基于随机森林算法建立甲状腺功能减退患病风险预测模型

doi:10.3969/j.issn.1000-8535.2023.07.003

广州医药 ›› 2023, Vol. 54 ›› Issue (7): 16-24.DOI: 10.3969/j.issn.1000-8535.2023.07.003

基于随机森林算法建立甲状腺功能减退患病风险预测模型

杨正霞¹, 王和勇¹, 贺施琪², 刘城³, 王天逸⁴, 张圣辉³, 毛晓健⁵

1 华南理工大学电子商务系（广州 510006）;
2 广州华银医学检验中心有限公司（广州 510670）;
3 广州市第一人民医院心内科,华南理工大学（广州 510180）;
4 广州医科大学（广州 510660）;
5 广州医科大学附属广州市妇女儿童医疗中心（广州 510623）

收稿日期:2023-01-13 出版日期:2023-07-20 发布日期:2023-08-15
通讯作者: 毛晓健,E-mail：xjamao@tom.com

Establishing a hypothyroidism risk prediction model based on random forest algorithm

YANG Zhengxia¹, WANG Heyong¹, HE Shiqi², LIU Cheng³, WANG Tiangyi⁴, ZHANG Shenghui³, MAO Xiaojian⁵

1 Department of Electronic Business, South China University of Technology, Guangzhou 510006, China;
2 Guangzhou Huayin Medical Laboratory Center Co., Ltd., Guangzhou 510670, China;
3 Department of Cardiology, Guangzhou First People's Hospital, South China University of Technology, Guangzhou 510180, China;
4 Guangzhou Medical University, Guangzhou 510660, China;
5 Guangzhou Women and Children's Medical Center Affiliated to Guangzhou Medical University, Guangzhou 510623, China

Received:2023-01-13 Online:2023-07-20 Published:2023-08-15

摘要/Abstract

摘要： 目的基于随机森林方法构建甲状腺功能减退（简称甲减）患病风险预测模型。方法从MIMIC-IV数据库纳入5 735名甲减患者为病例组,4 803名非甲减患者为对照组,基于随机森林模型进行建模。同时利用逻辑回归、贝叶斯正则化神经网络、XGBoost作为比较模型。最后用准确率、F1分数、精确率、召回率、特异性以及AUC值评价四个机器学习模型性能。结果随机森林模型准确率为0.85,F1分数为0.84,精确率为0.84,召回率为0.84,特异性为0.86,AUC值为0.91。在该模型中,促甲状腺激素、年龄、绝对淋巴细胞计数、血液中红细胞数、中性白细胞、性别、碱性磷酸酶、丙氨酸氨基转移酶、嗜酸性粒细胞绝对计数、尿素氮为甲减患者诊断重要性排前10的指标。结论采用随机森林方法构建的甲减患病预测模型为甲减的早期诊断有潜在应用价值。

关键词: 甲状腺功能减退症, 随机森林, 预测模型, MIMIC-IV数据库

Abstract: Objective To construct a risk prediction model for hypothyroidism based on the random forest model．Methods A total of 5 735 hypothyroidism patients were included from the MIMIC-IV database as the case group, and 4 803 non-hypothyroidism patients were included as the control group．Random forest models were constructed for both groups, and logistic regression, Bayesian regularized neural network, and XGBoost were used as comparative models．The performance of the four machine learning models was evaluated using accuracy, F1 score, precision, recall, specificity, and AUC value．Results The random forest model had an accuracy of 0.85, an F1 score of 0.84, a precision of 0.84, a recall of 0.84, a specificity of 0.86, and an AUC value of 0.91．In this model, thyroid-stimulating hormone, age, absolute lymphocyte count, red blood cell count in blood, neutrophil, gender, alkaline phosphatase, aspartate aminotransferase, absolute eosinophil count, and blood urea nitrogen were the top 10 indicators for diagnosing hypothyroidism patients．Conclusions The hypothyroidism disease prediction model constructed using the random forest method has potential application value for the early diagnosis of hypothyroidism．

Key words: hypothyroidism, random forest, predictive model, MIMIC-IV database

杨正霞, 王和勇, 贺施琪, 刘城, 王天逸, 张圣辉, 毛晓健. 基于随机森林算法建立甲状腺功能减退患病风险预测模型[J]. 广州医药, 2023, 54(7): 16-24.

