Application of quantitative electroencephalography in digital screening for mild cognitive impairment
10.3969/j.issn.1006-9771.2025.11.008
- VernacularTitle:定量脑电图在轻度认知障碍数字化筛查中的应用
- Author:
Jianpeng GU
1
;
Yulei SONG
;
Haiyan YIN
;
Tingting YIN
;
Fengyi SUN
;
Bingqing YANG
;
Minghui ZHAO
;
Guihua XU
;
Yamei BAI
Author Information
1. 南京中医药大学护理学院,江苏 南京市 210023
- Publication Type:Journal Article
- Keywords:
elderly;
mild cognitive impairment;
digital screening;
quantitative electroencephalography;
XGBoost
- From:
Chinese Journal of Rehabilitation Theory and Practice
2025;31(11):1314-1321
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the quantitative electroencephalography(qEEG)characteristics of the prefrontal cortex in patients with mild cognitive impairment(MCI)during digital screening tasks for MCI screening.Methods A total of 592 MCI patients(MCI group)and 317 normal cognitively elderly individuals(control group)were recruited from 40 communities in Nanjing,Jiangsu Province,from July to August,2024.All participants were as-sessed using Montreal Cognitive Assessment-Beijing Version(MoCA-BJ).Prefrontal EEG data were collected using a portable EEG device,and power spectral analysis was performed via Fast Fourier Transform.An XG-Boost algorithm was employed to construct an MCI identification model based on qEEG power features,and the model's performance was evaluated using receiver operating characteristic(ROC)curve.Results Compared with the control group,prefrontal δ,α,and β band power increased during screening tasks in MCI group(P<0.05);δ power was negatively correlated with MoCA-BJ total scores,and visuospatial/executive func-tion,attention and delayed recall scores(r=-0.269,-0.169,-0.133,-0.171,P<0.001);α power was negative-ly correlated with MoCA-BJ total scores,attention and delayed recall scores(r=-0.113,-0.075,-0.091,P<0.05).The XGBoost model based on δ and α power was excellent in MCI identification,with an area under the curve of 0.91,accuracy of 0.81,precision of 0.89,F1 score of 0.84,recall of 0.80,and specificity of 0.81.Conclusion MCI patients exhibit increased power in the prefrontal δ and α frequency bands during digital screening tasks,which is associated with cognitive decline.An XGBoost model based on qEEG power features can enable early prediction of MCI.
