Oxidative balance score as a predictor of MASLD-CVD comorbidity risk and mortality: Insights from traditional and machine learning approaches

Figures

• 

  Fig. 1. Schematic flow diagram of exclusion criteria for study cohort.

• 

  Fig. 2. Association between OBS and the prevalence of MASLD-CVD comorbidity, MASLD-only, and CVD-only cohorts. P values from multinomial logistic regression models adjusted for age, sex, race and ethnicity, education level, poverty income ratio, physical activity, smoking and alcohol consumption. CI, confidence interval; OBS, oxidative balance score; OR, odds ratio; MASLD, metabolic dysfunction-associated steatotic liver disease; CVD, cardiovascular disease. Q1, OBS<13; Q2, 13≤ OBS <21; Q3, 21≤ OBS <25; Q4, OBS ≥25.

• 

  Fig. 3. Association between OBS and all-cause mortality in MASLD-CVD comorbidity, MASLD-only, and CVD-only cohorts. Hazard ratios were adjusted for age, sex, race and ethnicity, education level, poverty income ratio, physical activity, smoking and alcohol consumption. CI, confidence interval; OBS, oxidative balance score; HR, hazard ratio; MASLD, metabolic dysfunction-associated steatotic liver disease; CVD, cardiovascular disease. Q1, OBS<13; Q2, 13≤ OBS <21; Q3, 21≤ OBS <25; Q4, OBS ≥25.

• 

  Fig. 4. The association of OBS with all-cause mortality in (a) MASLD-CVD comorbidity, (b) MASLD, (c) CVD cohorts visualized by restricted cubic spline. Hazard ratios were adjusted for age, sex, race and ethnicity, education level, poverty income ratio, physical activity, smoking and alcohol consumption.

• 

  Fig. 5. SHAP-based feature importance of machine learning models for (a) MASLD-CVD comorbidity status prediction (vs. monomorbidity): with radar chart visualization, (b) corresponding SHAP summary plot for comorbidity status, (c) all-cause mortality risk prediction among comorbid patients with radar chart visualization, and (d) corresponding SHAP summary plot for mortality risk. TyG, Triglyceride-Glucose Index; WHtR, Waist-to-Height Ratio; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; HOMA IR, Homeostatic Model Assessment for Insulin Resistance.

• 

  Fig. 6. Bioinformatic prediction of potential targets and mechanistic pathways. (a) Venn diagram showing overlapping targets between predicted targets of alcohol and cotinine (green) and MASLD-/CVD-associated genes (purple); (b) PPI network of the overlapping targets; (c) Hub genes identified within the PPI network using the CytoHubba plugin; (d) KEGG pathway enrichment analysis of the 99 overlapping targets; (e) Gene Ontology (GO) functional enrichment analysis of the 99 overlapping targets. BP, biological process; CC, cellular component; MF, molecular function; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; PPI, protein-protein interaction.