Enhancing Trust in AI-Generated Medical Narratives: A Transparent Approach for Simplifying Radiology Reports

As advancements in artificial intelligence (AI) continue to permeate the medical field, the necessity for transparency in their decision-making becomes paramount. This work investigates how NLP models turn complex medical reports into simpler stories for patients from traditional radiology reports. Given the critical nature of medical information and its impact on patient care, merely producing accurate narratives is not sufficient. Patients, healthcare providers, and other stakeholders need to trust these AI-generated narratives, which comes from understanding how the AI arrived at its conclusions. This paper presents our work in integrating explainability into NLP models, ensuring that every step in the narrative generation process is interpretable and justifiable. Through our approach, we aim to bolster confidence in AI-generated medical narratives, bridging the gap between complex radiology jargon and clear, patient-friendly reports without compromising transparency. Our findings underscore the importance of making AI tools not just powerful, but also clear and trustworthy, especially in sensitive domains like healthcare.