Deep Learning methods are often utilised for medical predictions, but they suffer from a lack of interpretability, thus the demand for transparent and interpretable models has become imperative. However, while methods for explaining model decisions have been extensively explored, temporal explanations remain largely underrepresented. To address this gap, we propose the Batch-Integrated Gradients (Batch-IG) library, building upon the foundation of the Integrated Gradients method. Batch-IG aims to provide temporal explanations by attributing the model’s predictions of input features over time. By integrating gradients along the temporal dimension, Batch-IG offers insights into how changes in input features contribute to model predictions across time batches. Our method extends the applicability of integrated gradients to temporal domains, thereby enhancing the interpretability of deep learning models operating on temporal data. This chapter presents a simple library for applying Batch-Integrated Gradients for Neural Networks developed in PyTorch.