一个根据血液化验结果建立的复杂性阑尾炎预测模型
A Blood Test Model for the Prediction of Complicated Appendicitis
DOI: 10.12677/ACM.2020.107191, PDF,   
作者: 张 驰, 臧守亚:青岛大学医学部第一临床医学院,山东 青岛;李振鲁, 高 鹏*:青岛大学附属医院,山东 青岛
关键词: 复杂性阑尾炎血液化验XGBoost预测模型Complicated Appendicitis Blood Test XGBoost Predictive Model
摘要: 目的:通过比较复杂性阑尾炎与非复杂性阑尾炎各项血液化验指标,分析复杂性阑尾炎的危险预测因素并建立预测模型。方法:记录2014年至2019年之间于青岛大学附属医院行阑尾切除术的患者,根据术后病理分为复杂性阑尾炎或非复杂性阑尾炎。采用统计分析方法比较两组间术前相关血液化验指标的差异,筛选出复杂性阑尾炎的危险因素。使用XGBoost建立模型,并使用SHAP对模型进行解释。结果:共记录阑尾切除术样本共566例,其中复杂性阑尾炎106例(复杂组),非复杂性阑尾炎460例(非复杂组)。两组比较,其中白细胞、中性粒细胞等11项指标存在显著性差异(p < 0.001)。随机抽取总体样本的70%作为训练集建立模型,以其余30%的样本设为验证集带入模型,计算该模型准确率为81.2%,AUC为0.93。通过SHAP对模型进行分析可发现:复杂性阑尾炎中C反应蛋白(80.93 ± 60.32比13.26 ± 23.26,p < 0.001)、纤维蛋白原(4.77 ± 1.39比3.22 ± 0.99,p < 0.001)及中性粒细胞(12.11 ± 4.38比7.35 ± 4.68,p < 0.001)明显升高,而淋巴细胞百分比明显降低(8.91 ± 5.08比22.41 ± 14.54,p < 0.001)。结论:通过血常规、C反应蛋白及纤维蛋白原等作为指标建立预测模型,能够较好于手术前区分复杂性阑尾炎与分复杂性阑尾炎,对于根据阑尾炎严重程度采取不同的治疗方案有重要的意义。
Abstract: Objective: Analyze the risk prediction factors of complicated appendicitis and establish a predictive model, by comparing the various blood test indicators of complicated appendicitis and uncomplicated appendicitis. Methods: The patients who underwent appendectomy at the Affiliated Hospital of Qingdao University from 2014 to 2019 were classified as complicated appendicitis or uncomplicated appendicitis according to the postoperative pathology. A statistical analysis method was used to compare the differences in preoperative blood test indexes between the two groups, and the risk factors for complicated appendicitis were screened out. XGBoost was used to build the predict model, and SHAP was used to explain the model. Results: A total of 566 samples of appendectomy were recorded, including 106 complicated appendicitis (complicated group) and 460 uncomplicated appendicitis (uncomplicated group). Compared between the two groups, there were significant differences in 11 indicators such as white blood cells and neutrophils (p < 0.001). 70% of the total samples are randomly selected as the training set to build the model, and the remaining 30% of the samples are used as the validation set to bring the model. The accuracy rate of the model is 81.2% and the AUC is 0.93. Analysis of the model through SHAP shows that in complicated appendicitis, C-reactive protein (80.93 ± 60.32 vs. 13.26 ± 23.26, p < 0.001), fibrinogen (4.77 ± 1.39 vs. 3.22 ± 0.99, p < 0.001), Neutrophil Count (12.11 ± 4.38 vs 7.35 ± 4.68, p < 0.001) was significantly increased, while the percentage of lymphocytes was significantly decreased (8.91 ± 5.08 vs 22.41 ± 14.54, p < 0.001). Conclusion: Using blood test results of C-reactive protein and fibrinogen as indicators to establish a predictive model can be used to distinguish complicated appendicitis from uncomplicated appendicitis before surgery, and is of great significance for different treatment options according to the severity of appendicitis.
文章引用:张驰, 李振鲁, 臧守亚, 高鹏. 一个根据血液化验结果建立的复杂性阑尾炎预测模型[J]. 临床医学进展, 2020, 10(7): 1252-1259. https://doi.org/10.12677/ACM.2020.107191

参考文献

[1] Tannoury, J. (2013) Treatment Options of Inflammatory Appendiceal Masses in Adults. World Journal of Gastroenterology, 19, 3942. [Google Scholar] [CrossRef] [PubMed]
[2] Podda, M., Gerardi, C., Cillara, N., et al. (2019) Antibiotic Treatment and Appendectomy for Uncomplicated Acute Appendicitis in Adults and Children: A Systematic Review and Meta-Analysis. Annals of Surgery, 270, 1028-1040. [Google Scholar] [CrossRef
[3] Di Saverio, S., Birindelli, A., Kelly, M.D., et al. (2016) WSES Jerusalem Guidelines for Diagnosis and Treatment of Acute Appendicitis. World Journal of Emergency Surgery, 11, Article No. 34.
