Model Results Using SLOA_CB: Gallstone Prediction Performance Metrics

Model performance results from a study using a CatBoost classifier and Sea Lion Optimization Algorithm to predict gallstone disease from tabular data. The dataset includes accuracy, F1-score, precision, and recall metrics for models using 38 features and a subset of 19 selected features. It was authored by Prosenjit Das and last updated on June 1, 2026.

Use Cases

• Benchmarking classification model performance for gallstone prediction based on reported accuracy and F1-scores.
• Analyzing the impact of feature selection on model metrics based on the comparison between 38 and 19 feature sets.
• Studying model explainability techniques based on the mention of SHAP, LIME, and DiCE being applied.
• Investigating the influence of specific biomarkers like C-Reactive Protein (CRP) and Vitamin D on predictions.

Strengths

• Performance metrics are quantified with specific numbers, including a mean accuracy of 80.42% for the optimized model.
• Model evaluation includes multiple metrics (accuracy, F1-score, precision, recall) and uses 5-fold cross-validation.
• Results are provided for both a full feature set (38 features) and an optimized subset (19 features).

Limitations

• The underlying patient data and the 38 specific features are not included in this dataset, only the model results.
• Row count for the original data is unknown, which may limit suitability assessment for modeling tasks.
• Column-level documentation for the original features is absent; field semantics must be inferred from the description.

Provenance

Source

Prosenjit Das via figshare

Collection Method

Results from applying CatBoost and Sea Lion Optimization Algorithm to a publicly available tabular dataset for gallstone prediction.

Freshness

Last updated 2026-06-01 17:51:50; freshness should be verified.