New AI tech can spot deadly infections before they strike

By Stephen Beech

Sepsis can be predicted in sick children two days in advance by AI, allowing doctors chance to save their lives.

Artificial intelligence analyzed electronic health records to identify youngsters in the hospital who didn’t have the potentially deadly condition, but were likely to develop it within 48 hours.

Deployment of the state-of-the-art technology allowed doctors to begin life-saving pre-emptive care, say scientists.

Sepsis, resulting from the presence of harmful microorganisms in the blood or other tissues, is a leading cause of death in children worldwide.

Now, American researchers have developed and validated AI models that accurately identify children at high risk of sepsis within 48 hours.

They explained that the predictive models used routine electronic health record (EHR) data from the first four hours the child spent in the emergency department, before organs began to fail.

More from this section

Can dogs slow down aging in women?

New AI tech can spot deadly infections before they strike

Kids aren’t getting as much sleep as parents think

The study, published in JAMA Paediatrics, is the first to use AI models to predict sepsis in children.

Lead author Professor Elizabeth Alpern, of Northwestern University Feinberg School of Medicine in Chicago, said: “The predictive models we developed are a huge step toward precision medicine for sepsis in children.

"These models showed robust balance in identifying children in the emergency department who will later develop sepsis, without overidentifying those who are not at risk.

"This is very important because we want to avoid aggressive treatment for children who don’t need it.”

The study included five health systems contributing to the Pediatric Emergency Care Applied Research Network (PECARN), which provided Alpern and her colleagues access to a large set of data.

Children who already had sepsis on arrival at the hospital or who developed it within the first hours of care were excluded, focusing the goal of the study on predicting sepsis.

Alpern said: “We evaluated our models to ensure that there were no biases.”

She added: “Future research will need to combine electronic health record-based AI models with clinician judgment to make even better predictions.”