Evaluating Artificial Intelligence and Traditional Learning Tools for Chest X-Ray Interpretation: A Descriptive Study

Background

Chest X-ray (CXR) interpretation is a fundamental yet challenging skill for medical students to master. Traditional resources like Radiopaedia offer extensive content, while newer artificial intelligence (AI) tools, such as Chester, provide pattern recognition and real-time feedback. This study aims to evaluate Radiopaedia and Chester's effectiveness as educational tools and to explore student perspectives on AI.

Approach

A teaching session on CXR interpretation fundamentals was delivered to establish a standardised baseline of knowledge among participants, followed by a live tutorial introducing students to the functionality of both Chester AI and Radiopaedia. Students engaged with both tools to answer a 25-item workbook assessing complex CXR pathologies. CXRs were deliberately selected for their complexity to examine student engagement with online learning tools amid diagnostic uncertainty, encouraging applied clinical reasoning.

Evaluation

Preclinical medical students were recruited and randomly assigned to the Chester AI (n = 5) or Radiopaedia group (n = 5). During the workbook task, participants were instructed to engage with the workbook using Radiopaedia and Chester AI. Post-session, participants took part in focus groups to share their experiences. Thematic analysis highlighted Chester's efficiency and potential as a revision tool but noted limitations with complex CXR pathologies. Radiopaedia was valued for its comprehensiveness but was less efficient for the workbook task due to its vast array of content.

Implications

AI tools such as Chester show promise as complementary resources alongside traditional learning materials. Combining Chester's efficiency and real-time feedback with Radiopaedia's in-depth content may optimise learning and improve CXR interpretation skills.