photographs in previous research. This study aims to evaluate this AI-FR system for TS screening in a prospective cohort in real-world clinic setting. We also aim to elucidate the impact of complexity of facial features on diagnostic accuracy of AI-FR in this cohort.
Patients were recruited in a single-gate prospective cohort at a clinic. Facial images were collected for AI-FR diagnosis. Karyotype analyses were performed as the gold-standard diagnosis. Diagnostic performance of the AI-FR system was evaluated. Individual facial recognition intensity (iFRI) was proposed to characterize the complexity of individual facial features in patients. iFRI was calculated based on 26 facial landmarks of seven parts. Group comparison and regression analysis were performed on iFRI according to diagnosis classification.
The prospective cohort included 218 patients (29 TS and 189 control). The AI-FR system showed performance: 89.0 % accuracy (95 % CI 84.1-92.5), 72.4 % sensitivity (95 % CI 54.3-85.3), and 91.5 % specificity (95 % CI 86.7-94.7). TS vs. control diagnosed by AI-FR had higher iFRI (p = 0.009), while TS vs. control diagnosed by gold standard showed no iFRI difference (p > 0.05). AI diagnosis classification positively correlated with iFRI (p = 0.033), while gold-standard classification did not (p > 0.05).
This was one of the pioneering prospective cohorts in disease diagnosis with AI-FR. This system for TS screening achieved ideal performance and improved the diagnosis of TS. iFRI was proposed and proved influential to AI-FR diagnostic performance.
基于人工智能的面部识别（AI-FR）在诊断具有明显面部特征的疾病方面具有前景。我们的团队在先前的研究中基于1295张面部照片回顾性构建了一个用于特纳综合征（TS）的AI-FR系统。本研究旨在在真实临床环境中的前瞻性队列中评估该AI-FR系统用于TS筛查的效果。我们还旨在阐明面部特征复杂性对该队列中AI-FR诊断准确性的影响。
在一家诊所的单门前瞻性队列中招募患者。收集面部图像用于AI-FR诊断。进行核型分析作为金标准诊断。评估AI-FR系统的诊断性能。提出个体面部识别强度（iFRI）来表征患者个体面部特征的复杂性。iFRI基于七个部位的26个面部标志点进行计算。根据诊断分类对iFRI进行组间比较和回归分析。
前瞻性队列包括218名患者（29名TS患者和189名对照）。AI-FR系统表现出如下性能：准确率89.0%（95%CI 84.1-92.5），灵敏度72.4%（95%CI 54.3-85.3），特异性91.5%（95%CI 86.7-94.7）。经AI-FR诊断的TS患者与对照相比具有更高的iFRI（p = 0.009），而经金标准诊断的TS患者与对照相比iFRI无差异（p>0.05）。AI诊断分类与iFRI呈正相关（p = 0.033），而金标准分类则不然（p>0.05）。
这是AI-FR用于疾病诊断的首批前瞻性队列研究之一。该TS筛查系统取得了理想的性能并改善了TS的诊断。提出了iFRI并证明其对AI-FR诊断性能有影响。
