Fig. 3From: Analysis of the benefits of imputation models over traditional QSAR models for toxicity predictionPerformance of multi-task QSAR models. A Ames dataset. B Tox21 dataset. Each box summarizes the MCC scores of 20 independent runs of the model on the test set with identical hyperparameters but differing random seeds. The XGB models (best performing single task QSAR model) are shown as a baseline model. Only the best performing Feature Net model (XGB-FN) is included.Back to article page