Objective:

Assess whether autonomous oxygen titration can improve oxygenation control in hospitalized adults receiving low-flow supplemental oxygen, while also exploring oxygen use and workflow considerations.

Approach:

• Trial design: The SAVE-O2 AI trial randomized 300 adults at four US hospitals to closed-loop autonomous titration with the O2matic PRO100 or standard manual titration.
• Participants and intervention: Participants received low-flow oxygen (1–10 L/min). The device used pulse oximetry readings to adjust oxygen delivery toward a clinician-selected SpO2 target. Patients requiring high-flow oxygen or ventilation were excluded.

Key Findings:

• Participants in the autonomous-titration group were within their target normoxemia range 85% of the time, compared with 63% under usual care.
• Automated titration significantly reduced the duration of both hypoxemia and hyperoxemia.
• Four participants experienced mild events, including epistaxis or discomfort.
• Staff feedback was not systematically collected; reported anecdotal responses were generally positive.

Interpretation:

The trial found improved oxygenation control with autonomous titration. Whether it reduces mortality or long-term complications remains undetermined. Reduced clinician workload and oxygen consumption are potential benefits discussed in the article, but were not established as trial outcomes.

Limitations:

• The study excluded patients requiring high-flow nasal oxygen or ventilation.
• The trial did not establish effects on mortality or long-term clinical complications.
• Staff feedback was anecdotal rather than systematically collected.
• The system depends on reliable pulse oximetry; poor perfusion, motion artifacts, or sensor detachment may produce faulty readings.
• Responsibility for failures involving autonomous titration remains unresolved.

Conclusion:

Further research and resolution of safety and accountability issues are needed before broader implementation. The article also describes possible applications in home oxygen, medical transport, and resource-limited settings.