Enhancing Explainability and Transparency in Deep Learning Models for Healthcare Applications

Recently, it has been observed that deep learning models can be very useful in handling healthcare facilities mostly medical diagnosis and management. However, they are not interpretable and transparent enough to be accepted and used in the clinical environments where the matching reasoning is important. This paper proposes a new strategy to improve explainability of deep learning approaches when used in healthcare data set. We introduce an IDNN design with explainable layers and postdoc approaches like SHAP and LIME for visualization and explanation of model predictions. Our framework is applied to multiple healthcare datasets such as MIMICIII and ChestXray14 where we will compare the traditional method with the proposed method by training accuracy and explainability evaluation metrics. Results indicate that this causes minimal compromise in accuracy on the other hand, other explainability measures of model fidelity, and comprehensibility reveal significant gains, signifying the potential of this method in advancing the use of AI solutions in healthcare contexts.