Breaking down barriers: Role of artificial intelligence in patient-friendly breast cancer education.

Abstract

e13721 Background: Breast cancer is the most common cancer and is the second leading cause of cancer death in women in the United States. Adequate and effective patient education on breast cancer is important in its management. The accessibility of educational information is increased with the assistance of artificial intelligence (AI) such as chatbots and virtual assistance tools. Despite the availability of educational information from several reputed professional societies, none are tailored to the patient’s literacy level. We aim to assess the efficacy of popular AI tools/chatbots in delivering complex medical information simply and understandably. Methods: Patient education material, with a focus on frequently asked questions was collected from the National Breast Cancer Foundation website. We analyzed the questions using ChatGPT (Open AI), Copilot (Microsoft), and Gemini (Google) to simplify it to a 6th-grade reading level (RL) or below. We compared the original and simplified responses using Flesch-Kincaid Grade Level (FKGL), Flesch Reading Ease Score (FRES), RL, and average words per sentence. FKGL assesses readability based on sentence length and word complexity, whereas FRES measures comprehensibility on a scale of 0 (very difficult) to 100 (very easy). Statistical significance was determined with ANOVA and a post-hoc test to assess differences in performance. P-value < 0.05 was considered statistically significant. Results: AI modifications were statistically significant (P < 0.01) in FKGL and FRES indicating AI responses were comparatively easy to read and comprehend. Additionally, Gemini responses were found to be better (P < 0.01) in FKGL compared to ChatGPT or Copilot. No statistical significance was noted in FRES among various AI models. AI responses had comparatively lesser word count (P < 0.01). Gemini generated responses with a lesser word count (P < 0.01) than ChatGPT and Copilot. No significant difference was noted in FKGL and average word count between ChatGPT and Copilot. Conclusions: Artificial intelligence is transforming medicine in various aspects such as diagnosis, treatment, and patient education. AI tools have the potential to simplify complicated medical information, improving accessibility and comprehensibility for individuals with various literacy levels. Our study suggests modifications made by ChatGPT, Copilot, and Gemini result in successful text simplification. AI can effectively communicate complex medical information on Breast Cancer and tailor it to the needs of a specific audience.

Ähnliche Arbeiten