背景：膈肌功能障碍是指多种病因导致的单侧或双侧膈肌暂时或永久性的无力或瘫痪，通常以膈肌收缩功能下降为特点，伴有肌纤维损伤和肌纤维萎缩。临床上ICU患者中膈肌功能障碍发生率较高，而膈肌功能障碍的存在直接影响患者脱机拔管的失败率。目前临床的医务人员聚焦于通过膈肌功能障碍预测患者的脱机拔管失败率，忽视了早期识别膈肌功能障碍发生的重要性。早期识别患者膈肌功能障碍的发生可以通过构建风险预测模型实现，构建模型的第一步是明确可能存在的影响因素，现有研究对于膈肌功能障碍危险因素的探索不够全面。鉴于此，有必要通过文献回顾和专家函询构建全面的ICU患者膈肌功能障碍发生风险评价指标，进而通过前瞻性研究探索膈肌功能障碍的预测因子，并运用逻辑回归的方式构建风险预测模型，辅助医疗及护理人员在临床工作中精准识别高危患者。

目的：构建全面的ICU患者膈肌功能障碍发生风险评价指标，开展前瞻性研究确定膈肌功能障碍预测因子，构建ICU患者膈肌功能障碍风险预测模型并进行内部验证。

方法：（1）通过文献回顾、头脑风暴及德尔菲专家函询法及层次分析法构建ICU患者膈肌功能障碍发生风险评价指标。（2）采用前瞻性研究方法，对2022年12月1日至2023年12月31日在北京市某三甲医院ICU住院的患者进行研究，分别在患者ICU住院的第1天、第3天、第5天时使用膈肌超声的方式评估患者膈肌功能障碍的发生情况，采用第一阶段构建的ICU患者膈肌功能障碍发生风险评价指标收集：患者一般资料、肌肉功能指标、治疗方式、实验室指标、药物使用、器官功能状态、呼吸功能指标，通过对数据进行单因素分析及多因素二元Logistic回归分析，明确膈肌功能障碍发生现状及影响因素。（3）采用logistic回归分析法分析ICU患者膈肌功能障碍发生的预测因子，构建ICU患者膈肌功能障碍风险预测模型并进行内部验证，通过区分度和校准度评价所得模型的预测效应。

结果：本研究通过两轮专家函询构建了包括7个一级指标、34个二级指标、34个三级指标的全面的ICU患者膈肌功能障碍发生风险评价指标。基于此，本研究共纳入221例患者，最终进入模型开发的共计221例患者，ICU住院期间共126例患者发生膈肌功能障碍，发病率为57%。多因素分析显示：心脏手术（OR=2.812 ，95%Cl= 1.046-7.887）、血清钙离子浓度（OR=4.687，95%Cl=1.574-15.69）、CRRT治疗（OR=7.263，95%Cl=0.921-102.0）、俯卧位治疗（OR=7.766，95%Cl=1.563-61.35）、机械通气（OR=2.902，95%Cl=0.801-11.37）、胰岛素使用（OR=8.313，95%Cl=2.527-33.30）、脓毒症（OR=2.304，95%Cl=0.848-6.644）是ICU患者发生膈肌功能障碍的危险因素，基于logistic回归分析构建的风险预测模型列线图预测效果良好（AUC=0.886，Acc=0.812），该列线图预测模型的校准曲线与实际曲线基本吻合；Hosmer-Lemeshow 拟合优度检验结果显示X-squared = 4.4174, df = 8, p-value = 0.8176＞0.05，且Brier=0.136＜0.25，说明该列线图预测模型拟合效果好。

结论：本研究构建的ICU患者膈肌功能障碍发生风险评价指标科学、全面，基于真实数据明确了膈肌功能障碍的发生现状和影响因素，运用logistic回归分析构建的膈肌功能障碍风险预测模型列线图具有较高的可复制性、普适性和可行性，能够有效识别膈肌功能障碍高危人群。该模型在进行临床运用前，仍需要进行外部验证研究。

