U of Maryland Medical Systems develops machine learning model to better predict readmissions

Baltimore-based University of Maryland Medical Systems has deployed an in-house machine learning model, called the Baltimore score, that aims to better predict which discharged patients are likely to be readmitted.

Daniel Morgan, MD, an associate professor of epidemiology and public health at Baltimore-based University of Maryland School of Medicine, led a study on the model's success. He and his team analyzed data from 14,000 patients from three UMMS hospitals.

The B score was compared against existing readmission risk-assessment tools, including the Lace index, Hospital score and Maxim/RightCare score. Between Sept. 1, 2014, and Aug. 31, 2016, Dr. Morgan evaluated more than 8,000 possible data variables from patients' EHRs to develop the B score. 

The final machine learning model includes 382 variables, including demographics, lab test results, if the patient required breathing assistance, body mass index, affiliation with a specific church, marital status, employment, medication usage and substance abuse.

Across the three hospitals, the B score was able to better identify patients at risk of readmission than the other scores, Dr. Morgan found. The score was most accurate for patients at highest risk.

Patients who scored in the top 10 percent of the B score at discharge had a 37.5 percent chance of 30-day unplanned readmission. A patient in the top 5 percent of the B score at discharge had a 43.1 percent chance of readmission.

"The widespread use of electronic health records has enhanced information flow from all clinicians involved in a patient's treatment," UMSOM Dean E. Albert Reece, MD, PhD, said. "This study underscores how patient data may also help solve the readmission puzzle and, ultimately improve the quality of patient care."

The research has been published in JAMA Network Open.

More articles about AI:
Novant Health opens Institute of Innovation & AI
8 new hospital applications for machine learning: AWS Textract, an algorithm predicting ICU outcomes & more
Facial recognition tech can be used to monitor ICU patient safety

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months