Application of machine learning for the diagnosis and prognosis of sepsis-induced acute respiratory distress syndrome: a systematic review and meta-analysis.

Background: Sepsis is a critical condition that can lead to Acute Respiratory Distress Syndrome (ARDS), resulting from both pulmonary and extrapulmonary etiologies. This systematic review evaluates the application of machine learning (ML) approaches for the early diagnosis and prognosis of ARDS in sepsis patients. Methods:This study is a systematic review and meta-analysis, focusing on studies that employ machine learning to forecast the incidence and outcomes of acute respiratory distress syndrome in adult sepsis patients. Databases including PubMed, Scopus, and Web of Science were queried for relevant studies. The quality of the eligible study was assessed, and the diagnostic accuracy data was retrieved. Results: A total of 11 studies were incorporated into the meta-analysis. The aggregated sensitivity for diagnostic models was 0.76(95% CI:0.75 to 0.77), whilst the aggregated specificity was 0.70(95% CI:0.69 to 0.71). Prognostic models demonstrated a combined sensitivity of 0.74(95% CI:0.72 to 0.75) and specificity of 0.72(95% CI:0.71 to 0.74). The areas under the curve for these models were 0.821(95% CI:0.781 to 0.861) and 0.793(95% CI:0.767 to 0.819). Conclusions: These findings indicate the potential utility of machine learning in improving the early diagnosis and prognosis of ARDS in septic patients. However, further study is required to substantiate these findings and to investigate the incorporation of machine learning methods in practical clinical environments. Trial Registration: The review protocol was registered and approved on the International Prospective Statistical Review Registry (PROSPERO) prior to the start of the study on 23rd June 2025(CRD420251079338).