Predictive value of fine motor deficits for mild cognitive impairment in the elderly based on machine learning
10.3969/j.issn.1009-0126.2025.06.004
- VernacularTitle:基于机器学习策略探究精细运动缺陷对老年轻度认知功能障碍患者的预测价值
- Author:
Yejing ZHAO
1
;
Yanyan ZHAO
1
;
Jie ZHANG
1
;
Han CUI
1
;
Ji SHEN
1
;
Ying YUAN
1
;
Hong SHI
1
;
Jing LI
1
Author Information
1. 100730 北京医院老年医学科 国家老年医学中心 中国医学科学院老年医学研究院
- Publication Type:Journal Article
- Keywords:
aged;
Alzheimer disease;
cognitive dysfunction;
forecasting;
logistic models
- From:
Chinese Journal of Geriatric Heart Brain and Vessel Diseases
2025;27(6):705-711
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the characteristics of fine motor deficits in the elderly individuals with MCI due to AD through a new wearable inertial motion capture device,and then construct a prediction model for MCI.Methods A total of 260 elderly subjects were recruited in community from November,2022 to April,2023,and based on diagnosis,they were divided into a MCI group(134 cases)and a control group(126 cases).A new wearable inertial motion capture device,which was self-designed and developed based on MEMS inertial sensor,was used to capture the fine mo-tor movements of the hands,and the obtained data were analyzed with a computerized assessment system to make the quantitative evaluation of fine motor.LASSO learning algorithm and logistic regression analysis were employed to identify the predictive factors for MCI,and then a nomo-gram was constructed based on these factors.ROC curve was plotted to evaluate the predictive ability of the model by calculating its AUC value.DC A,CIC,and Bootstrap method were applied to evaluate and validate the clinical utility and stability of the model.Results The total score of MoCA(22.18±2.84 vs 27.60±1.10)and scores of the dimensions were significantly lower in the MCI group than the control group(all P＜0.01).In the five digital assessment tasks,the MCI group showed obviously poorer fine motor performance of both hands than the control group(P＜0.05,P＜0.01).ROC curve analysis showed that the AUC value of our nomogram model in predicting MCI was 0.762(95％CI:0.705-0.819).DCA,CIC,and Bootstrap methods demonstra-ted good and relatively stable discrimination,calibration,and clinical applicability of the model.Conclusion MEMS inertial sensor motion capture technology can make digital evaluation of fine motor.For the elderly,fine motor deficits are significantly associated with risk for MCI.Our no-mogram model based on fine motion parameters shows good predictive efficacy in assessing the risk of MCI.
