Prediction model of insomnia disorder in perimenopausal women based on machine learning method
10.3760/cma.j.cn211501-20231115-01022
- VernacularTitle:基于机器学习法构建围绝经期女性失眠障碍的预测模型
- Author:
Jinli HU
1
;
Jiebai SHI
;
Fangli LIAO
;
Lixiang ZHANG
Author Information
1. 丽水市妇幼保健院中医科，丽水 323000
- Keywords:
Perimenopause;
Insomnia disorder;
Machine learning method;
Risk prediction model
- From:
Chinese Journal of Practical Nursing
2024;40(20):1535-1542
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the influencing factors of insomnia disorder in perimenopausal women and construct a prediction model of insomnia disorder in perimenopausal women based on machine learning method.Methods:A case-control study was used in this study. A total of 140 perimenopausal women who were examined in Lishui Maternal and Child Health Hospital from January 2019 to June 2021 were selected as the study objects for retrospective analysis by convenient sampling method. They were divided into occurrence and non-occurrence groups based on the presence of insomnia disorders. Relevant data of the patients were collected and risk factors analysis was conducted. Multivariate Logistic regression, decision classification regression tree (CRT) and back propagation neural network (BPNN) algorithm based on machine learning, the prediction model of insomnia disorder in perimenopausal women was constructed.Results:A total of 140 perimenopausal women were included, including 88 patients (62.86%) in the occurrence group, aged (50.16 ± 4.73) years old, and 52 patients (37.14%) in the non-occurrence group, aged (47.33 ± 4.54) years old. Multivariate Logistic regression analysis showed that percapita family monthly income ( OR = 0.019, 95% CI 0.001-0.422, P<0.05), Hamilton Depression Scale (HAMD) score ( OR = 1.665, 95% CI 1.108-2.502, P<0.05) and Self-rating Anxiety Scale (SAS) score ( OR = 1.407, 95% CI 1.085-1.826, P<0.05) of the two groups were independent risk factors for the occurrence of insomnia disorder in perimenopausal women. The prediction model constructed by CRT showed that SAS score, HAMD score and percapita family monthly income were the influencing factors for the occurrence of insomnia disorder in perimenopausal women. The results of BPNN model showed that the importance ranking of influencing factors was SAS score>percapita family monthly income>HAMD score>body mass index>age>work status>daily exercise cumulative time. Among the models constructed by the three machine learning algorithms, the area under the curve of multivariate Logistic regression analysis was 0.998, the sensitivity was 96.6%, the specificity was 100.0%, which had the best predictive performance. Conclusions:In this study, the prediction model of insomnia disorder in perimenopausal women based on machine learning method has good prediction efficiency, among which the multivariate Logistic regression model has the best diagnostic efficiency, and the established prediction model has good prediction accuracy.
