Short-Term Prognosis in Acute Asthma Exacerbations: A Comparative Evaluation of Machine Learning Models Using...

Acute asthma exacerbations are a common cause of emergency department visits and require rapid risk stratification to guide disposition decisions. Although spirometry is the primary objective method used to assess clinical severity, it may not reliably predict clinical outcomes. In this context, machine learning-based approaches have gained increasing attention for improving prognostic assessment in emergency settings. To evaluate the prognostic relevance of spirometric parameters, tracheal respiratory sound-derived acoustic features, and machine learning-based classification models for short-term outcomes in adults presenting to the emergency department with acute asthma exacerbations.

In this prospective cohort study, adults with acute asthma exacerbations underwent spirometry and tracheal respiratory sound recording before and after emergency department treatment. Short-term prognosis was defined as hospitalization during the index visit or emergency department re-presentation within seven days. Changes in spirometric and acoustic features were analyzed, and machine learning models incorporating demographic, spirometric, acoustic, and combined feature sets were developed and compared.

Baseline airflow limitation at presentation was the strongest determinant of short-term prognosis. Patients with poor outcomes had significantly lower pre-treatment and post-treatment FEV₁ and PEF values. Acoustic feature changes showed minimal correlation with ΔFEV₁ and ΔPEF; however, selected frequency-domain features differed between prognosis groups, indicating complementary physiological information. Machine learning models incorporating spirometric variables achieved the highest performance, with accuracy up to 84.4% and balanced classification metrics. Sound-based models demonstrated moderate but clinically meaningful performance (accuracy approximately 67-77%), while multimodal models did not outperform spirometry-only models.

In acute asthma exacerbations, baseline spirometric impairment remains the most reliable predictor of short-term outcomes. Tracheal respiratory sound analysis combined with machine learning may provide complementary prognostic information, particularly when spirometry is unavailable or unreliable, supporting its role as a clinical decision-support tool rather than a replacement for conventional assessment.