Human‐multimodal deep learning collaboration in ‘precise’ diagnosis of lupus erythematosus subtypes and similar skin diseases

• DOI: 10.1111/jdv.20031
• Publication Date: 2024-04-15T14:00:16Z
• Authors: Qianwen Li; Zhi Yang; Kaili Chen; Ming Zhao; Hai Long; Yueming Deng; Haoran Hu; Chen Jia; Meiyu Wu; Zhidan Zhao; Huan Zhu; Suqing Zhou; Mingming Zhao; Pengpeng Cao; Shengnan Zhou; Yang Song; Guishao Tang; Juan Liu; Jiao Jiang; Wei Liao; Wenhui Zhou; Bingyi Yang; Feng Xiong; Suhan Zhang; Xiaofei Gao; Yiqun Jiang; Wei Zhang; Bo Zhang; Yan‐ling He; Liwei Ran; Chunlei Zhang; Wenting Wu; Quzong Suolang; Hanhuan Luo; Xiaojing Kang; Caoying Wu; Hongzhong Jin; Lei Chen; Qing Guo; Guangji Gui; Shanshan Li; Henan Si; Shuping Guo; Hong‐Ye Liu; Xiguang Liu; Guo‐Zhang Ma; Danqi Deng; Limei Yuan; Jianyun Lu; Jinrong Zeng; Xian Jiang; Xiaoyan Lyu; Liuqing Chen; Bin Hu; Juan Tao; Yuhao Liu; Gang Wang; Guannan Zhu; Zhirong Yao; Qianyue Xu; Bin Yang; Yu Wang; Yan Ding; Xianxu Yang; Hu Kai; Haijing Wu; Qianjin Lu
• Keywords: Skin lesion; Teledermatology; Human skin; Clinical Diagnosis

Abstract Background Lupus erythematosus (LE) is a spectrum of autoimmune diseases. Due to the complexity cutaneous LE (CLE), clinical skin image‐based artificial intelligence still experiencing difficulties in distinguishing subtypes LE. Objectives We aim develop multimodal deep learning system (MMDLS) for human‐AI collaboration diagnosis subtypes. Methods This multi‐centre study based on 25 institutions across China assist subtypes, other eight similar diseases and healthy subjects. In total, 446 cases with 800 images, 3786 multicolor‐immunohistochemistry (multi‐IHC) images data were collected, EfficientNet‐B3 ResNet‐18 utilized this study. Results multi‐classification task, overall performance MMDLS 13 conditions much higher than single or dual modals (Sen = 0.8288, Spe 0.9852, Pre 0.8518, AUC 0.9844). Further, MMDLS‐based diagnostic‐support help improves accuracy dermatologists from 66.88% ± 6.94% 81.25% 4.23% ( p 0.0004). Conclusions These results highlight benefit human‐MMDLS collaborated framework telemedicine by assisting rheumatologists differential