Fig. 4

Model accuracy and loss of HormoNet before and after feature selection techniques on our benchmark datasets. Where, A is the model accuracy of HormoNet for HDI before using feature selection, B, C, and D are model accuracy of HormoNet for HDI after using RF, lsvc, and XGBoost. E is model accuracy of HormoNet on our benchmark dataset for hormone-drug interaction risk level. F, G, and H are model accuracy of HormoNet for risk level after RF, lsvc, and XGBoost. I is the model loss of HormoNet for HDI before using feature selection, J, K, and L are model loss of HormoNet for HDI after using RF, lsvc, and XGBoost. M is model loss of HormoNet on our benchmark dataset for hormone-drug interaction risk level. N, O, and p are model loss of HormoNet for risk level after RF, lsvc, and XGBoost.