Development and validation of an explainable machine learning model using routine laboratory biomarkers for identifying prevalent MASLD: Evidence from two observational studies
Simple blood tests and measurements can identify fatty liver disease as reliably as expensive imaging, making screening more accessible.
Among 11 ML algorithms, an Extra Trees model using routine labs (diabetes, waist circumference, age, hypertension, atherogenic index) achieved AUC 0.879 internally and 0.822 in Korean external validation for identifying prevalent MASLD. SHAP visualization enhances clinical interpretability and supports bedside application.
What the study was
- Study design
- ML model development (NHANES 2017-2020) with external validation (KNHANES 2019-2021)
- Population
- Training: N=2,760; Internal: N=1,184; External: KNHANES N not specified but large survey cohort
- Sample size
- 3944
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Clin Exp Med
Why it surfaced
Externally validated ML MASLD screening tool using routine labs. Cross-cultural validation (US + Korean cohort) is a strength. SHAP interpretability supports clinical adoption. Useful adjacent to GLP-1/SGLT2 monitoring.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.