Development and evaluation of a machine learning prediction model for short-term mortality in patients with diabetes or hyperglycemia at emergency department admission
Authors' conclusion
A machine learning model based on routinely collected data in the ED accurately predicted 30-day mortality with high specificity and sensitivity. This approach shows promise in identifying high-risk patients requiring close monitoring and timely interventions.
Study context
- Population
- Patients aged > 18 years with diabetes or hyperglycemia (glucose > 11 mmol/L) at all nine emergency department (EDs) in Region Skåne, 2017 to 2018
- Participants
- 34,280 patients74,611 ED visits
- Exposure
- Men 40-69 vs Men ≥ 70 vs Women 40-69 vs Women ≥ 70
- Outcome
- 30-day mortality prediction
- Follow-up
- 30 days
Methodological detailsprognostic model development and validation, machine learning, stochastic gradient boosting, trained in some hospitals, tested in others
- Prognostic model development and validation
- Machine learning
- Stochastic gradient boosting
- Trained in some hospitals, tested in others
Main finding
As reported in the abstract
Model performance for 30-day mortality across the four groups
- Sensitivity
- 86-97%
- Specificity
- 86-94%
- Accuracy
- 86% to 94%
- AUC
- 0.84 to 0.93
- Cohen's kappa
- 0.34 to 0.45
- PPV
- 23% to 37%
Observational data: associations, not effects.
Further findings
Key predictors (normalized relative influence (NRI) scores)
- Laboratory values (pH, base excess, pCO2, standard bicarbonate, oxygen saturation, lactate, CRP, leukocytes), age, triage category and time to doctor consultation
Original publication
Development and evaluation of a machine learning prediction model for short-term mortality in patients with diabetes or hyperglycemia at emergency department admission. Cardiovasc Diabetol. Published 3 October 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →