Acute respiratory distress syndrome (ARDS) is a severe critical illness characterized by respiratory failure, regional lung collapse, and high mortality. Pediatric and neonatal ARDS poses additional challenges compared with adult ARDS due to developmental differences in lung structure, physiology, and treatment responses [1]. ARDS pathophysiology spans multiple spatial scales, involving organ-level physiological alterations, tissue-level alveolar collapse and reduced lung compliance, and molecular-level dysfunction of pulmonary surfactant at the alveolar air-liquid interface [2,3]. In this study, we investigate pediatric and neonatal ARDS across multiple spatial scales (shown in Figure 1). At the organ and organ-system levels, bedside clinical data are analysed using machine learning to assess ARDS risk, and predict disease severity, supporting clinical decision-making. At the tissue level, electrical impedance tomography (EIT) is applied to quantify regional lung collapse and compliance changes during positive end-expiratory pressure (PEEP) titration, enabling exploration of personalized ventilator settings. At the molecular scale, molecular dynamics simulations are used to examine previously unverified protein-lipid interactions within pulmonary surfactant. By integrating insights across scales, this work aims to advance the understanding of pediatric and neonatal ARDS pathophysiology, and provide a mechanistic foundation for improving surfactant replacement therapies in neonatal patients.