The Pediatric Recruitment Engine: A Transparent Computational Model for Inspecting Assumptions During Positive End-Expiratory Pressure Titration

Positive end-expiratory pressure (PEEP) can reduce lung collapse, while higher pressures may cause overdistension and unfavorable dead space or pressure effects. The Pediatric Recruitment Engine (PRE) v0.7 is an open-source computational model for supervised teaching about these competing assumptions. It represents 2,560 heterogeneous virtual lung units in 16 conceptual regions. A fixed sequence of increasing and decreasing PEEP updates each unit's probability of remaining open and produces model-derived indices of recruitment, collapse, mechanics, shunt, oxygenation, dead space, carbon dioxide clearance, and regional balance. In one seeded scenario for a 2-year-old, 12-kg child under the moderate impairment preset, several balance and clearance indices selected 16 cmH2O, while the intentionally monotonic oxygenation surrogate selected the upper tested value of 24 cmH2O. Re-running the frozen code reproduced its archived numerical outputs, and 10 scripted execution and expected-behavior checks passed. These checks do not establish independent code verification or physiologic validity. A focused exploratory sensitivity analysis showed that some selected indices changed when illustrative model assumptions were varied. PRE uses no patient-level data, has not been calibrated or externally validated, and cannot recommend patient-specific PEEP. Its code and archived release are publicly available; the full mathematical specification and new exploratory analyses accompany this revision.