Machine Learning Prediction of Mortality After Prone Positioning in ARDS

A Machine Learning Model to Predict Mortality in Patients With Acute Respiratory Distress Syndrome After Prone Positioning

Acute respiratory distress syndrome (ARDS) is a life-threatening condition with high mortality. Prone position ventilation (PPV) is an evidence-based therapy that improves oxygenation and survival in patients with moderate to severe ARDS; however, outcomes remain heterogeneous. Early identification of patients at high risk of mortality after PPV may improve clinical decision-making and individualized management.

This retrospective observational study developed and validated a machine learning model to predict intensive care unit (ICU) mortality in patients with ARDS receiving prone position ventilation. Clinical, laboratory, and treatment variables obtained from ICU electronic medical records were used to construct prediction models using multiple machine learning algorithms. The performance of these models was evaluated and compared to identify the optimal model for mortality prediction.

Inclusion Criteria:

• Diagnosis of ARDS according to the Berlin definition;
• Receipt of at least one session of prone position ventilation (PPV) during hospitalization;
• Requirement for mechanical ventilation.

Exclusion Criteria:

• Age <18 years;
• PPV duration <6 hours;
• ICU length of stay <2 days;
• Pregnancy;
• Receipt of awake prone positioning.
• Missing key clinical data.