Assessment of Machine Learning vs Standard Prediction Rules for Predicting Hospital Readmissions
| Autor: | Stacy D. Garrett-Ray, Søren M. Bentzen, Daniel J. Morgan, Walt Ettinger, Paul Zimand, Bill Bame, Patrick Dooley, Yuanyuan Liang, Anthony D. Harris, Kerri A. Thom, J. Kathleen Tracy |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Male
MEDLINE Health Informatics Medicare Machine learning computer.software_genre Patient Readmission Risk Assessment Tertiary care Machine Learning Risk Factors Humans Medicine Hospital patients Original Investigation Maryland Receiver operating characteristic business.industry Research General Medicine Length of Stay Middle Aged Prognosis Readmission rate United States Community hospital Hospitalization Online Only Rank score Research Design Female Artificial intelligence business computer Medicaid |
| Zdroj: | JAMA Network Open |
| ISSN: | 2574-3805 |
| Popis: | This prognostic study compares standard readmission risk assessment scores with a machine learning score, the Baltimore score, for predicting 30-day unplanned hospital readmissions calculated in real time. Key Points Question Does a machine learning score better predict hospital readmissions than standard readmission scores? Findings In this prognostic study including 14 062 individual patients and 16 649 discharges from 3 hospitals, a machine learning score (the Baltimore score) was better able to predict hospital readmissions compared with standard readmission scores. Area under the receiver operating characteristic curve for the Baltimore score at discharge was 25.5% to 54.9% more efficient compared with commonly used scores with the difference between scores being statistically significant. Meaning The findings suggest that an individual hospital-tailored, machine learning–based score better predicts readmissions compared with standard methods. Importance Hospital readmissions are associated with patient harm and expense. Ways to prevent hospital readmissions have focused on identifying patients at greatest risk using prediction scores. Objective To identify the type of score that best predicts hospital readmissions. Design, Setting, and Participants This prognostic study included 14 062 consecutive adult hospital patients with 16 649 discharges from a tertiary care center, suburban community hospital, and urban critical access hospital in Maryland from September 1, 2016, through December 31, 2016. Patients not included as eligible discharges by the Centers for Medicare & Medicaid Services or the Chesapeake Regional Information System for Our Patients were excluded. A machine learning rank score, the Baltimore score (B score) developed using a machine learning technique, for each individual hospital using data from the 2 years before September 1, 2016, was compared with standard readmission risk assessment scores to predict 30-day unplanned readmissions. Main Outcomes and Measures The 30-day readmission rate evaluated using various readmission scores: B score, HOSPITAL score, modified LACE score, and Maxim/RightCare score. Results Of the 10 732 patients (5605 [52.2%] male; mean [SD] age, 54.56 [22.42] years) deemed to be eligible for the study, 1422 were readmitted. The area under the receiver operating characteristic curve (AUROC) for individual rules was 0.63 (95% CI, 0.61-0.65) for the HOSPITAL score, which was significantly lower than the 0.66 for modified LACE score (95% CI, 0.64-0.68; P |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|