Combining Deep Learning and Explainable AI for Toxicity Prediction of Chemical Compounds
Eduard Popescu, Adrian Groza, Andreea Cernat
TL;DR
This paper tackles toxicity prediction for chemical compounds using the public Tox21 dataset. It proposes an image-based pipeline using DenseNet121 to extract features from 2D molecular images and applies Grad-CAM to visualize regions driving predictions, combined with an ensemble ML classifier. The approach is benchmarked against fingerprint-based ML, DL on SMILES, and graph neural networks, with the image-based DenseNet method showing competitive performance and added interpretability. A global confidence score further quantifies prediction certainty, underscoring the method's potential for transparent, efficient toxicology screening.
Abstract
The task here is to predict the toxicological activity of chemical compounds based on the Tox21 dataset, a benchmark in computational toxicology. After a domain-specific overview of chemical toxicity, we discuss current computational strategies, focusing on machine learning and deep learning. Several architectures are compared in terms of performance, robustness, and interpretability. This research introduces a novel image-based pipeline based on DenseNet121, which processes 2D graphical representations of chemical structures. Additionally, we employ Grad-CAM visualizations, an explainable AI technique, to interpret the model's predictions and highlight molecular regions contributing to toxicity classification. The proposed architecture achieves competitive results compared to traditional models, demonstrating the potential of deep convolutional networks in cheminformatics. Our findings emphasize the value of combining image-based representations with explainable AI methods to improve both predictive accuracy and model transparency in toxicology.