Answers are grounded in this paper and cite their sources.
Supplemental oxygen is among the most widely administered hospital therapies, yet it remains one of the least precisely controlled. Both hypoxemia and hyperoxemia are independently associated with adverse outcomes, and the decade of pragmatic oxygen-target trials has shifted attention from the numeric choice of target toward the fidelity with which any chosen target is sustained. Manual titration is intrinsically episodic and cannot close this fidelity gap, particularly outside high-intensity monitored environments. Automated closed-loop oxygen control has matured rapidly across four delivery paradigms—conventional oxygen flow (low-to-moderate flow up to 15–20 L/min via nasal cannula or mask), high-flow nasal cannula, invasive ventilation, and neonatal respiratory support—with consistent gains in time-in-target and reductions in extremes of oxygenation across all classes. This review surveys the contemporary device landscape (FreeO2, O2matic, iPOC, the Hamilton O2 assist controller evaluated in the HILOOP study, and the TSF high-flow controller, INTELLiVENT-ASV, and the neonatal CLiO2/OxyGenie/PRICO/SPOC/IntellO2/CLAC family), compares control architectures (proportional-integral-derivative, rule-based, and emerging machine-learning approaches), and synthesizes the efficacy, safety, workload, and economic evidence. We then use EMILY, a rule-based, network-independent, source-agnostic, closed-loop flow controller deployed in an intermediate respiratory care unit as a working exemplar to illustrate where the field is heading: multi-parameter inference, network-independent operation, portability across oxygen sources, and instrumented clinician override. We discuss its specific advantages and limitations in detail, and close with an agenda for the outcome trials, interoperability standards, and equity safeguards the field now requires.
No takes yet. Share an insight, caveat, or question.
Prieto et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: