63/100

Moderate evidence

Cite
More
EmailLinkedInX

In 1,838 stroke survivors, machine learning using a liver blood score predicted risk of death

Study title: Predictive Value of the Albumin-Bilirubin Score for All-Cause Mortality in Individuals with Stroke: Machine Learning Prediction Models with SHAP-Based Interpretation

Evidence type
Observational study
Sample
n = 1,838
Sources
1 sourcePeer-reviewed
Published
1 min read

Study title: Predictive Value of the Albumin-Bilirubin Score for All-Cause Mortality in Individuals with Stroke: Machine Learning Prediction Models with SHAP-Based Interpretation

Researchers evaluated the albumin-bilirubin score as a predictor of all-cause mortality in 1,838 stroke survivors from NHANES 1999-2018, with 78 months median follow-up. They selected variables with Boruta, tested associations with Kaplan-Meier and Cox models, and built five machine learning models with 10-fold cross-validation assessed by C-index, Brier score, time-dependent AUC, and calibration, interpreting the best model with SHAP.

Read the study abstract

Background The albumin-bilirubin (ALBI) score, an objective liver function and prognostic indicator, has shown predictive value in various diseases, but its long-term prognostic significance in stroke survivors remains unclear. Objective To evaluate the ALBI score's predictive value for all-cause mortality (ACM) in individuals with stroke. Methods Data from the National Health and Nutrition Examination Survey (1999-2018) were analyzed. The Boruta algorithm selected key variables associated with ACM. Kaplan-Meier…

Predictive Value of the Albumin-Bilirubin Score for All-Cause Mortality in Individuals with Stroke: Machine Learning Prediction Models with SHAP-Based Interpretation
Granada - Hospital Doctor Olóriz 1 by Zarateman. CC0 · http://creativecommons.org/publicdomain/zero/1.0/deed.en

The gain

In 1,838 stroke survivors, machine learning models incorporating the albumin-bilirubin score predicted all-cause mortality and enabled risk stratification for individualized management.

Main points

  1. Study analyzed NHANES 1999-2018 data for 1,838 stroke survivors with median follow-up of 78 months.
  2. Boruta algorithm selected variables including age, sex, cardiovascular disease, lymphocyte count, uric acid, and creatinine alongside ALBI.
  3. Five ML models were built with 10-fold cross-validation and evaluated by C-index, integrated Brier score, time-dependent AUC, and calibration curves.
  4. SHapley Additive exPlanations (SHAP) was used to interpret the best-performing model.

The full story

The analysis used National Health and Nutrition Examination Survey data from 1999-2018 and applied the Boruta algorithm to select predictors including marital status, body mass index, aspartate/alanine aminotransferase, polyunsaturated fatty acids, and basophil percentage.

Associations were examined with Kaplan-Meier curves, Cox regression, and restricted cubic splines to assess nonlinearity between ALBI and mortality risk, with log-rank testing showing survival differences across quartiles.

Sources

1 on file
  1. Journal of Stroke and Cerebrovascular Diseasesdoi:10.1016/j.jstrokecerebrovasdis.2026.108759
    Peer-reviewed2026-10-01

Talk it over

What readers make of this record

What readers say