静脉血栓栓塞症患者住院时间机器学习预测模型的建立

Predicting hospitalization time for patients with venous thromboembolism through artificial intelligence technology

摘要

目的 通过比较5种监督学习算法在预测静脉血栓栓塞症(VTE)患者住院时间方面的性能,开发一个有效的机器学习预测模型,以深入理解影响VTE患者住院时长的因素。方法 研究西安交通大学第一附属医院2018年至2023年间收集的1 854例VTE患者数据,涵盖数据预处理、缺失数据处理、属性编码、数据简化、模型构建和模型测试等6个核心步骤。采用5种机器学习算法进行模型训练和比较。结果 支持向量机(SVM)模型在预测住院时长方面表现最佳,准确率高达94.53%。通过分析发现,血红蛋白、心脏病、年龄、高血压和肌酐是影响住院时长的主要因素。此外药物使用也显示出对住院时长有显著影响。结论 本研究成功开发了一个基于SVM的预测模型,能够准确预测VTE患者的住院时间。该模型不仅为医院提供了优化床位资源配置和运营效率的工具,也为患者提供了更好的治疗规划和时间安排支持。

Abstract

Objective To compare the performance of five supervised learning algorithms in predicting the length of hospital stay for venous thromboembolism (VTE) patients, so as to develop an effective machine learning prediction model to gain a deeper understanding of the factors that affect the length of hospital stay for VTE patients. Methods This study utilized data from 1 854 VTE patients collected from a tertiary hospital in northwest China from 2018 to 2023, covering six core steps: data preprocessing, missing data processing, attribute encoding, data simplification, model construction, and model testing. Five machine learning algorithms were used for model training and comparison. Results The support vector machine (SVM) model performed the best in predicting hospitalization duration, with an accuracy rate of 94.53%. Through analysis, it was found that hemoglobin, heart disease, age, hypertension, and creatinine were the main factors affecting hospitalization duration. In addition, drug use (such as warfarin sodium) also showed a significant impact on hospitalization duration. Conclusion This study successfully developed an SVM-based prediction model that can accurately predict the length of hospital stay for VTE patients. This model not only provides a tool for hospitals to optimize bed resource allocation and operational efficiency, but also provides better treatment planning and scheduling support for patients.

关键词

静脉血栓栓塞症(VTE) / 人工智能 / 住院时间预测 / 机器学习 / 医疗管理优化

Key words

venous thromboembolism (VTE) / artificial intelligence / prediction of hospitalization time / machine learning / medical management optimization

参考文献

[1] KEARON C, AKL E A, ORNELAS J, et al. Antithrombotic therapy for VTE disease: CHEST guideline and expert panel report[J]. Chest, 2016, 149(2): 315-352.

[2] LEBRUN L A. Effects of length of stay and language proficiency on health care experiences among immigrants in Canada and the United States[J]. Soc Sci Med, 2012, 74(7): 1062-1072.

[3] SPRIVULIS P C, DA SILVA J A, JACOBS I G, et al. The association between hospital overcrowding and mortality among patients admitted via Western Australian emergency departments[J]. Med J Aust, 2006, 184(5): 208-212.

[4] GOULD M K, GARCIA D A, WREN S M, et al. Prevention of VTE in nonorthopedic surgical patients: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence—based clinical practice guidelines[J]. Chest, 2012, 141(2 Suppl): e227S.

[5] 申丹丹 . 基于机器学习算法的患者康复需求预测模型构建[J]. 电子设计工程, 2025, 33(11): 182-186.

[6] SHEN D D. Construction of patient rehabilitation demand prediction model based on machine learning algorithm[J]. Electron Des Eng, 2025, 33(11): 182-186.

[7] 陈舸, 李观海, 杨朔, 等 . 乙肝肝硬化患者次年住院风险及高医疗费用预测模型的构建[J]. 华南预防医学, 2025, 51(2): 142-147.

[8] CHEN G, LI G H, YANG S, et al. Prediction models of the following year hospitalization risk and high medical cost for patients with hepatitis B cirrhosis[J]. South China J Pre Med, 2025, 51(2): 142-147.

[9] ARAB M, ZAREI A, RAHIMI A, et al. Analysis of factors affecting length of stay in public hospitals in Lorestan province, Iran[J]. J Res Med Sci, 2010, 15(4): 265-272.

[10] 杨冰倩 . 基于医疗大数据环境的疾病预测模型设计[J]. 科技资讯, 2024, 22(16): 41-44.

[11] YANG B Q. Design of disease prediction model based on medical big data environment[J]. Sci Technol Inf, 2024, 22(16): 41-44.

[12] 龙建英 . 基于机器学习的患者非计划重返ICU风险预测模型的构建与验证[D].兰州: 兰州大学, 2024.

[13] LONG J Y. Construction and validation of risk prediction model for unplanned readmission to ICU based on machine learning[D]. Lanzhou:Lanzhou University, 2024.

[14] 梁芳 . 腹腔镜卵巢癌根治术后静脉血栓栓塞症风险因素的Logistic回归分析[J]. 吉林医学, 2025, 46(7): 1603-1606.

[15] LIANG F. Logistic regression analysis of risk factors for venous thromboembolism after laparoscopic radical ovarian cancer surgery[J]. Jilin Med J, 2025, 46(7): 1603-1606.

