Heart failure & cardiomyopathies
Machine Learning-Based Model for Worsening Heart Failure Risk in Chinese Chronic Heart Failure Patients
Authors' conclusion
This study identifies NT-proBNP, Cr, UA, Hb, and emotional area scores as crucial predictors of WHF in CHF patients. Among the nine ML algorithms assessed, the RF model showed the highest predictive accuracy. SHAP analysis further emphasized NT-proBNP, UA, and Cr as the most significant predictors.
Study context
- Population
- Patients with chronic HF, China-Japan Friendship Hospital, September 2019 to December 2022
- Participants
- 200 patients
- Outcome
- WHF occurrence
- Follow-up
- 3-month follow-up
Methodological detailssingle-centre, 80% training / 20% test split, LASSO variable selection, SHAP interpretation, nine machine learning (ML) algorithms compared, Online risk predictor based on the random forest (RF) model developed
- Single-centre
- 80% training / 20% test split
- LASSO variable selection
- SHAP interpretation
- Nine ML algorithms compared
- Online risk predictor based on the RF model developed
Main finding
As reported in the abstract
Prediction of WHF (test set)
| Group | AUC | Accuracy | Sensitivity | Specificity | NPV | PPV |
|---|---|---|---|---|---|---|
| RF | 0.842 (95% CI 0.675-1.000) | 0.775 | 0.900 | 0.833 | 0.800 | 0.600 |
RF
AUC0.842 (95% CI 0.675-1.000)
Accuracy0.775
Sensitivity0.900
Specificity0.833
NPV0.800
PPV0.600
Observational data: associations, not effects.
Further findings
Worsening heart failure (WHF) incidence · 3-month follow-up
- Participants developing WHF
- 60 (30%)
Critical predictors of WHF
- N-terminal pro-brain natriuretic peptide (NT-proBNP), creatinine, uric acid, haemoglobin and Minnesota HF QoL emotional score were critical predictors
- SHAP highlighted NT-proBNP, uric acid and creatinine as significant predictors
Original publication
Machine Learning-Based Model for Worsening Heart Failure Risk in Chinese Chronic Heart Failure Patients. ESC Heart Fail. Published 7 September 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →