Research Article
Improvement of Adequate Digoxin Dosage: An Application of Machine Learning Approach
Table 6
Performance evaluation of the classifiers for the non-DDI and DDI groups.
| Group | Method | Sensitivity | Specificity | Accuracy | AUC |
| Non-DDI | C4.5 | 0.705/0.091 | 0.806/0.078 | 0.759/0.061 | 0.784/0.065 | CART | 0.696/0.095 | 0.825/0.067 | 0.765/0.055 | 0.791/0.057 | RF | 0.782/0.090 | 0.888/0.054 | 0 0.839/0.041 | 0.912/0.032 | kNN | 0.619/0.172 | 0.547/0.162 | 0.592/0.068 | 0.606/0.070 | LGR | 0.566/0.145 | 0.715/0.117 | 0.648/0.078 | 0.661/0.097 | MLP | 0.741/0.091 | 0.871/0.057 | 0.809/0.059 | 0.813/0.071 |
| DDI | C4.5 | 0.701/0.060 | 0.759/0.050 | 0.732/0.029 | 0.774/0.030 | CART | 0.728/0.051 | 0.776/0.050 | 0.754/0.031 | 0.795/0.031 | RF | 0.790/0.050 | 0.817/0.043 | 0.805/0.027 | 0.892/0.020 | kNN | 0.545/0.094 | 0.651/0.087 | 0.602/0.042 | 0.634/0.048 | LGR | 0.464/0.139 | 0.621/0.145 | 0.551/0.042 | 0.556/0.058 | MLP | 0.745/0.058 | 0.799/0.042 | 0.774/0.037 | 0.777/0.051 |
|
|