YANG Zhengxia, WANG Heyong, HE Shiqi, LIU Cheng, WANG Tiangyi, ZHANG Shenghui, MAO Xiaojian. Establishing a hypothyroidism risk prediction model based on random forest algorithm[J]. Guangzhou Medical Journal, 2023, 54(7): 16-24.

参考文献

[1] DUNTAS L H,YEN P M.Diagnosis and treatment of hypothyroidism in the elderly[J].Endocrine,2019,66(1):63-69.
[2] CHAKER L,BIANCO A C,JONKLAAS J,et al.Hypothyroidism[J].Lancet,2017,390(10101):1550-1562.
[3] BOUCAI L,HOLLOWELL J G,SURKS M I.An approach for development of age-,gender-,and ethnicity-specific thyrotropin reference limits[J].Thyroid,2011,21(1):5-11.
[4] SHARMA A K,ARYA R,MEHTA R,et al.Hypothyroidism and cardiovascular disease:Factors,mechanism and future perspectives[J].Curr Med Chem,2013,20(35):4411-4418.
[5] BREIMAN L.Random forests[J].Machine Learning,2001,45(1):5-32.
[6] SAI P V,RAJALAKSHMI T,SNEKHALATHA U.Non-invasive thyroid detection based on electroglottogram signal using machine learning classifiers[J].Proc Inst Mech Eng H,2021,235(10):1128-1145.
[7] LUONG G,IDARRAGA AJ,HSIAO V,et al.Risk stratifying indeterminate thyroid nodules with machine learning[J].J Surg Res,2022,1(270):214-220.
[8] ZHANG H,ZHANG Z,LIU X,et al.DNA methylation haplotype block markers efficiently discriminate follicular thyroid carcinoma from follicular adenoma[J].J Clin Endocrinol Metab,2021,106(4):1011-1021.
[9] JOHNSON A,BULGARELLI L,POLLARD T,et al.Data from:MIMIC-IV (version 2.0).2022[DB].http://physionet.org/content/mimiciv/2.0/
[10] GOLDBERGER A L,AMARAL L A N,GLASS L,et al.PhysioBank,PhysioToolkit,and PhysioNet:components of a new research resource for complex physiologic signals[J].Circulation,2000,101(23):e215-e220.
[11] STEKHOVEN D J,BÜHLMANN P.MissForest—non-parametric missing value imputation for mixed-type data[J].Bioinformatics,2012,28(1):112-118.
[12] HOSMER Jr D W,LEMESHOW S,STURDIVANT R X.Applied logistic regression[M].John Wiley & Sons,2013.
[13] BURDEN F,WINKLER D.Bayesian regularization of neural networks[J].Artificial neural networks:methods and applications,2009:23-42.
[14] CHEN T,GUESTRIN C.Xgboost:A scalable tree boosting system[C]//Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining.2016:785-794.
[15] KIM W,CHO Y A,KIM D C,et al.Factors associated with thyroid-related adverse events in patients receiving PD-1 or PD-L1 inhibitors using machine learning models[J].Cancers,2021,13(21):5465.
[16] SISKIND S M,LEE S Y,PEARCE E N.Investigating hypothyroidism[J].BMJ,2021,1(373):n993.
[17] JABBAR A,PINGITORE A,PEARCE S H S,et al.Thyroid hormones and cardiovascular disease[J].Nature Reviews Cardiology,2017,14(1):39-55.
[18] CHAKER L,RAZVI S,BENSENOR I M,et al.Hypothyroidism[J].Nat Rev Dis Primers,2022,8(1):30.

基于随机森林算法建立甲状腺功能减退患病风险预测模型

Establishing a hypothyroidism risk prediction model based on random forest algorithm

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 2

编辑推荐

Metrics

本文评价

[1]	谢恋, 卢慧英, 王瑞瑜, 马为, 赵俊, 唐带君. 老年吸入性肺炎的危险因素分析及风险预测模型构建[J]. 广州医药, 2022, 53(2): 12-16.
[2]	古嘉基, 连泳欣, 丘福满. m6A甲基化基因风险评估模型在卵巢癌预后的临床意义[J]. 广州医药, 2021, 52(3): 1-8.