| 基于随机森林算法的支原体肺炎患儿并发器官功能障碍预测模型构建 |
| 投稿时间：2025-07-15 修订日期：2025-08-05 点此下载全文 |
| 引用本文：徐莉,曹长春,茆丽丽,李光珍,潘克香,王会芳.基于随机森林算法的支原体肺炎患儿并发器官功能障碍预测模型构建[J].医学研究杂志,2025,54(12):82-87 |
| DOI：
10.11969/j.issn.1673-548X.2025.12.014 |
| 摘要点击次数: 22 |
| 全文下载次数: 21 |
|
| 基金项目:江苏省卫生健康委员会医学科研项目(ZB202026)；江苏省淮安市卫生健康科研项目(HAWJ202104) |
|
| 中文摘要:目的 对支原体肺炎(mycoplasma pneumoniae pneumonia, MPP)患儿并发器官功能障碍(organ dysfunction, OD)的影响因素进行分析,并构建基于随机森林算法的风险预测模型,以为临床医务人员制定针对性的防治措施提供科学的理论依据。方法 回顾性选取2021年1月～2024年1月南京医科大学附属淮安第一医院收治的160例MPP患儿作为研究对象,根据是否发生OD,将其分为OD组和非OD组。收集两组患儿的相关临床资料,采用单因素分析和多因素Logistic回归分析筛选影响MPP患儿并发OD的危险因素,并采用R 4.3.3软件构建MPP患儿并发OD的风险预测模型。结果 160例MPP患儿中有41例发生OD,发生率为25.63%。OD组和非OD组患儿的热程、热峰、C反应蛋白(C-reactive protein, CRP)、凝血酶原时间(prothrombin time, PT)以及D-二聚体(D-dimer, D-D)水平比较,差异均有统计学意义(P＜0.05)。多因素Logistic回归分析结果显示,热程、热峰、CRP、PT以及D-D为MPP患儿发生OD的危险因素。基于上述因素构建MPP患儿并发OD的随机森林模型,结果显示,影响MPP患儿并发OD的因素的重要性排序依次为PT、CRP、D-D、热峰以及热程。随机森林模型预测MPP患儿并发OD的曲线下面积(area under the curve, AUC)值为0.907(95%CI:0.851～0.947),Logistic回归模型预测MPP患儿并发OD的AUC值为0.849(95%CI:0.784～0.900)。使用Delong检验对2种模型的AUC值进行比较,结果显示,z=2.073,P=0.0381,表明随机森林模型的预测效能优于Logistic回归模型。结论 热程、热峰、CRP、PT以及D-D为MPP患儿并发OD的危险因素,临床医务人员应针对该类患儿采取针对性的防治措施,减少OD的发生。本研究构建的MPP患儿并发OD的随机森林模型准确性较高,能够为临床预防和治疗提供科学的理论依据。 |
| 中文关键词:随机森林算法 支原体肺炎 患儿 器官功能障碍 预测模型 |
| |
| Construction of Predictive Model for Organ Dysfunction in Children with Mycoplasma Pneumonia Based on Random Forest Algorithm. |
|
|
| Abstract:Objective To analyze the influencing factors of organ dysfunction (OD) in children with mycoplasma pneumoniae pneumonia (MPP), and construct a risk prediction model based on random forest algorithm, to provide a scientific theoretical basis for clinical medical staff to formulate targeted preventive and treatment measures. Methods A retrospective study was conducted on 160 children with MPP admitted to the hospital from January 2021 to January 2024. The children were divided into the OD group and non-OD group based on the occurrence of OD. The relevant clinical data of the two groups of children were collected, and univariate analysis and multivariate Logistic regression analysis were used to screen the risk factors affecting the occurrence of OD in children with MPP. The R 4.3.3 software was used to construct a risk prediction model for OD in children with MPP. Results Among the 160 children with MPP, 41 children developed OD, with an incidence of 25.63%. The differences were all statistically significant in terms of fever duration, peak fever, C-reactive protein (CRP), prothrombin time (PT), and D-dimer (D-D) levels between the OD group and non-OD group. The results of the multivariate Logistic regression analysis showed that fever duration, peak fever, CRP, PT and D-D were the risk factors for OD in children with MPP. The random forest model based on the above factors for OD in children with MPP showed that the importance of factors affecting the occurrence of OD in children with MPP was in the order of PT, CRP, D-D, peak fever, and fever duration. The area under the curve (AUC) value of the random forest model for predicting OD in children with MPP was 0.907 (95%CI:0.851-0.947), and the AUC value of the Logistic regression model was 0.849 (95%CI:0.784-0.900). The Delong test was used to compare the AUC value of the two models, and the result showed z=2.073, P=0.0381, indicating that the predictive efficacy of the random forest model was superior to that of the Logistic regression model. Conclusion Fever duration, peak fever, CRP, PT and D-D are the risk factors for OD in children with MPP. Clinical medical staff should take targeted preventive and treatment measures for such children to reduce the occurrence of OD. The random forest model for OD in children with MPP constructed in this study has a high accuracy, and can provide a scientific theoretical basis for clinical prevention and treatment. |
| keywords:Random forest algorithm Mycoplasma pneumonia Children Organ dysfunction Prediction model |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
