U.K. Scientists Tap AI for Better Ventilation for Covid Patients

(Bloomberg) --

A trial is underway in the U.K. exploring how artificial intelligence can be used to help monitor the lung physiology of Covid-19 patients to optimize ventilation treatments.

Patients with severe Covid-19 may develop a condition known as Acute Respiratory Distress Syndrome, which prevents the lungs from providing the body with enough oxygen. The condition is life-threatening and often requires mechanical ventilation in intensive care. Selecting the wrong ventilator setting can damage the patient’s lungs further, which is why researchers at Imperial College London are studying how AI can help personalize ventilator settings to individual patients.

The system could be helpful to the National Health Service’s response to a second wave of Covid-19 during the winter months when demand increases, Brijesh Patel, clinical senior lecturer at Imperial College London, said in an interview.

“In winter, we really don’t know whether we’re going to see pure Covid-19 patients come through hospitals as we have recently, or whether it’ll be Covid-19 patients with bacterial infection on top of that, or flu patients,” he said. “So our interest is in making sure we find a way to set the ventilators appropriately regardless of the cause.”

The study is testing a new AI device called the Beacon Care system, which uses mathematical calculations to recommend ventilator settings more suited to an individual’s needs. Clinicians input data on a patient’s condition and the AI device advises health-care staff on how to treat the patient.

At the same time, the system measures how the lung is aerated and the way blood flow goes through and captures oxygen, which can allow for better understanding of treatment responses.

Patel’s ambition is for the data gathered to also enable the team to detect the trajectory of patients and how their condition might evolve. “It might predict whether patients are going to get worse or respond to different maneuvers we apply to them so that we can do more targeted interventions,” he said.

©2020 Bloomberg L.P.