AI Readiness in Healthcare through Storytelling XAI

TL;DR

The paper tackles the challenge of trustworthy AI in healthcare by proposing Storytelling XAI, an audience-centric explainability framework that combines knowledge distillation and multi-task learning with both model-agnostic and model-specific interpretability techniques. Using chest X-ray analysis as a use case, the approach trains task-specific teachers and distills their knowledge into a single student model that can perform abnormality detection, segmentation, and report generation while providing textual and visual explanations. The key contributions are the end-to-end, audience-tailored XAI pipeline, the integration of multiple datasets to improve generalization, and the demonstration that concept-based explanations can be delivered alongside technical interpretations for ML practitioners. This framework has practical significance for deploying trustworthy, explainable AI in healthcare and can be extended to other domains with suitable datasets.

Abstract

Artificial Intelligence is rapidly advancing and radically impacting everyday life, driven by the increasing availability of computing power. Despite this trend, the adoption of AI in real-world healthcare is still limited. One of the main reasons is the trustworthiness of AI models and the potential hesitation of domain experts with model predictions. Explainable Artificial Intelligence (XAI) techniques aim to address these issues. However, explainability can mean different things to people with different backgrounds, expertise, and goals. To address the target audience with diverse needs, we develop storytelling XAI. In this research, we have developed an approach that combines multi-task distillation with interpretability techniques to enable audience-centric explainability. Using multi-task distillation allows the model to exploit the relationships between tasks, potentially improving interpretability as each task supports the other leading to an enhanced interpretability from the perspective of a domain expert. The distillation process allows us to extend this research to large deep models that are highly complex. We focus on both model-agnostic and model-specific methods of interpretability, supported by textual justification of the results in healthcare through our use case. Our methods increase the trust of both the domain experts and the machine learning experts to enable a responsible AI.

Figures (3)

• Figure 1: A sample chest X-ray with the corresponding report with original text and extended text. Left: X-ray image of a human chest. Top right: Original text content describing the prognosis of the chest x-ray in medical jargon. Bottom right: Augmented text describing the patient's lung in a concise and clear manner.
• Figure 2: Storytelling XAI Framework. Three teacher networks are trained to perform the specific tasks numbers 1, 2, and 3. Using knowledge distillation, the student model acquires knowledge to perform all the distinct tasks alone. The student model learns the underlying relationship between different features enhancing the interpretability.
• Figure 3: Overview of the Storytelling XAI framework applied in the medical imaging domain. The input image is provided as an input to the student model. The student model performs abnormality detection, lung segmentation, and report text generation. The individual results are provided as input to the interpretability module to generate interpretations. The attention map visualization shows different detected abnormalities.