Geisinger, IBM build predictive tool to gauge sepsis risk

Geisinger has tapped IBM’s AI expertise and come up with a way to predict hospital patients’ risk of sepsis. In the process, the method can increase chances of survival in those who have the tricky and potentially life-threatening condition.

The integrated 13-hospital health system, which operates in Pennsylvania and New Jersey, worked with IBM’s data science and AI teams to train the predictive model on clinical data from thousands of de-identified sepsis patients spanning a decade, according to a Geisinger news release.

Geisinger says it hopes to leverage the new model to develop more personalized clinical care plans for at-risk sepsis patients.

Researchers testing the model used open-source tools from IBM Watson Studio to predict patient mortality during the hospitalization period or during the 90 days following their hospital stay.

“The model helped researchers identify clinical biomarkers associated with higher rates of mortality from sepsis by predicting death or survival of patients in the test data,” Geisinger reports.

The project showed numerous risk factors can combine to increase a patient’s chances of getting sepsis. These included age, prior cancer diagnosis, decreased blood pressure, number of hospital transfers and time spent on blood-pressure drugs, as well as the type of the culprit pathogen.

Geisinger notes that sepsis has historically proven hard to catch early. It hits about 1.7 million American adults, contributing to the deaths of more than a quarter-million people a year.