Construction of a Right Ventricular Function Assessment Model in Patients Undergoing Invasive Mechanical Ventilation Based on VExUS Grading and the Classification and Regression Tree Algorithm

Gao JunPan GaoYuanyuan DuZhang Jianjun^*Jinrong Wang^*

• Harrison International Peace Hospital, Hengshui, China

Objective: Investigate the correlation between right ventricular function ultrasound indicators and the Venous Excess Ultrasound (VExUS) grading system in patients undergoing invasive mechanical ventilation (IMV), and develop a right ventricular function assessment model using VExUS grading and the Classification and Regression Tree (CART) algorithm. Methods: This cross-sectional study collected data from patients in the Department of Critical Care at Harrison International Peace Hospital who underwent tracheal intubation within 24–48 hours from October 2023 to December 2024. Collected information comprised general clinical data, hematological indices, hemodynamic parameters, severity scores, right ventricular function ultrasound indicators, the maximum diameter of the inferior vena cava (IVCdmax), and Doppler spectra from the hepatic, portal and intrarenal veins. Patients were divided into two groups（VExUS 0 and VExUS ≥1）based on VExUS grading system to reflect the presence or absence of venous congestion. Baseline data were compared between groups, and Spearman’s correlation analysis was performed to assess the relationship between VExUS grading and right ventricular function ultrasound indicators. Subsequently, a right ventricular function assessment model was constructed using the CART algorithm. Results: A total of 80 patients were enrolled, with 26 in the VExUS 0 group and 54 in the VExUS ≥1 group. In the VExUS ≥1 group, the proportion of male patients, levels of NT-proBNP, cTnI, central venous pressure, cumulative fluid balance, and the usage rate of diuretics were significantly higher than in the VExUS 0 group (P < 0.05). IVCdmax was positively correlated with VExUS grading (r=0.773, P < 0.05), whereas tricuspid annular plane systolic excursion (TAPSE) and the tricuspid E-wave peak were negatively correlated (r= –0.670 and –0.648, respectively; P < 0.05). In the CART model analysis, the importance of the variables was ranked as: IVCdmax > TAPSE > tricuspid E-wave peak. When validated with a test set, the model achieved an overall accuracy of 87.5%. Conclusion: In patients undergoing IMV, IVCdmax, TAPSE, and the tricuspid E-wave peak show strong correlations with VExUS grading. A right ventricular function assessment model based on VExUS grading and the CART algorithm can effectively evaluate right ventricular performance and may serve as a useful tool in identifying venous congestion.