[16] 马香爱, 何有华, 俞英, 等 . 腹腔镜前列腺癌根治术后患者静脉血栓栓塞症风险预测模型的构建[J]. 温州医科大学学报, 2025, 55(2): 112-120.

[17] MA X A, HE Y H, YU Y, et al. Construction of a risk prediction model for venous thromboembolism in patients after laparoscopic radical prostatectomy[J]. J Wenzhou Med Univ, 2025, 55(2): 112-120.

[18] 王蒙蒙, 曾焕, 董梓扬, 等 . 脓毒症合并急性静脉血栓预测模型的建立与验证[J]. 同济大学学报(医学版), 2025, 46(1): 38-45.

[19] WANG M M, ZENG H, DONG Z Y, et al. Construction and validation of a predictive model for acute venous thromboembolism in patients with sepsis[J]. J Tongji Univ (Med Sci), 2025, 46(1): 38-45.

[20] LIN C L, LIN P H, CHOU L W, et al. Model—based prediction of length of stay for rehabilitating stroke patients[J]. J Formos Med Assoc, 2009, 108(8): 653-662.

[21] JIANG X, QU X, DAVIS L. Using data mining to analyze patient discharge data for an urban hospital[C]//Proceedings of the International Conference on Data Mining. Las Vegas, Nevada, USA: DMIN, 2010: 1015-1020.

[22] TANUJA S, SHAILESH K, DINESH U A. Comparison of different data mining techniques to predict hospital length of stay[J]. J Pharm Biomed Sci, 2011, 3(2): 189-194.

[23] 方建冰, 黎秀婵, 赖海燕, 等 . 脑肿瘤术后患者发生静脉血栓栓塞症的危险因素及列线图预测模型的构建[J]. 广西医学, 2024, 46(12): 1877-1885.

[24] FANG J B, LI X C, LAI H Y, et al. Risk factors for the occurrence of venous thromboembolism in patients after brain tumor surgery and the establishment of a nomogram prediction model[J]. Guangxi Med J, 2024, 46(12): 1877-1885.

[25] ZHONG W, CHOW R, HE J. Clinical charge profiles prediction for patients diagnosed with chronic diseases using multilevel support vector machine[J]. Expert Syst Appl, 2012, 39(1): 1474-1483.

[26] FREITAS A, SILVA—COSTA T, LOPES F, et al. Factors influencing hospital high length of stay outliers[J]. BMC Health Serv Res, 2012, 12: 265.

[27] 田双, 黄晓莉, 周翌, 等 . 高出血风险患者左心耳封堵术后静脉血栓栓塞发生风险的临床研究[J]. 血管与腔内血管外科杂志, 2024, 10(12): 1506—1510, 1523.

[28] TIAN S, HUANG X L, ZHOU Y, et al. Clinical study of left atrial appendicular closure for reducing the risk of venous thromboembolism in patients with high bleeding risk[J]. J Vasc Endovasc Surg, 2024, 10(12): 1506-1510, 1523.

[29] 肖月琼, 陈宋璋 . 基于人工神经网络在基层医疗冠心病早期筛查中的应用[J]. 心电图杂志(电子版), 2020, 9(1): 80-81.

[30] XIAO Y Q, CHEN S Z. Application of artificial neural network in early screening of coronary heart disease in primary medical care[J]. J Electrocardiogram (Elec Ed), 2020, 9(1): 80-81.

[31] BYUN H, JEON S, YI E S. Analysis and prediction of older adult sports participation in South Korea using artificial neural networks and logistic regression models[J]. BMC Geriatr, 2023, 23(1): 676.

[32] 邢红雨, 毛毅敏 . 肺癌合并静脉血栓栓塞症危险因素及预测模型研究进展[J]. 中国现代医药杂志, 2024, 26(10): 102-106.

[33] XING H Y, MAO Y M. Research progress on risk factors and prediction models of lung cancer complicated with venous thromboembolism[J]. Mod Med J China, 2024, 26(10): 102-106.

[34] ZHENG B, ZHANG J, YOON S W, et al. Predictive modeling of hospital readmissions using metaheuristics and data mining[J]. Expert Syst Appl, 2015, 42(20): 7110-7120.

[35] 陈娅娟, 周蕊, 孙丹, 等 . 住院患儿静脉血栓栓塞症风险预测模型的构建及验证[J]. 中华护理杂志, 2024, 59(16): 1966-1972.

[36] CHEN Y J, ZHOU R, SUN D, et al. Construction and validation of a risk prediction model for venous thromboembolism in hospitalized children[J]. Chin J Nurs, 2024, 59(16): 1966-1972.

[37] 祝甜, 肖珍, 何秀, 等 . 基于Khorana评分构建老年肺癌患者静脉血栓栓塞症风险预测模型[J]. 中国防痨杂志, 2024, 46(S1): 103-105.

[38] ZHU T, XIAO Z, HE X, et al. Construction of risk prediction model for venous thromboembolism in elderly lung cancer patients based on Khorana score[J]. Chin J Antituberculosis, 2024, 46(S1): 103-105.

[39] CHERIF W. Optimization of K—NN algorithm by clustering and reliability coefficients: application to breast—cancer diagnosis[J]. Procedia Comput Sci, 2018, 127: 293-299.

基金资助

国家卫生健康委医院管理研究所2024年医疗人工智能临床应用研究课题(YLXX24AIA021)