Exploring and Identifying Prognostic Phenotypes of Patients with Heart Failure Guided by Explainable Machine
Xue Zhou1, Keijiro Nakamura2, Naohiko Sahara2
1Biomedical Information Engineering Lab, The University of Aizu, Aizuwakamatsu 965-8580, Japan.
Machine learning identified three heart failure patient phenotypes with distinct mortality risks. This approach aids in personalized medicine by classifying new patients and guiding tailored management strategies.
Area of Science:
- Cardiology
- Medical Informatics
- Machine Learning
Background:
- Precision medicine requires identifying patient subgroups with different prognoses.
- Heart failure (HF) patient stratification is crucial for effective management.
- Machine learning (ML) offers potential for discovering prognostic phenotypes.
Purpose of the Study:
- To explore patient phenotypes in heart failure (HF) linked to mortality risk.
- To develop and validate ML models for classifying these prognostic phenotypes.
- To interpret the factors influencing phenotype classification.
Main Methods:
- Unsupervised ML was used to identify phenotypes in a derivation cohort (n=562).
- Supervised ML models were trained for phenotype classification and validated on an independent cohort (n=168).
- Shapley additive explanations (SHAP) were employed for model interpretability.
Main Results:
- Three distinct phenotypes with stratified mortality risks (high, intermediate, low) were identified.
- Significant differences in survival curves and hazard ratios were observed among phenotypes.
- Random forest models achieved high AUCs (0.721-0.815) for phenotype classification, with age and creatinine clearance as key predictors.
Conclusions:
- ML effectively identifies prognostic patient phenotypes in heart failure.
- This facilitates personalized patient management based on predicted mortality risk.
- The identified phenotypes and predictors can guide clinical decision-making.
More Related Videos
09:21Author Spotlight: Generating Neuronal Phenotypic Profiles - A Protocol to Culture and Image Human Midbrain Dopaminergic Neurons
Published on: July 7, 2023
04:05Author Spotlight: Unveiling Prognostic Indicators in Heart Failure - The Role of Phase Angle and Bioelectrical Impedance Analysis
Published on: June 30, 2023
Related Concept Videos
Heart Failure IV: Classification and Diagnostic Evaluation
Pathophysiology of Heart Failure
Heart Failure II: Pathophysiology
Heart Failure I: Introduction
Heart Failure III: Clinical Manifestations
Heart Failure V: Medical Management
