Establishment of a prediction model for extubation failure risk in ICU patients using bedside ultrasound technology

Background: Mechanical ventilation (MV) is crucial for managing critically ill patients; however, extubation failure, associated with adverse outcomes, continues to pose a significant challenge. Objective: The purpose of this prospective observational study was to develop and validate a predictive numerical model utilizing bedside ultrasound to forecast extubation outcomes in ICU patients. Methods: We enrolled 300 patients undergoing MV, from whom clinical variables, biomarkers, and ultrasound parameters were collected. Patients were randomly assigned to two groups at a 6:4 ratio: the derivation cohort (n = 180) and the validation cohort (n = 120). A nomogram prediction model was developed using significant predictors identified through multivariate analysis and its performance was assessed and validated by evaluating its discrimination, calibration, and clinical utility. Results: A total of 300 patients (mean age 72 years; 57.3 % male) were included, with an extubation failure rate of 26.7 %. The model, including diaphragm thickening fraction (OR: 0.890, P = 0.009), modified lung ultrasound score (OR: 1.371, P < 0.001), peak relaxation velocity (OR: 1.515, P = 0.015), and APACHE II (OR: 1.181, P = 0.006), demonstrated substantial discriminative capability, as indicated by an area under the receiver operating characteristic curve (AUC) of 0.886 (95 % CI: 0.830-0.942) for the derivation cohort and 0.846 (95 % CI: 0.827-0.945) for the validation cohort. Hosmer-Lemeshow tests yielded P-values of 0.224 and 0.212 for the derivation and validation cohorts. Conclusions: We have established a risk prediction model for extubation failure in mechanically ventilated ICU patients. This risk model base on bedside ultrasound parameters provides valuable insights for identifying high- risk patients and preventing extubation failure.