A Viable Oxygenation-Ultrasound Assessment Paradigm for the Rapid Identification of Severe Blast Lung Injury in the Early Stage

Background To develop and verify a viable assessment paradigm that integrates lung ultrasound and oxygenation parameters for the early detection of severe blast lung injury (BLI), particularly useful in resource-constrained situations. Methods Forty-two Bama miniature pigs wounded after being exposed to controlled shock waves (22) or TNT explosions (20) were divided into severe and non-severe BLI groups in this retrospective study. Arterial partial pressure of oxygen (PaO ₂ ), pulse oximetry saturation (SpO ₂ ), and lung ultrasound scores (LUS) were recorded pre-injury and at 1-3h and 3-6h post-injury. For assessing group differences across periods, a repeated - measures ANOVA was conducted. Predictive models for BLI severity were developed using logistic regression, and their performance was systematically evaluated. The optimal threshold of the models was identified using the maximum Youden value. Results A non-invasive predictive model that combined SpO2 and LUS always showed the highest prediction performance. Meanwhile, an Area Under the Curve (AUC) of 0.92 (95% CI: 0.81–0.99) at 1-3h and 0.93 (95% CI: 0.82-1.00) at 3-6h were achieved. The SpO2-LUS model significantly improved discriminative power over the best single-indicator model, and also had excellent calibration (p > 0.05) as well as superior net clinical benefit in Decision Curve Analysis compared to other models. Conclusions The combined SpO2-LUS model, which provides a non-invasive and precise tool for the early identification of severe BLI, is of great for optimizing triage in pre-hospital and resource-limited contexts.