Background: Diaphragm dysfunction refers to the temporary or permanent weakness or paralysis of the unilateral or bilateral diaphragm caused by a variety of causes, usually characterized by the decreased contractile function of the diaphragm, accompanied by muscle fiber damage and muscle fiber atrophy. The incidence of diaphragm dysfunction is higher in clinical ICU patients, and the presence of diaphragm dysfunction directly affects the failure rate of weaning extubation. Current clinical medical staff focus on predicting the failure rate of weaning extubation in patients through diaphragm dysfunction, ignoring the importance of early identification of the occurrence of diaphragm dysfunction. Early identification of patients with the occurrence of diaphragm dysfunction can be achieved through the construction of risk prediction model, the first step of the model construction is to clarify the possible influencing factors, the existing research for the exploration of the risk factors of diaphragm dysfunction is not comprehensive enough. In view of this, it is necessary to through the literature review and expert enquiry to build comprehensive ICU patients with diaphragm dysfunction risk evaluation index, and through prospective studies to explore the predictor of diaphragm dysfunction, and use the way of logical regression to build risk prediction model, auxiliary medical and nursing staff accurately identify high-risk patients in clinical work.

Objective: A comprehensive risk evaluation index for the occurrence of diaphragmatic dysfunction in ICU patients was constructed, prospective studies were conducted to determine the predictors of diaphragmatic dysfunction, and the risk prediction model of diaphragmatic dysfunction in ICU patients was constructed for internal validation.

Method: (1) The risk evaluation index of diaphragm dysfunction in ICU patients was constructed through literature review, brainstorming, Delphi expert letter inquiry and hierarchical analysis. (2) by using a prospective study approach, A study of patients admitted to the ICU of a Grade A hospital in Beijing from December 1,2022 to December 31,2023, The occurrence of diaphragm dysfunction was evaluated on day 1,3 and 5 days of ICU admission, Risk evaluation indicators of diaphragmatic dysfunction in ICU patients constructed in the first stage: general patient data, muscle function indicators, treatment methods, laboratory indicators, drug use, organ function status, respiratory function indicators, Through univariate analysis of the data and multivariate binary Logistic regression analysis, Clarify the occurrence status and influencing factors of diaphragm dysfunction. (3) Use logistic regression analysis to analyze the predictors of diaphragm dysfunction in ICU patients, construct the risk prediction model of diaphragm dysfunction in ICU patients and conduct internal verification, and evaluate the prediction effect of the model by differentiation and calibration.

Results: In this study, a comprehensive risk evaluation index of diaphragm dysfunction in ICU patients including 7 primary indicators, 34 secondary indicators and 34 tertiary indicators was constructed through two rounds of expert eninquiry. Based on this, 221 patients were included in this study, and a total of 221 patients finally entered the model development. A total of 126 patients developed diaphragm dysfunction during ICU hospitalization, with an incidence of 57%. Multivariate analysis showed that: cardiac surgery (OR = 2.812, 95% Cl = 1.046-7.887), serum calcium ion concentration (OR = 4.687, 95%Cl=1.574-15.69), CRRT treatment (OR = 7.263, 95%Cl=0.921-102.0), treatment in prone position (OR = 7.766, 95%Cl=1.563-61.35), mechanical ventilation (OR = 2.902, 95%Cl=0.801-11.37), use of insulin (OR = 8.313, 95%Cl=2.527-33.30), sepsis (OR = 2.304, 95%Cl=0.848-6.644) is a risk factor for developing diaphragmatic dysfunction in ICU patients, The nomogram of the risk prediction model constructed based on the logistic regression analysis had a good prediction effect (AUC = 0.886, Acc=0.812), The calibration curve of the nomogram prediction model is basically consistent with the actual curve; The results of the Hosmer-Lemeshow goodness-of-fit test showed that X-squared = 4.4174, df = 8, p-value = 0.8176>0.05, And Brier = 0.136 <0.25, This shows that the nomogram predicted model fits well.

Conclusion: This study of ICU patients of diaphragm dysfunction risk evaluation index scientific, comprehensive, based on the real data clear the occurrence of diaphragm dysfunction and influencing factors, diaphragm dysfunction risk prediction model constructed using logistic regression analysis with high replicability, universality and feasibility, and can effectively identify the high-risk groups of diaphragm dysfunction. This model still needs an external validation study before clinical application.