A machine learning algorithm has been developed that can detect which patients with Covid-19 may get worse and do not respond positively to turning on their stomach in intensive care units (ICUs).
This technique, known as proning, is often used in this environment to improve oxygenation of the lungs, but it is not suitable for all patients.
Researchers at Imperial College London gave the algorithm each patient’s data daily, rather than just on admission, so it could track their condition more accurately.
They believe the system can be used to improve guidelines in clinical practice in the future and be applied to potential future waves of pandemics and other diseases treated in similar clinical settings.
First author of the study Dr. Brijesh Patel said, “Most studies look at a patient’s health when admitted to the ICU and whether they were discharged or sadly died.
In the ICU, there is a tremendous amount of information that we use at the bedside to manage patients on a daily basis, and our research focuses on how the patient’s condition changed on a daily basis.
“This helped focus our attention on which specific parameters are most important and how the importance of each parameter changes over time.
This dynamic understanding is vital for understanding a new life-threatening disease and for knowing when and for whom each intervention is working.
The prone position is used in ICUs to help improve blood oxygenation in people with severe acute respiratory distress syndrome and has been widely used during the pandemic.
However, Proning did not help all Covid-19 patients and may – if used in patients who do not benefit Delaying the initiation of other treatments The findings show that the AI model identified factors that determined which patients were likely to get worse and did not respond to interventions such as prone position.
The researchers found that during the first wave of the pandemic, patients with blood clots or inflammation in the lungs, lower oxygen levels, lower blood pressure, and lower lactate levels benefited less from pronation.
Overall, continued oxygenation improved in only 44 percent of patients. The researchers analyzed data from 633 mechanically ventilated Covid-19 patients in 20 UK ICUs during the first wave of the outbreak that began last March.
They examined the importance of factors related to disease progression, such as blood clots and inflammation in the lungs, as well as the treatments given and whether the patient eventually died or was discharged.
They used this data, collected daily by legions of medical students, nurses, doctors, audit, research, and data personnel, to design and train the AI tool that then predicted factors that determine outcomes.
In November 2020, another group of researchers developed an AI that can detect signs of Covid-19 by looking at X-rays of a patient’s lungs.