Heart failure & cardiomyopathies
Prediction of 90 Day Mortality in Elderly Patients with Acute Heart Failure (HF) from e-Health Records Using Artificial Intelligence
Authors' conclusion
The study proposed a robust model for the automatic prediction of 90 day mortality risk 48 h after hospitalization for decompensated HF. This could be used to identify high-risk patients for intensification of therapeutic management.
Study context
- Population
- HF-related admissions; derivation cohort admitted to a single hospital from 2015 to 2020, evaluation cohorts admitted to two hospitals between March and December 2021admissions in the derivation cohort, evaluation cohorts 348 and 388 admissions · Derivation cohort 2,257 · First evaluation cohort 348 · Second evaluation cohort 388
- Participants
- 2,257 admissionsEvents: 248 deaths after hospitalization (derivation cohort); 34 and 38 deaths in the evaluation cohorts
- Predictors
- Electronic health record (EHR) data 48 h after admission; Forty-two independent variables selected with least absolute shrinkage and selection operator; logistic regression model
- Predicted outcome
- 90 day post-discharge mortality
Methodological detailslogistic regression after least absolute shrinkage and selection operator variable selection, using artificial intelligence, single hospital derivation cohort, two independent evaluation cohorts from two hospitals, natural language processing applied to textual information, electronic health record (EHR) data collected 48 h after admission
- Logistic regression after least absolute shrinkage and selection operator variable selection, using artificial intelligence
- Single hospital derivation cohort
- Two independent evaluation cohorts from two hospitals
- Natural language processing applied to textual information
- Electronic health record (EHR) data collected 48 h after admission
Model performance
As reported in the abstract
AUC
Derivation cohort
0.817
First evaluation cohort
0.750
Second evaluation cohort
0.723
| Test or model | Value | 95% CI |
|---|---|---|
| Derivation cohort | 248 events | 0.789-0.845 |
| First evaluation cohort | 34 events | 0.672-0.829 |
| Second evaluation cohort | 38 events | 0.644-0.803 |
Value
Derivation cohort248 events
First evaluation cohort34 events
Second evaluation cohort38 events
95% CI
Derivation cohort0.789-0.845
First evaluation cohort0.672-0.829
Second evaluation cohort0.644-0.803
Further findings
Calibration: observed 90 day mortality rate by predicted risk category
Calibration: observed 90 day mortality rate by predicted risk category
- Calibration was good
Original publication
Prediction of 90 Day Mortality in Elderly Patients with Acute HF from e-Health Records Using Artificial Intelligence. ESC Heart Fail. Published 13 February 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →