摘要
目的基于美国重症监护医学信息数据库Ⅳ(MIMIC-Ⅳ),利用机器学习技术探究糖尿病酮症酸中毒(DKA)患者的最佳降血糖方案,以提升患者的个性化治疗效果。方法基于MIMIC-Ⅳ数据库,分析2008至2019年贝斯以色列女执事医疗中心重症监护病房(ICU)2096例DKA患者的病例资料。构建机器学习模型,绘制受试者工作特征曲线(ROC曲线)和精确率-召回率曲线(PR曲线)评估模型对4种常见不良结局〔低血糖、低钾血症、格拉斯哥昏迷评分(GCS)下降和长期住院〕的预测效能;分析不良结局发生风险随血糖下降速度的变化趋势;采用单因素和多因素Logistic回归分析相关因素与低钾血症风险之间的关系;利用个性化风险解释方法和预测技术,对患者的最佳血糖控制范围进行个体化分析。结果所构建的机器学习模型在预测DKA患者不良结局方面展现出优异性能,其预测低血糖、低钾血症、GCS评分下降和长期住院的ROC曲线下面积(AUROC)及95%可信区间(95%CI)分别为0.826(0.803~0.849)、0.850(0.828~0.870)、0.925(0.903~0.946)、0.901(0.883~0.920)。患者血糖下降速度与4种不良结局风险的关系分析显示,最大血糖下降速度>6.26 mmol·L^(-1)·h^(-1)时,低血糖风险显著升高(P<0.001);最大血糖下降速度>2.72 mmol·L^(-1)·h^(-1)时,低钾血症风险显著升高(P<0.001);最大血糖下降速度>5.53 mmol·L^(-1)·h^(-1)时,GCS评分下降风险显著降低(P<0.001);最大血糖下降速度>8.03 mmol·L^(-1)·h^(-1)时,长期住院风险显著降低(P<0.001)。多因素Logistic回归分析显示,最大碳酸氢盐水平、血尿素氮水平和总胰岛素使用量与低钾血症风险显著相关(均P<0.01)。个性化最佳治疗阈值范围的制定方面,若假设低血糖、低钾血症、GCS评分下降、长期住院4种不良结局的最佳血糖下降阈值分别为x_(1)、x_(2)、x_(3)、x_(4),降低患者低血钾和低血糖发生风险的最佳血糖下降速度应≤min{x_(1),x_(2)};降低GCS评分下降和长期住院风险的血糖下降速度建议应≥max{x_(3),x_(4)};当两者有重叠,即max{x_(3),x_(4)}≤min{x_(1),x_(2)}时,该区间即推荐的最佳血糖下降速度范围。若最佳阈值范围无交集,即max{x_(3),x_(4)}>min{x_(1),x_(2)},需综合考虑患者各种不良结局风险的个体差异,进行动态调整治疗策略。结论机器学习模型在预测DKA患者不良结局方面表现良好,有助于个性化降血糖管理,具有重要的临床应用前景。
Objective To explore the optimal blood glucose-lowering strategies for patients with diabetic ketoacidosis(DKA)to enhance personalized treatment effects using machine learning techniques based on the United States Critical Care Medical Information Mart for Intensive Care-Ⅳ(MIMIC-Ⅳ).Methods Utilizing the MIMIC-Ⅳdatabase,the case data of 2096 patients with DKA admitted to the intensive care unit(ICU)at Beth Israel Deaconess Medical Center from 2008 to 2019 were analyzed.Machine learning models were developed,and receiver operator characteristic curve(ROC curve)and precision-recall curve(PR curve)were plotted to evaluate the model's effectiveness in predicting four common adverse outcomes:hypoglycemia,hypokalemia,reductions in Glasgow coma scale(GCS),and extended hospital stays.The risk of adverse outcomes was analyzed in relation to the rate of blood glucose decrease.Univariate and multivariate Logistic regression analyses were conducted to examine the relationship between relevant factors and the risk of hypokalemia.Personalized risk interpretation methods and predictive technologies were applied to individualize the analysis of optimal glucose control ranges for patients.Results The machine learning models demonstrated excellent performance in predicting adverse outcomes in patients with DKA,with areas under the ROC curve(AUROC)and 95%confidence interval(95%CI)for predicting hypoglycemia,hypokalemia,GCS score reduction,and extended hospital stays being 0.826(0.803-0.849),0.850(0.828-0.870),0.925(0.903-0.946),and 0.901(0.883-0.920),respectively.Analysis of the relationship between the rate of blood glucose reduction and the risk of four adverse outcomes showed that a maximum glucose reduction rate>6.26 mmol·L^(-1)·h^(-1) significantly increased the risk of hypoglycemia(P<0.001);a rate>2.72 mmol·L^(-1)·h^(-1) significantly elevated the risk of hypokalemia(P<0.001);a rate>5.53 mmol·L^(-1)·h^(-1) significantly reduced the risk of GCS score reduction(P<0.001);and a rate>8.03 mmol·L^(-1)·h^(-1) significantly shortened the length of hospital stay(P<0.001).Multivariate Logistic regression analysis indicated significant correlations between maximum bicarbonate levels,blood urea nitrogen levels,and total insulin doses with the risk of hypokalemia(all P<0.01).In terms of establishing personalized optimal treatment thresholds,assuming optimal glucose reduction thresholds for hypoglycemia,hypokalemia,GCS score reduction,and extended hospital stay were x_(1),x_(2),x_(3),x_(4),respectively,the recommended glucose reduction rates to minimize the risks of hypokalemia and hypoglycemia should be≤min{x_(1),x_(2)},while those to reduce GCS score decline and extended hospital stay should be≥max{x_(3),x_(4)}.When these ranges overlap,i.e.,max{x_(3),x_(4)}≤min{x_(1),x_(2)},this interval was the recommended optimal glucose reduction range.If there was no overlap between these ranges,i.e.,max{x_(3),x_(4)}>min{x_(1),x_(2)},the treatment strategy should be dynamically adjusted considering individual differences in the risk of various adverse outcomes.Conclusion The machine learning models shows good performance in predicting adverse outcomes in patients with DKA,assisting in personalized blood glucose management and holding important clinical application prospects.
作者
王瑞瑞
吴利娟
李惠先
李欣
Wang Ruirui;Wu Lijuan;Li Huixian;Li Xin(Guangdong Cardiovascular Institute,Department of Emergency Medicine,Guangdong Provincial People's Hospital(Guangdong Academy of Medical Sciences),Guangzhou 510080,Guangdong,China;Department of Emergency Medicine,Institute of Sciences in Emergency Medicine,Guangdong Provincial People's Hospital(Guangdong Academy of Medical Sciences),Guangzhou 510080,Guangdong,China;Medical Big Data Center,Guangdong Provincial People's Hospital(Guangdong Academy of Medical Sciences),Guangzhou 510080,Guangdong,China)
出处
《中华危重病急救医学》
CAS
CSCD
北大核心
2024年第6期635-642,共8页
Chinese Critical Care Medicine
基金
国家重点研发计划政府间重点项目(2023YFE0114300)。
