| Title: | Development of an artificial intelligence model for predicting cardiovascular disease in MASLD |
| Journal: | Scientific Reports |
| Published: | 3 Aug 2026 |
| DOI: | https://doi.org/10.1038/s41598-026-64829-6 |
Publication 21224
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
Abstract
Metabolic dysfunction-associated steatotic liver disease (MASLD) is associated with cardiovascular disease (CVD). Conventional CVD risk scores often fail to reflect MASLD-specific factors. We aimed to develop and validate an artificial intelligence (AI) model for 10-year CVD risk prediction in MASLD using UK Biobank data. We included 39,691 participants with a fatty liver index ≥ 60 and at least one cardiometabolic risk factor. Incident CVD events were identified using International Classification of Diseases, 10th Revision codes. The M-CARD AI model was developed using extreme gradient boosting, incorporating conventional CVD risk factors, liver-related blood tests, and handgrip strength. Training and internal validation used an England cohort (n = 35,810); a separate Scotland/Wales cohort (n = 3,881) served as a geographically distinct test set to assess transportability. In testing, M-CARD AI achieved an area under the receiver operating characteristic curve (AUROC) of 0.76 and area under the precision-recall curve of 0.20. It outperformed conventional scores, including atherosclerotic cardiovascular disease risk score (AUROC 0.72), Framingham risk score (AUROC 0.59), fibrosis-4 index (AUROC 0.58), and steatosis-associated fibrosis estimator (AUROC 0.57). By incorporating liver and muscle indicators, M-CARD AI addresses a gap in MASLD-related CVD assessment. Future prospective and multi-ethnic validation studies are needed to confirm clinical applicability.</p>
4 Authors
- Tae Seop Lim
- Chanmin Park
- SungA Bae
- Dukyong Yoon
1 Application
| Application ID | Title |
|---|---|
| 85037 | Prediction of drug response using pharmacogenomics-driven machine learning approach |
Enabling scientific discoveries that improve human health