No more than 15 authors can be listed.Please ensure the authors are listed in the right order.If there are multiple affiliations, please use the "+" button next to "Select Country/Region" to add additional affiliation fields.
Severe community-acquired pneumonia (sCAP) is associated with high mortality and healthcare burden. Early identification remains challenging, particularly in resource-limited settings where advanced diagnostics are not readily available. Clinical severity scores and artificial intelligence (AI) represent accessible and increasingly valuable tools in modern clinical practice.
We conducted a prospective study of 311 adults hospitalized with CAP at two tertiary hospitals in Vietnam (June 2025–January 2026). Clinical variables, Charlson Comorbidity Index (CCI), CURB-65, ROX index, neutrophil-to-lymphocyte ratio (NLR), and AI-assisted chest X-ray (deep learning-based abnormality detection, Qure.AI's chest X-ray) were evaluated. Logistic regression analyses were used to identify independent predictors of severe CAP, defined according to 2019 ATS/IDSA criteria.
Severe community-acquired pneumonia occurred in 30.8% (96/311) of patients. Independent factors included respiratory failure, altered mental status, elevated NLR, lower ROX index, higher CURB-65 score, and ≥3 radiographic abnormalities on chest X-ray (all p < 0.05). A prediction model incorporating CURB-65, AI-derived chest X-ray score, and ROX index demonstrated good discriminative performance for sCAP (AUC 0.819; 95% CI 0.773–0.865; p < 0.001).
A combined approach using simple clinical parameters, validated severity scores, and AI-assisted imaging provides a practical and available strategy for early identification of sCAP in resource-limited settings.