【作者】 杨怡；

【导师】 张黎；

【作者基本信息】 重庆医科大学 ， 护理（专业学位）， 2025， 硕士

【摘要】 目的:探索老年患者医院获得性压力性损伤(Hospital-acquired pressure injury,HAPI)的风险因素,构建并验证老年患者HAPI风险预测模型,为临床准确筛查老年HAPI高危患者提供可靠工具,以预防及减少老年患者HAPI的发生。方法:1.通过文献回顾法和专家咨询法,收集并整理老年患者HAPI相关影响因素,并制成《老年患者医院获得性压力性损伤风险因素信息提取表》作为数据采集工具。2.收集重庆医科大学医学数据研究院大数据平台2016年1月1日至2022年12月31日符合纳排标准的HAPI老年患者住院电子病历相关信息,以1:4的比例随机抽取同期同科室非HAPI老年患者作为对照,形成建模组数据集,按照7:3比例随机划分为训练集和验证集,分别构建逻辑回归(Logistic regression,LR)、随机森林(Random forest,RF)、朴素贝叶斯(Naive bayes,NB)和极端梯度提升(Extreme gradient boosting,XGBoost)四种老年患者HAPI风险预测模型;使用特异性(Sensitivity)、灵敏度(Specificity)、阳性预测值(Positive predictive value,PPV)、阴性预测值(Negative predictive value,NPV)、几何平均数(Geometric mean,G-mean)、Brier评分(Brier score)、受试者工作特征曲线下面积(Area under the receiver operating characteristic curve,AUROC)和精确率-召回率曲线下面积(Area under the precision-recall curve,AUPRC)对模型性能进行评估。3.收集重庆市人民医院的2016年1月1日至2022年12月31日老年患者住院电子病历HAPI相关信息,形成外部验证组数据集对模型进行外部验证,以评价模型的可靠性和实用性。结果:1.本次研究共收集3020例老年患者,包括建模组2445例和外部验证组575例。发生HAPI的老年患者共604例,包括建模组489例和外部验证组115例。总样本包含男性1644例,女性1376例,患者年龄中位数及四分位数间距为75(67～82)岁,住院时长中位数及四分位数间距为10(6～17)天。2.皮肤潮湿、移动能力、冠心病、住院时长、意识障碍、下肢深静脉血栓、日常食物获取、血钾、白细胞、水肿、吸烟史、摩擦力/剪切力、高血压、白蛋白、感知觉、身体活动度、糖尿病这17项因素是老年患者HAPI的独立风险因素(P<0.05)。3.RF模型在四种算法构建的老年患者HAPI模型中具有最佳的预测性能:内部验证中,其特异性为0.93,灵敏度为0.83,阳性预测值为0.75,阴性预测值为0.96,G-mean值为0.88,Brier score值为0.07,AUROC值为0.96,AUPRC值为0.91;外部验证中,其特异性为0.83,灵敏度为0.88,阳性预测值为0.57,阴性预测值为0.97,G-mean值为0.79,Brier score值为0.12,AUROC值为0.88,AUPRC值为0.55。结论:1.老年患者HAPI风险因素复杂多样,皮肤潮湿、移动能力、冠心病、住院时长、意识障碍、下肢深静脉血栓、日常食物获取、血钾、白细胞、水肿、吸烟史、摩擦力/剪切力、高血压、白蛋白、感知觉、身体活动度、糖尿病等17项风险因素对老年患者HAPI有重要预测价值。2.利用真实临床数据,基于RF算法构建的老年患者HAPI模型兼具指标易获取性与预测高准确性的优势,有助于临床护理人员精准识别老年HAPI高危患者,进而制定个体化预防措施,有效减少老年患者HAPI的发生,实现护理资源合理分配与患者安全质量提升的双重目标。更多还原

【Abstract】 Objective:To investigate risk factors for hospital-acquired pressure injury(HAPI)in elderly patients,develop and validate a risk prediction model for HAPI in this population,and provide a clinically reliable tool for early identification of high-risk elderly patients,thereby enabling targeted preventive interventions to reduce the incidence of HAPI in geriatric care settings.Methods:1.By employing a literature review and expert consultation approach,we compiled relevant influencing factors for HAPI in elderly patients and developed an information extraction form to serve as a data collection instrument.2.A retrospective study was used to collect related information from the inpatient electronic medical records of elderly patients with HAPI who met the inclusion criteria and the exclusion criteria from January 1,2016,to December 31,2022,through the big data platform of the Institute of Medical Data of Chongqing Medical University,and randomly select non-HAPI elderly patients from the same department during the same period of time as a control in a ratio of 1:4 to form a modeling group dataset.The modeling group dataset was randomly divided into a training set and a validation set in a 7:3 ratio to construct four risk prediction models for HAPI in elderly patients using Logistic Regression(LR),Random Forest(RF),Naive Bayes(NB),and Extreme Gradient Boosting(XGBoost)algorithms.Model performance was assessed using sensitivity,specificity,positive predictive value(PPV),negative predictive value(NPV),geometric mean(G-mean),Brier score,area under the receiver operating characteristic curve(AUROC),and area under the precision-recall curve(AUPRC).3.According to the same collection method of the modeling group dataset,the related information about the inpatient electronic medical records of elderly patients in Chongqing People’s Hospital from January 1,2016 to December 31,2022 was retrospectively collected to form the model external validation group dataset to evaluate the reliability and clinic practical utility of the model.Results:1.In this study,a total of 3,020 elderly patients were enrolled,including2,445 patients in the modeling group and 575 patients in the external validation group.A total of 604 elderly patients developed HAPI,among which 489 cases were in the modeling group and 115 cases were in the external validation group.The total sample consisted of 1,644 male patients and 1,376 female patients.The median age and interquartile range of the patients were 75(67-82)years old,and the median length of hospitalization and interquartile range of patients were 10(6-17)days.2.The 17 influencing factors,including moist skin,mobility,coronary heart disease,length of hospitalization,consciousness,deep vein thrombosis,daily food intake,serum potassium,white blood cells,edema,smoking history,friction and shear,hypertension,albumin,sensation,physical activity and diabetes mellitus were identified as risk factors for HAPI in elderly patients(P<0.05).3.The RF model demonstrated superior predictive performance among the four algorithms constructed for modeling HAPI in elderly patients:in the internal validation,the specificity was 0.93,the sensitivity was 0.83,the PPV was 0.75,the NPV was 0.96,the Brier score was 0.07,the G-mean was 0.88,the AUROC was 0.96 and the AUPRC was 0.91;in the external validation,the specificity was 0.83,the sensitivity was 0.88,the PPV was 0.57,the NPV was 0.97,the Brier score was 0.12,the G-mean was 0.79,the AUROC was0.88 and the AUPRC was 0.55.Conclusion:1.The risk factors for HAPI in elderly patients are complex and diverse.Seventeen risk factors,including moist skin,mobility,coronary heart disease,length of hospitalization,consciousness,deep vein thrombosis,daily food intake,serum potassium,white blood cells,edema,smoking history,friction and shear,hypertension,albumin,sensation,physical activity and diabetes mellitus,have important predictive values for HAPI in elderly patients.2.Utilizing real clinical data,the model for HAPI in elderly patients constructed based on the RF algorithm possesses the advantages of easy accessibility of indicators and high prediction accuracy.This model is conducive for clinical nursing staff to accurately identify elderly patients at high risk of HAPI,and subsequently formulate individualized preventive measures.These measures can effectively reduce the occurrence of HAPI in elderly patients,achieving the dual goals of rational allocation of nursing resources and enhancement of patient safety and healthcare quality.更多还原