From Blackbox to Explainable AI in Healthcare: Existing Tools and Case Studies

Table 7

Values representing the accuracy of various classification models with SHAP.

Approaches	Accuracy	Sensitivity	Specificity	Precision	F1 Score	Interpretability score
Random forest	87.6	79.2	91.6	82.0	80.4	0.50
Extra trees	87.1	79.3	90.8	80.7	79.9	0.58
Ada boost	85.2	75.1	90.1	78.7	76.4	0.50
Gradient boosting	84.7	78.0	88.0	76.0	76.6	0.50
XGBoost	87.1	79.2	90.8	80.6	80.6	0.25
Max voting	86.6	76.3	91.5	81.6	81.6	0.58