Heart failure & cardiomyopathies
Explainable Machine Learning and Online Calculators to Predict Heart Failure Mortality in Intensive Care Units
Authors' conclusion
The study developed an LDA and a nomogram model for predicting mortality in HF patients in the ICU.
Study context
- Population
- Patients with heart failure (HF) at first intensive care unit (ICU) stay lasting 24 h to 28 days (Medical Information Mart for Intensive Care IV (MIMIC-IV))1,808 survived, 535 died
- Participants
- 2,343 records
- Comparison
- 15 machine learning models; Linear discriminant analysis (LDA) model and logistic regression nomogram
- Outcome
- All-cause mortality
- Follow-up
- 28 days
Methodological detailsretrospective, MIMIC-IV database, external validation in eICU database, machine learning, least absolute shrinkage and selection operator (LASSO) feature selection, SHAP explainability
- Retrospective
- MIMIC-IV database
- External validation in eICU database
- Machine learning
- LASSO feature selection
- SHAP explainability
Performance
As reported in the abstract
Prediction of 28-day all-cause mortality
| Test or model | C-index, training cohort | C-index, testing cohort | AUC, external validation | AUC, training cohort | AUC, internal validation cohort | AUC, external validation cohort | Accuracy, training cohort | Accuracy, testing cohort | Accuracy, precision, recall and F1 score, external validation |
|---|---|---|---|---|---|---|---|---|---|
| LDA | 0.7972 | 0.8125 | 0.79 | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract | 0.8354 | 0.8563 | ≈0.9 |
| Nomogram | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract | 0.852 | 0.855 | 0.770 | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract |
C-index, training cohort
LDA0.7972
Nomogram— not stated in the abstract
C-index, testing cohort
LDA0.8125
Nomogram— not stated in the abstract
AUC, external validation
LDA0.79
Nomogram— not stated in the abstract
AUC, training cohort
LDA— not stated in the abstract
Nomogram0.852
AUC, internal validation cohort
LDA— not stated in the abstract
Nomogram0.855
AUC, external validation cohort
LDA— not stated in the abstract
Nomogram0.770
Accuracy, training cohort
LDA0.8354
Nomogram— not stated in the abstract
Accuracy, testing cohort
LDA0.8563
Nomogram— not stated in the abstract
Accuracy, precision, recall and F1 score, external validation
LDA≈0.9
Nomogram— not stated in the abstract
— not stated in the abstract
Original publication
Explainable Machine Learning and Online Calculators to Predict Heart Failure Mortality in Intensive Care Units. ESC Heart Fail. Published 19 September 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →