Key result
A machine learning model predicted in-hospital mortality for adult patients at admission with an AUC of 0.86 (95% CI, 0.83-0.90) in prospective validation.
Why the study?
Few machine learning models predicting in-hospital mortality are broadly applicable across a health system, and none have been prospectively evaluated and externally validated.
Does a machine learning model using electronic health record data accurately predict in-hospital mortality in adult patients at the time of admission?
Population
75 247 hospital admissions representing adult patients across three hospitals
Comparison
Machine learning model to predict in-hospital mortality at admission vs observed outcomes
Design
Prognostic study with retrospective and prospective validation cohorts
Authors
Loading...
May aid admission risk stratification using EHR data; leaves open whether implementation improves outcomes.
Observational (n=75,247)
Yes
Does a machine learning model using electronic health record data accurately predict in-hospital mortality in adult patients at the time of admission?
Effect estimate: AUC 0.86 (95% CI 0.83-0.90)
A machine learning model using commonly available electronic health record data at admission demonstrated good prospective and external validation for predicting in-hospital mortality.
Brajer et al. (2020) conducted an observational in In-hospital mortality (n=75,247). Machine learning model was evaluated on In-hospital mortality (AUC 0.86, 95% CI 0.83-0.90). A machine learning model predicted in-hospital mortality for adult patients at admission with an AUC of 0.86 (95% CI, 0.83-0.90) in prospective validation.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: