Expert consensus on deploying AI diagnostic models to support electroencephalogram-based epilepsy diagnosis

doi:10.3969/j.issn.1004-583X.2026.02.011

Abstract

Abstract:

This expert consensus establishes a comprehensive framework and practical roadmap for applying artificial intelligence (AI) diagnostic models to support electroencephalogram (EEG) analysis in epilepsy. In light of the current drawbacks of manual EEG interpretation, primarily subjectivity and limited efficiency, the consensus frames AI as a clinical decision-support tool that augments, rather than replaces, physician expertise. It issues standardized, full-lifecycle recommendations spanning data acquisition and annotation, algorithm development and validation, clinical integration and workflow implementation, and governance covering safety, ethical considerations, and regulatory oversight. The objective is to enable the scientific, safe, and effective incorporation of AI into epilepsy care, thereby enhancing diagnostic accuracy, efficiency, and inter-rater consistency, and to provide practical guidance for researchers, clinicians, developers, and regulators.

Key words: epilepsy, electroencephalogram, artificial intelligence, assisted diagnosis, expert consensus

CLC Number:

R742.1

Task Force on the Expert Consensus for Artificial Intelligence Diagnostic Model-Assisted Electroencephalography Diagnosis of Epilepsy in the Epilepsy Subspecialty Group of the Department of Neurosurgery of the Second Hospital of Hebei Medical University. Expert consensus on deploying AI diagnostic models to support electroencephalogram-based epilepsy diagnosis[J]. Clinical Focus, 2026, 41(2): 164-168.

Add to citation manager EndNote|Ris|BibTeX

URL: https://www.lchc.cn/EN/10.3969/j.issn.1004-583X.2026.02.011

https://www.lchc.cn/EN/Y2026/V41/I2/164

Figures/Tables 2

References 13

[1]	GBD 2016 Epilepsy Collaborators. Global, regional, and national burden of epilepsy, 1990-2016: A systematic analysis for the Global Burden of Disease Study 2016[J]. Lancet Neurol, 2019, 18(4):357-375. doi: 10.1016/S1474-4422(18)30454-X.
[2]	Kural MA, Duez L, Sejer Hansen V, et al. Criteria for defining interictal epileptiform discharges in EEG: A clinical validation study[J]. Neurology, 2020, 94(20):e2139-e2147. doi: 10.1212/WNL.0000000000009439.
[3]	Ibrahim FE, Emara HM, El-Shafai W, et al. Deep-learning-based seizure detection and prediction from electroencephalography signals[J]. Int J Numer Method Biomed Eng, 2022, 38(6):e3573. doi: 10.1002/cnm.3573.
[4]	Wang C, Yuan X, Jing W. Artificial intelligence in electroencephalography analysis for epilepsy diagnosis and management[J]. Front Neurol, 2025, 16:1615120. doi: 10.3389/fneur.2025.1615120.
[5]	Rao W, Guo J, Zhu C, et al. Interictal Epileptiform Discharge Detection using dual-domain features and GAN[J]. IEEE J Biomed Health Inform, 2025 Sep 5;PP. doi: 10.1109/JBHI.2025.3605257. Epub ahead of print.
[6]	Chen Y, Ou Z, Zhou D, Wu X. Advancements and challenges of artificial intelligence-assisted electroencephalography in epilepsy management[J]. J Clin Med, 2025, 14(12):4270. doi: 10.3390/jcm14124270.
[7]	Kinney MO, Kovac S, Diehl B. Structured testing during seizures: A practical guide for assessing and interpreting ictal and postictal signs during video EEG long term monitoring[J]. Seizure, 2019, 72:13-22. doi: 10.1016/j.seizure.2019.08.008. pmid: 31546090
[8]	Kellaway P. Sleep and epilepsy[J]. Epilepsia, 1985, 26(Suppl 1):S15-S30. doi: 10.1111/j.1528-1157.1985.tb05720.x.
[9]	Chahid A, Albalawi F, Alotaiby TN, et al. QuPWM: Feature extraction method for epileptic spike classification[J]. IEEE J Biomed Health Inform, 2020, 24(10):2814-2824. doi: 10.1109/JBHI.2020.2972286.
[10]	Steyerberg EW, Moons KG, van der Windt DA, et al. Prognosis research strategy (PROGRESS) 3: Prognostic model research[J]. PLoS Med, 2013, 10(2):e1001381. doi: 10.1371/journal.pmed.1001381.
[11]	Rieke N, Hancox J, Li W, et al. The future of digital health with federated learning[J]. NPJ Digit Med, 2020, 3:119. doi: 10.1038/s41746-020-00323-1. pmid: 33015372
[12]	Tjoa E, Guan C. A Survey on explainable artificial intelligence (XAI): Toward medical XAI[J]. IEEE Trans Neural Netw Learn Syst, 2021, 32(11):4793-4813. doi: 10.1109/TNNLS.2020.3027314.
[13]	中国医疗器械行业协会人工智能医疗器械专业委员会. 深度学习辅助决策医疗器械软件审评要点[EB/OL]. (2025-09-30). http://aimd.org.cn/newsinfo/1339997.html.

