Table 3 Mean Square Error (MSE) and Mean Absolute Error (MAE) between actual and predicted responses using 3 fold cross validation of integrated Random Forest, separate RF model for each cancer and Heterogeneity Aware Random Forest (HARF) for different cancer types in the CCLE dataset.
From: Heterogeneity Aware Random Forest for Drug Sensitivity Prediction
Drug Name | Cancer Types | Number of Samples | Mean AUC of Cancer Types | Random Forest | Cancer Specific RF | HARF | |||
|---|---|---|---|---|---|---|---|---|---|
MSE | MAE | MSE | MAE | MSE | MAE | ||||
AZD6244 | Skin & CNS | 40 & 29 | 0.304 & 0.090 | 0.0168 | 0.1061 | 0.0163 | 0.0957 | 0.0140 | 0.0861 |
AZD6244 | Skin & Ovary | 40 & 28 | 0.304 & 0.114 | 0.0168 | 0.1083 | 0.0162 | 0.0982 | 0.0145 | 0.0918 |
Lapatinib | Breast & CNS | 29 & 29 | 0.148 & 0.030 | 0.0083 | 0.0703 | 0.0094 | 0.0709 | 0.0073 | 0.0590 |
Nilotinib | CNS & HLT | 29 & 71 | 0.039 & 0.168 | 0.0245 | 0.1065 | 0.0241 | 0.1024 | 0.0225 | 0.0986 |
Nilotinib | Ovary & HLT | 28 & 71 | 0.046 & 0.168 | 0.0247 | 0.1082 | 0.0223 | 0.0968 | 0.0230 | 0.1002 |
PD-0325901 | CNS & Skin | 29 & 40 | 0.130 & 0.434 | 0.0325 | 0.1507 | 0.0275 | 0.1316 | 0.0269 | 0.1311 |
PD-0325901 | Pancreas & Breast | 30 & 30 | 0.343 & 0.136 | 0.0230 | 0.1210 | 0.0166 | 0.0994 | 0.0137 | 0.0918 |