[4] Leung, B., Madhuripan, N., Bittner, K., et al. (2019) Clinical Outcomes Following Identification of Tip Appendicitis on Ultrasonography and CT Scan. Journal of Pediatric Surgery, 54, 108-111. [Google Scholar] [CrossRef] [PubMed]
[5] Leeuwenburgh, M.M., Wiezer, M.J., Wiarda, B.M., et al. (2014) Accuracy of MRI Compared with Ultrasound Imaging and Selective Use of CT to Discriminate Simple from Perforated Appendicitis. British Journal of Surgery, 101, e147- e155. [Google Scholar] [CrossRef] [PubMed]
[6] Atema, J.J., Gans, S.L., Beenen, L.F., et al. (2015) Accuracy of White Blood Cell Count and C-Reactive Protein Levels Related to Duration of Symptoms in Patients Suspected of Acute Appendicitis. Academic Emergency Medicine, 22, 1015-1024. [Google Scholar] [CrossRef] [PubMed]
[7] Ahmed, N. (2017) C-Reactive Protein: An Aid For Diagnosis of Acute Appendicitis. Journal of Ayub Medical College Abbottabad, 29, 250-253.
[8] Yardimci, S., Ugurlu, M.U., Coskun, M., et al. (2016) Neutrophil-Lymphocyte Ratio and Mean Platelet Volume Can Be a Predictor for Severity of Acute Appendicitis. Ulus Travma Acil Cerrahi Derg, 22, 163-168. [Google Scholar] [CrossRef] [PubMed]
[9] Bhangu, A., Søreide, K., Di Saverio, S., et al. (2015) Acute Appendicitis: Modern Understanding of Pathogenesis, Diagnosis, and Management. The Lancet, 386, 1278-1287. [Google Scholar] [CrossRef
[10] Chen, T. and Guestrin, C. (2016) XGBoost: A Scalable Tree Boosting System. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery, San Francisco, 785-794. [Google Scholar] [CrossRef
[11] Lundberg, S.M., Nair, B., Vavilala, M.S., et al. (2018) Explainable Machine-Learning Predictions for the Prevention of Hypoxaemia during Surgery. Nature Biomedical Engineering, 2, 749-760. [Google Scholar] [CrossRef] [PubMed]
[12] Strumbelj, E. and Kononenko, I. (2014) Explaining Prediction Models and Individual Predictions with Feature Contributions. Knowledge and Information Systems, 41, 647-665. [Google Scholar] [CrossRef
[13] Pepys, M.B. and Baltz, M.L. (1983) Acute Phase Proteins with Special Reference to C-Reactive Protein and Related Proteins (Pentaxins) and Serum Amyloid A Protein. Advances in Immunology, 34, 141-212. [Google Scholar] [CrossRef
[14] Eriksson, S. and Granstrom, L. (1995) Randomized Controlled Trial of Appendicectomy versus Antibiotic Therapy for Acute Appendicitis. British Journal of Surgery, 82, 166-169. [Google Scholar] [CrossRef] [PubMed]
[15] Vons, C., Barry, C., Maitre, S., et al. (2011) Amoxicillin plus Clavulanic Acid versus Appendicectomy for Treatment of Acute Uncomplicated Appendicitis: An Open-Label, Non-Inferiority, Randomised Controlled Trial. The Lancet, 377, 1573-1579. [Google Scholar] [CrossRef
[16] Foley, T.A., Earnest, F., Nathan, M.A., et al. (2005) Differentiation of Nonperforated from Perforated Appendicitis: Accuracy of CT Diagnosis and Relationship of CT Findings to Length of Hospital Stay. Radiology, 235, 89-96. [Google Scholar] [CrossRef] [PubMed]
[17] Li, S., Cheng, L., Li, Y., et al. (2018) Analysis of High Risk Factors for Acute Complex Appendicitis in Adults. Chinese Journal of Gastrointestinal Surgery, 21, 1374-1379.
[18] Levine, C.D., Aizenstein, O., Lehavi, O., et al. (2005) Why We Miss the Diagnosis of Appendicitis on Abdominal CT: Evaluation of Imaging Features of Appendicitis Incorrectly Diagnosed on CT. American Journal of Roentgenology, 184, 855-859. [Google Scholar] [CrossRef] [PubMed]
[19] McGowan, D.R., Sims, H.M., Zia, K., et al. (2013) The Value of Biochemical Markers in Predicting a Perforation in Acute Appendicitis. ANZ Journal of Surgery, 83, 79-83. [Google Scholar] [CrossRef] [PubMed]
[20] Yu, C.W., Juan, L.I., Wu, M.H., et al. (2013) Systematic Review and Meta-Analysis of the Diagnostic Accuracy of Procalcitonin, C-Reactive Protein and White Blood Cell Count for Suspected Acute Appendicitis. British Journal of Surgery, 100, 322-329. [Google Scholar] [CrossRef] [PubMed]