核心功能	核心算法	预期用途
伪差去除功能	基于信号处理技术的算法	辅助医生阅读脑电图,排除干扰信号
通过深度学习模型获取包含生理波形片段	基于注意力机制的深度学习模型	区分只包含生理波形的脑电和出现异常放电的脑电
基于特征的异常放电检测算法	机器学习分类器	辅助医生找到脑电图中癫痫样放电的位置
判断脑电是否有癫痫样放电	基于自适应阈值的算法	辅助医生给出脑电图关于癫痫样放电的结论
癫痫样放电的导联定位	基于自适应阈值的算法	辅助医生将癫痫样放电定位在具体导联上

核心功能	核心算法	预期用途
伪差去除功能	基于信号处理技术的算法	辅助医生阅读脑电图,排除干扰信号
通过深度学习模型获取包含生理波形片段	基于注意力机制的深度学习模型	区分只包含生理波形的脑电和出现异常放电的脑电
基于特征的异常放电检测算法	机器学习分类器	辅助医生找到脑电图中癫痫样放电的位置
判断脑电是否有癫痫样放电	基于自适应阈值的算法	辅助医生给出脑电图关于癫痫样放电的结论
癫痫样放电的导联定位	基于自适应阈值的算法	辅助医生将癫痫样放电定位在具体导联上

[1]	Liu Yijing, Zuo Chun, Luo Linzheng, Zhao Mingmin, Ren Bowen, Li Na. Youth stroke presenting with seizure-like onset: A case report and literature review [J]. Clinical Focus, 2026, 41(2): 155-159.
[2]	Wang Xiaoxiao, Hu Yanfeng, Qi Xiufeng, Zhao Junjie. Acute pulmonary embolism with epilepsy attack as the first clinical manifestation: A case report and literature review [J]. Clinical Focus, 2025, 40(6): 532-536.
[3]	Lin Bincheng, Lin Gangxi. Effects of etiology and preoperative course of disease on the efficacy of drug-resistant epilepsy in children [J]. Clinical Focus, 2025, 40(6): 527-531.
[4]	Jiao Fei, Yao Zhiping, Wang Langui. Risk factors of seizures in patients with central nervous system cystic echinococcosis [J]. Clinical Focus, 2024, 39(8): 712-715.
[5]	Wang Caizhen, Miao Lina, Chen Yuan, Li Shuangcheng. Effectiveness of high-frequency vagus nerve stimulation in the treatment of drug-resistant epilepsy: A meta-analysis [J]. Clinical Focus, 2024, 39(7): 593-597.
[6]	Zhao Xin, Liu Yun. Generative artificial intelligence: Ethical review on the empowerment in medical humanities education [J]. Clinical Focus, 2024, 39(1): 65-69.
[7]	Zhao Shuzhen, Wang Sanping, Zhao Xiaoyun. SLC32A1 gene mutation in hereditary epilepsy with febrile convulsion addition: A case report and literature review [J]. Clinical Focus, 2023, 38(5): 451-454.
[8]	Ding Siqi, Liu Shihua, Zhang Chao, Zhong Ping, Cao Li. Risk factors for epilepsy after delayed post-stroke epilepsy and its clinical correlation with blood Hcy, hs-CRP and D-D [J]. Clinical Focus, 2023, 38(10): 893-897.
[9]	Song Yiran, Liang Xiaonan, Li Chenyang, Zhang Xiaolan. Interpretation on Chinese Expert Consensus of Irritable Bowel Syndrome in 2020 [J]. Clinical Focus, 2021, 36(7): 628-631.
[10]	Yang Xin, Yao Shuyan, Yang Hongjie. Intelligent system in out-of-hospital management of patients with type 2 diabetes mellitus [J]. Clinical Focus, 2021, 36(10): 927-932.
[11]	Nie Ximing1, Wang Long2, Liu Liping1,Yang Zhonghua1. Application of artificial intelligence in neurocritical care [J]. Clinical Focus, 2018, 33(4): 277-281.
[12]	Wang Mina, Gao Weib, Hu Xiaoweia, Wu Chaob, Xu Mina. Preoperative localization and surgical outcome analysis of multifocal refractory epilepsy [J]. Clinical Focus, 2018, 33(12): 1031-1035.
[13]	Dong Lia， Li Xianghaib. Curative effect of oxcarbazepine on infants with partial epilepsy and immune function [J]. Clinical Focus, 2017, 32(10): 881-884.
[14]	Zhao Sena， Zhang Songb， Song Ninga， Jia Weihuaa， Yuan Shengfanga， Guo Xianlia. Interpretation of consensus guidelines for investigation and management of encephalitis in adults and children in Australia and New Zealand [J]. Clinical Focus, 2016, 31(12): 1370-1376.