AUTHOR=Li Hao , Tao Xiang , Liang Tuo , Jiang Jie , Zhu Jichong , Wu Shaofeng , Chen Liyi , Zhang Zide , Zhou Chenxing , Sun Xuhua , Huang Shengsheng , Chen Jiarui , Chen Tianyou , Ye Zhen , Chen Wuhua , Guo Hao , Yao Yuanlin , Liao Shian , Yu Chaojie , Fan Binguang , Liu Yihong , Lu Chunai , Hu Junnan , Xie Qinghong , Wei Xiao , Fang Cairen , Liu Huijiang , Huang Chengqian , Pan Shixin , Zhan Xinli , Liu Chong 

TITLE=Comprehensive AI-assisted tool for ankylosing spondylitis based on multicenter research outperforms human experts

JOURNAL=Frontiers in Public Health

VOLUME=Volume 11 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2023.1063633

DOI=10.3389/fpubh.2023.1063633

ISSN=2296-2565

ABSTRACT=Early diagnosis and individualized treatment of ankylosing spondylitis (AS) remain extremely challenging, particularly in less-developed countries, which lack experts. To that goal, in this study, a comprehensive artificial intelligence (AI) tool was developed for AS diagnosis and clinical prediction. First, an ensemble deep learning (DL) model was established to diagnose AS by using pelvic radiographs. In addition, clinical prediction models for identifying high-risk patients and triaging patients were constructed. In a multicenter external test set, the ensemble DL model exhibited satisfactory performance, with precision, recall, and area under curve values of 0.90, 0.89, and 0.96, respectively. Its performance surpassed that of human experts. Moreover, the model aided greatly in the experts’ diagnostic performance. Furthermore, the model’s diagnosis results based on smartphone-captured images are comparable to those of human experts. In addition, a clinical prediction model that accurately categorizes patients with AS into high-and low-risk groups with distinct clinical trajectories was established. This lays the groundwork for individualized care. Overall, in this study, an exceptionally comprehensive AI tool was developed for the diagnosis and management of AS in complex clinical scenarios, particularly in underdeveloped or rural areas, which lack experts.