| 4 | 0 | 1 |
| 下载次数 | 被引频次 | 阅读次数 |
为了解决传统抑郁症预测模型因过于依赖单一模型而难以有效应对数据复杂性的问题，提出了一种基于ABS-Stacking算法的抑郁症预测模型。在传统Stacking模型基础上采用最佳优先搜索算法构建基分类器筛选层，以自适应选择最优的基分类器组合。通过5折交叉验证，根据各基模型在验证集上的AUC(area under curve)值对预测结果进行加权平均，使得表现较好的基模型在最终预测中贡献更大，从而提升模型的整体预测性能。在中老年结构化数据上的实验结果表明，ABS-Stacking模型在泛化能力和抑郁症预测效果上均优于单一模型和传统集成方法。该方法不仅有效解决了基分类器组合选择和性能加权的问题，还显著提高了模型的自适应性和泛化能力，为抑郁症预测提供了新的方法参考。
Abstract：In order to address the issue that traditional depression prediction models are overly dependent on a single model, making it challenging to effectively handle data complexity, a depression prediction model based on ABS-Stacking is proposed.This model incorporates a base classifier screening layer built on the traditional stacking model using a best-first search algorithm to adaptively select the optimal combination of base classifiers.Furthermore, through 5-fold cross-validation, the prediction results are weighted and averaged based on the area under curve values of each base model on the validation set.This approach ensures that better-performing base models contribute more to the final prediction, thereby enhancing the overall predictive performance of the model.Experimental results on structured data from middle-aged and elderly individuals demonstrate that the ABS-Stacking model outperforms both single models and traditional ensemble methods in terms of generalization ability and depression prediction effectiveness.This method not only effectively resolves issues related to base classifier combination selection and performance weighting but also significantly improves the model′s adaptability and generalization capability, providing a new methodological reference for depression prediction.
[1] WORLD H O.Depression and other common mental disorders:global health estimates[R/OL].(2017-03-01)[2024-03-20].https://www.who.int/publications/i/item/depression-global-health-estimates.
[2] SANTOMMAURO D F,HERRER A M M,SHADID J,et al.Global prevalence and burden of depressive and anxiety disorders in 204 countries and territories in 2020 due to the COVID-19 pandemic[J].Lancet,2021,398(10 312):1 700-1 712.
[3] SAU A,BHAKTA I.Artificial neural network (ANN) model to predict depression among geriatric population at a slum in Kolkata,India[J].Journal of Clinical and Diagnostic Research:JCDR,2017,11(5):VC01-VC04.
[4] MOREIRA M W L,RODRIGUES J J P C,KUMAR N,et al.Postpartum depression prediction through pregnancy data analysis for emotion-aware smart systems[J].Information Fusion,2019,47:23-31.
[5] 朱徐乐，徐林燕，赵敏，等.老年人阈下抑郁危险因素分析及风险预测模型的构建与验证[J].现代预防医学，2023,50(14):2 639-2 644.
[6] BHAUMIK R,JENKINS L M,GOWINS J R,et al.Multivariate pattern analysis strategies in detection of remitted major depressive disorder using resting state functional connectivity[J].NeuroImage Clinical,2017,16:390-398.
[7] ZHANG B T,CAI H S,SONG Y B,et al.Computer-aided recognition based on decision-level multimodal fusion for depression[J].IEEE Journal of Biomedical and Health Informatics,2022,26(7):3 466-3 477.
[8] 谷明轩，范冰冰.基于多模态特征融合的抑郁症识别[J].计算机与现代化，2023,39(10):17-22.
[9] ZILCHA-MANO S,ROOSE S P,BROWN P J,et al.Abrupt symptom improvements in antidepressant clinical trials:transient placebo effects or therapeutic reality?[J].The Journal of Clinical Psychiatry,2018,80(1):18m12353.
[10] BAILEY N W,HOY K E,ROGASCH N C,et al.Differentiating responders and non-responders to rTMS treatment for depression after one week using resting EEG connectivity measures[J].Journal of Affective Disorders,2019,242:68-79.
[11] GENG D Y,AN Q,FU Z G,et al.Identification of major depression patients using machine learning models based on heart rate variability during sleep stages for pre-hospital screening[J].Computers in Biology and Medicine,2023,162:107 060.
[12] KIM T,PARK U,KANG S W.Prediction model for potential depression using sex and age-reflected quantitative EEG biomarkers[J].Frontiers in Psychiatry,2022,13:913 890.
[13] DU M H,LIU S,WANG T,et al.Depression recognition using a proposed speech chain model fusing speech production and perception features[J].Journal of Affective Disorders,2023,323:299-308.
[14] KOUR H,GUPTA M K.An hybrid deep learning approach for depression prediction from user tweets using feature-rich CNN and bi-directional LSTM[J].Multimedia Tools and Applications,2022,81(17):23 649-23 685.
[15] 胡以婵，梁铭，谢灿荣，等.基于Stacking模型融合的高性能混凝土强度预测方法[J].硅酸盐通报，2023,42(11):3 914-3 926.
[16] 冯鑫磊，俞凯，袁贞明.改进Stacking算法在妊娠期糖尿病预测中的应用[J].杭州师范大学学报(自然科学版),2023,22(2):126-134.
[17] 北京大学国家发展研究院.2020年全国追踪调查[EB/OL].(2023-11-16)[2023-11-28].https://charls.charlsdata.com/pages/data/111/zh-cn.html.
[18] 刘悦.基于机器学习的老年人抑郁症状的预测[D].济南：山东大学，2023:23-24.
[19] 赵臣啸，薛惠锋，王磊，等.基于孤立森林算法的取用水量异常数据检测方法[J].中国水利水电科学研究院学报，2020,18(1):31-39.
[20] ANDREA S B,SIEGEL S A R,TEO A R.Social support and health service use in depressed adults:findings from the national health and nutrition examination survey[J].General Hospital Psychiatry,2016,39:73-79.
[21] KATZ S,FORD A B,MOSKOWITZ R W,et al.Studies of illness in the aged.the index of ADL:a standardized measure of biological and psychosocial function[J].Journal of the American Medical Association,1963,185:914-919.
[22] 刘芹，陆杰华.中老年人体质指数对日常生活活动能力的影响探究：基于CHARLS数据的验证[J].人口与发展，2020,26(6):40-51.
[23] 张宇波，雷鹏，卜阳，等.基于机器学习算法肝细胞癌伴微血管侵犯术前预测模型的构建与评估[J].中华肝胆外科杂志，2023,29(11):801-807.
[24] 刘辉玲，陶洁，邱磊.基于Python的One-hot编码的实现[J].武汉船舶职业技术学院学报，2021,20(3):136-139.
[25] 严颖，黄奇，李娜.基于优化后集成学习模型的特征选择与疾病高效预警研究：以老年抑郁焦虑为例[J].数据分析与知识发现，2023,7(7):74-88.
[26] 谭朋柳，张露玉，徐光勇等.基于多粒度自注意力机制的抑郁症预测模型[J].计算机应用，2023,43(S2):34-40.
基本信息:
DOI：10.19724/j.cnki.jmju.2025.02.007
中图分类号:R749.4;TP18
引用信息:
[1]张梓坪,林耿,龙素娟等.基于ABS-Stacking算法的抑郁症预测模型[J].闽江学院学报,2025,46(02):51-62.DOI:10.19724/j.cnki.jmju.2025.02.007.
基金信息:
国家自然科学基金(12001259); 福建省自然科学基金项目(2024J011180,2023J05251)