The field of medical diagnosis is witnessing a significant shift towards explainable AI, with a focus on developing transparent and trustworthy models that can provide accurate predictions and insights into the decision-making process. Recent studies have demonstrated the potential of multimodal deep learning frameworks, generative adversarial networks, and prototype learning models in detecting lung diseases, classifying skin lesions, and estimating ejection fractions. These innovative approaches have achieved high accuracy and outperformed traditional methods, while also providing interpretable explanations and visualizations of the results. Notably, the integration of explainable AI techniques, such as Grad-CAM, SHAP, and LIME, has enabled clinicians to understand the underlying factors driving the predictions and build trust in the models. The applications of these models are vast, ranging from telemedicine and point-of-care diagnostics to real-world respiratory screening and continuous neurocognitive monitoring. Some noteworthy papers in this regard include: The study on explainable multimodal deep learning for automatic lung-disease detection, which achieved strong generalization and outperformed ablated variants. The proposal of ProtoEFNet, a novel video-based prototype learning model for continuous ejection fraction regression, which provided clinically relevant insights and achieved accuracy on par with non-interpretable counterparts.