Machine-Learning Models Show Promise at Identifying Sepsis in NICU

March 4, 2019
CHOP researchers developing a machine-learning model able to recognize sepsis at least four hours before clinical suspicion

Machine-learning models have shown promise in identifying which sick infants in a neonatal intensive care unit (NICU) have sepsis hours before clinicians recognize the life-threatening condition, according to a recent study at the Children’s Hospital of Philadelphia (CHOP).

A team of data researchers and physician-scientists tested machine-learning models in a NICU population, drawing only on routinely collected data available in electronic health records (EHRs).

The study aimed to develop a machine-learning model able to recognize sepsis in NICU infants at least four hours before clinical suspicion. "To our knowledge, this was the first study to investigate machine learning to identify sepsis before clinical recognition using only routinely collected EHR data," says Aaron Masino, Ph.D., who led the study team's machine-learning efforts, in a prepared statement. Masino is an assistant professor in the Department of Anesthesiology and Critical Care Medicine and a member of the Department of Biomedical and Health Informatics at CHOP.

"Because early detection and rapid intervention is essential in cases of sepsis, machine-learning tools like this offer the potential to improve clinical outcomes in these infants," adds Masino. "Follow-up clinical studies will allow researchers to evaluate how well such systems perform in a hospital setting."

The study team evaluated how well eight machine-learning models were able to analyze patient data to predict which infants had sepsis. Because the data came from a retrospective sample of NICU infants, the researchers were able to compare each model's predictions to subsequent findings--whether or not an individual patient was found to develop sepsis.

The study team drew on EHR data from 618 infants in the CHOP NICU from 2014 to 2017. Many of the infants in the patient registry were premature; the cohort had a median gestational age of 34 weeks. Co-occurring conditions included chronic lung disease, congenital heart disease, necrotizing enterocolitis (a severe intestinal infection) and surgical conditions. 

Six of the eight models performed well in accurately predicting sepsis up to four hours before clinical recognition of the condition, the researchers said.

The team's findings are a preliminary step toward developing a real-time clinical tool for hospital practice, Masino says. The researchers plan to do further research to refine their models and investigate the software tools in a carefully designed prospective clinical study. "If research validates some of these models, we may develop a tool to support clinical decisions and improve outcomes in critically ill infants," he adds.

The research team published its findings in the retrospective case-control study Feb. 22 in PLOS ONE.

Sponsored Recommendations

2024's Healthcare Buyer Journey: New Research and Insights

Join us on April 30th for a webinar unveiling insights from the latest study on the Healthcare IT Buying Journey! Discover evolving challenges, effective health data management...

Improving care with AI-powered solutions

Don't miss our April 23rd webinar delving into the transformative impact of AI-powered solutions on healthcare. Join industry leaders Reid Conant and Dr. Patrick McGill as they...

Shield your health system against cyber threats

You won't want to miss out on this imperative April 4th webinar about how you can protect your healthcare organization. Join us to learn how to fortify your health system against...

Healthcare Trends 2024: Trends & Strategies for Future Success

Explore the future of healthcare in 2024 with insights from the Healthcare Industry Trends Report. Stay ahead of the curve as we delve into the latest industry developments and...