Deep learning-based relapse prediction of neuromyelitis optica spectrum disorder with anti-aquaporin-4 antibody

Liang Wang; Lei Du; Qinying Li; Fang Li; Bei Wang; Yuanqi Zhao; Qiang Meng; Wenyu Li; Juyuan Pan; Junhui Xia; Shitao Wu; Jie Yang; Heng Li; Jianhua Ma; Jingzi ZhangBao; Wenjuan Huang; Xuechun Chang; Hongmei Tan; Jian Yu; Lei Zhou; Chuanzhen Lu; Min Wang; Qiang Dong; Jiahong Lu; Chongbo Zhao; Chao Quan

doi:10.3389/fneur.2022.947974

Deep learning-based relapse prediction of neuromyelitis optica spectrum disorder with anti-aquaporin-4 antibody

Front Neurol. 2022 Aug 5:13:947974. doi: 10.3389/fneur.2022.947974. eCollection 2022.

Authors

Liang Wang^{1

2}, Lei Du³, Qinying Li⁴, Fang Li^{2

5}, Bei Wang⁶, Yuanqi Zhao⁷, Qiang Meng⁸, Wenyu Li⁹, Juyuan Pan¹⁰, Junhui Xia¹⁰, Shitao Wu¹¹, Jie Yang¹², Heng Li¹³, Jianhua Ma³, Jingzi ZhangBao^{1

2}, Wenjuan Huang^{1

2}, Xuechun Chang^{1

2}, Hongmei Tan^{1

2}, Jian Yu¹⁴, Lei Zhou^{1

2}, Chuanzhen Lu^{1

2}, Min Wang¹⁴, Qiang Dong^{1

2}, Jiahong Lu^{1

2}, Chongbo Zhao^{1

2}, Chao Quan^{1

2}

Affiliations

¹ Department of Neurology, Huashan Rare Disease Center, Huashan Hospital, Shanghai Medical College, Fudan University, Shanghai, China.
² National Center for Neurological Disorders (NCND), Shanghai, China.
³ Department of Neurology, The First Affiliated Hospital of Xinjiang Medical University, Urumqi, Xinjiang Uygur Autonomous Region, China.
⁴ Department of Rehabilitation Medicine, Jing'an District Centre Hospital of Shanghai, Fudan University, Shanghai, China.
⁵ Department of Rehabilitation Medicine, Huashan Hospital, Shanghai Medical College, Fudan University, Shanghai, China.
⁶ Department of Neurology, Jing'an District Centre Hospital of Shanghai, Fudan University, Shanghai, China.
⁷ Department of Neurology, The Second Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China.
⁸ Department of Neurology, The First People's Hospital of Yunnan Province, Kunming, China.
⁹ Department of Neurology, Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University, Hangzhou, China.
¹⁰ Department of Neurology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China.
¹¹ Department of Neurology, The Fifth Affiliated Hospital of Zhengzhou University, Zhengzhou, China.
¹² Department of Neurology, Wuhan No.1 Hospital, Wuhan, China.
¹³ Department of Neurology, Central Hospital Affiliated to Shandong First Medical University, Jinan, China.
¹⁴ Department of Ophthalmology and Vision Science, Eye and ENT Hospital, Shanghai Medical College, Fudan University, Shanghai, China.

Abstract

Objective: We previously identified the independent predictors of recurrent relapse in neuromyelitis optica spectrum disorder (NMOSD) with anti-aquaporin-4 antibody (AQP4-ab) and designed a nomogram to estimate the 1- and 2-year relapse-free probability, using the Cox proportional hazard (Cox-PH) model, assuming that the risk of relapse had a linear correlation with clinical variables. However, whether the linear assumption fits real disease tragedy is unknown. We aimed to employ deep learning and machine learning to develop a novel prediction model of relapse in patients with NMOSD and compare the performance with the conventional Cox-PH model.

Methods: This retrospective cohort study included patients with NMOSD with AQP4-ab in 10 study centers. In this study, 1,135 treatment episodes from 358 patients in Huashan Hospital were employed as the training set while 213 treatment episodes from 92 patients in nine other research centers as the validation set. We compared five models with added variables of gender, AQP4-ab titer, previous attack under the same therapy, EDSS score at treatment initiation, maintenance therapy, age at treatment initiation, disease duration, the phenotype of the most recent attack, and annualized relapse rate (ARR) of the most recent year by concordance index (C-index): conventional Cox-PH, random survival forest (RSF), LogisticHazard, DeepHit, and DeepSurv.

Results: When including all variables, RSF outperformed the C-index in the training set (0.739), followed by DeepHit (0.737), LogisticHazard (0.722), DeepSurv (0.698), and Cox-PH (0.679) models. As for the validation set, the C-index of LogisticHazard outperformed the other models (0.718), followed by DeepHit (0.704), DeepSurv (0.698), RSF (0.685), and Cox-PH (0.651) models. Maintenance therapy was calculated to be the most important variable for relapse prediction.

Conclusion: This study confirmed the superiority of deep learning to design a prediction model of relapse in patients with AQP4-ab-positive NMOSD, with the LogisticHazard model showing the best predictive power in validation.

Keywords: anti-aquaporin-4 antibody; deep learning; machine learning; neuromyelitis optica spectrum disorder; relapse prediction.