Explainability methods for machine learning systems for multimodal medical datasets: research proposal