AI predicts early onset of sepsis

Researchers from Emory University in Atlanta have developed an artificial intelligence (AI) algorithm capable of predicting the onset of sepsis, according to a study published in Critical Care Medicine.

Researchers hope their algorithm will detect the onset of sepsis—a leading cause of morbidity and mortality—while it can still be treated with antibiotics. In this study, researchers tested the accuracy of the Artificial Intelligence Sepsis Expert algorithm in the prediction of early onset sepsis.

Data was collected from more than 31,000 admissions to Emory University hospital intensive care units as well as 52,000 ICU patients from a public database. After exclusion, data on vital signs and electronic health records from 27,000 patient were used to train the algorithm. In testing the algorithms accuracy, 42,000 patients were evaluated by a set of 65 measures to predict the onset of sepsis within 12, eight, six or four hours.

Results showed the Artificial Intelligence Sepsis Expert achieved predictions at a range of 0.83 to 0.85.

“Using data available in the ICU in real-time, Artificial Intelligence Sepsis Expert can accurately predict the onset of sepsis in an ICU patient 4-12 hours prior to clinical recognition,” concluded first author Shamim Nemati and colleagues. “A prospective study is necessary to determine the clinical utility of the proposed sepsis prediction model.”