A university hospital in Spain has become the first to deploy an artificial intelligence (AI)-based oxygen therapy system designed to automatically adjust oxygen delivery to individual patients in real time, according to its developers.
©Fundación Jiménez Díaz University Hospital/Emily.AI
The Emily.AI platform has been developed through a collaboration between Madrid’s Fundación Jiménez Díaz University Hospital and a Spanish healthcare technology company named after the technology.
The system continuously monitors patient data and automatically adjusts oxygen flow to maintain target oxygen saturation levels. Developers say the technology is intended to reduce the burden of manual oxygen titration on healthcare staff while helping patients remain within prescribed oxygen ranges.
More than 50 patients have received oxygen therapy using the system within the hospital’s Intermediate Respiratory Care Unit (IRCU), covering patients breathing spontaneously as well as those receiving invasive and non-invasive mechanical ventilation.
According to the hospital, the device has operated continuously for periods of up to 72 hours without incident.
“Oxygen is an essential medicine, yet oxygen therapy still relies heavily on manual adjustments,” said Dr Sara Heili, Associate Head of Pulmonology and Head of the hospital’s IRCU. The technology aims to maintain patients within target oxygen saturation ranges while reducing episodes of hyperoxia.
The project originated during the Covid-19 pandemic, when founder Daniel Oliva began exploring whether software and AI could automate the frequent oxygen adjustments typically performed by healthcare professionals.
The resulting system combines medical hardware, software and AI algorithms in a closed-loop design that continuously monitors patient status and adjusts oxygen delivery accordingly.
While the technology has been deployed clinically at Fundación Jiménez Díaz, it remains in the development and validation stage.
The company is currently pursuing ISO certification requirements needed for CE marking and estimates the platform could be ready for commercialisation within approximately two years.
Future plans include validating the system in conventional hospital wards, expanding deployment to additional healthcare centres, and exploring use in home care and remote monitoring settings.
Developers also intend to incorporate predictive capabilities that could identify patient deterioration before oxygen saturation levels begin to fall.
Why it matters
The project reflects a broader trend towards the application of AI and automation in respiratory care. Oxygen remains one of the most widely used medical gases globally, yet its administration often relies on manual adjustment by clinicians and nursing staff.
If successfully commercialised and adopted more widely, closed-loop oxygen delivery systems could help improve treatment consistency while reducing staff workload in hospitals facing increasing resource pressures.
