Prevention & lipidsResearch article
Multichamber Myocardial Strain for Cardiovascular Risk Stratification in Liver Transplant Recipients Using Interpretable Machine Learning
- retrospectively studied
Authors' conclusion
Machine-learning models showed only modest discrimination for post-transplant MACE and did not outperform conventional approaches or established LT risk scores; the incremental value of strain was not statistically significant. Nevertheless, multichamber strain consistently ranked among the strongest contributors in feature attribution, suggesting complementary biological information. These findings are hypothesis-generating and support further evaluation of strain within LT risk models in larger, externally validated cohorts.
Study context
- Population
- Liver transplantation (LT) recipients with pretransplant echocardiographyLT recipients with pretransplant echocardiography
- Participants
- 309 recipients
- Comparison
- Penalized Cox regression, random survival forest (RSF), and gradient boosting survival models were developed and benchmarked against coronary artery disease in LT and cardiovascular risk in orthotopic LT scores
- Outcome
- post-transplant MACE
- Follow-up
- Median 4.6 years
Main finding
As reported in the abstract
patients experienced major adverse cardiovascular events (MACE)
65 (21.0%)
Further findings
| Outcome | Penalized Cox regression | random survival forest (RSF) | gradient boosting survival models | coronary artery disease in LT and cardiovascular risk in orthotopic LT scores |
|---|---|---|---|---|
| Model discrimination · none in penalized Cox regression | ||||
| C-index | Not reported in the abstract | 0.636 | Not reported in the abstract | Not reported in the abstract |
| integrated Brier score | Not reported in the abstract | 0.138 | Not reported in the abstract | Not reported in the abstract |
- Model discrimination — Adding multichamber strain, RSF ΔC-index
- +0.019
- Model discrimination — Adding multichamber strain, gradient boosting ΔC-index
- +0.028
- Model discrimination: small, statistically nonsignificant changes in discrimination
Model discrimination
- RSF showed the highest discrimination
- Adding multichamber strain, penalized Cox regression
- The models did not outperform penalized Cox regression or the established LT-specific risk scores
Feature importance (SHapley Additive exPlanations)
- strain ranked among the most important contributors
Unsupervised clustering
- Clustering identified "cardiometabolic" and "hepatic" phenotypes
Baseline characteristics
- mean age
- 55.8 ± 11.4 years
- females
- 33.3%
Original publication
Summarised from the published abstract; every number is checked against it