Journal Article

7838 Development and integration of an AI-driven tool to enhance paediatric trainees clinical diagnostic reasoning: a case-control study

Prashanth Gowda Parameshwara, Ismail, Salim, Kurbet Santosh

2025

Overview

Why did you do this work?Diagnostic errors, often arising from inadequate clinical reasoning skills, remain a concern in healthcare (Lawson & Daniel, 2011). Artificial intelligence (AI) has the potential to revolutionize medical training though innovative curricular elements to finetune trainee aptitude to help perform metacognitive tasks (Gordon et al., 2024). AI presents a promising avenue to revolutionize medical training by enhancing clinical diagnostic reasoning (CDR) skills in postgraduate paediatric training (Ba et al., 2024). This study sought to develop an AI-based tool to support paediatric trainees in enhancing their CDR skills, thereby improving diagnostic accuracy.What did you do?A multicenter case-control study was conducted among 48 postgraduate paediatric trainees. Phase 1 constituted the development of a CDR framework (Croskerry 2009) by educational experts. A software developers team integrated the framework into an AI-powered tool accessible via a web-based platform (figure 1). The interface employed client-server architecture with the back end consisting of a system for storing and processing clinical data points. Algorithms were developed to query this database and dynamically generate case details based on user interactions. Trainees engaged with the AI tool through chatbot function to gather virtual case history, physical findings, and lab results. A training workshop was conducted to orient trainees to CDR concepts (Phase-2). Participants completed a 12-week clinical posting in general paediatrics during which they used the AI tool. Trainees’ CDR skills were evaluated post-intervention by script concordance approach and feedback obtained using a pre-validated questionnaire.What did you find?Overall CDR skills of trainees assessed using script concordance test showed a significant improvement compared to control group (mean difference = 2.5 points, p=0.007). Specific improvements were observed in generating and testing hypotheses (mean difference = 2.7 points, p=0.004), identifying and prioritizing differential diagnoses (mean difference = 1.9 points, p=0.001), and justifying provisional diagnoses (mean difference = 2.3 points, p=0.005). The percentage of trainees able to generate at least four relevant hypotheses improved from 73.6% to 87.7% while those accurately prioritising differential diagnoses rose from 67.4% to 82.5%.Thematic analysis of trainee feedback suggested the tool’s impact on ‘enhancing structured critical thinking’ and ‘facilitating diagnostic accuracy through safe, iterative practice.’ The majority agreed that the tool significantly improved diagnostic reasoning in a supportive, practice-based environment. Analysis of feedback pre-post intervention corroborated these sentiments, citing the tool’s ability to foster ‘confidence in decision-making’ and provide ‘critical reflection opportunities’ on diagnostic errors. Data triangulation confirmed the practical utility of the tool.What does it mean?Our findings demonstrate the potential of AI-driven tools to enhance CDR skills in postgraduate education. Significant performance gains validated the AI tool’s potential for improving CDR. Receptive attitudes of trainees suggest that integrating AI into paediatric curricula could improve clinical training. Adequate resource allocation, interdisciplinary coordination, and continued refinement based on user feedback are key for optimising such innovations (Liang at al., 2019). Future research should explore the scalability of AI-driven educational tools in medical training.Abstract 7838 Figure 1[Image Omitted. See PDF.]ReferencesBa H, Zhang L, Yi Z. Enhancing clinical skills in pediatric trainees: a comparative study of ChatGPT-assisted and traditional teaching methods. BMC Med Educ. 2024;24(1):558.Croskerry P. A universal model of diagnostic reasoning. Acad Med. 2009;84(8):1022–1-28.Gordon M, Daniel M, Ajiboye A, et al. A scoping review of artificial intelligence in medical education: BEME guide no. 84. Med Teach. 2024;46(4):446–470.Lawson AE, Daniel ES. Inferences of clinical diagnostic reasoning and diagnostic error. J Biomed Inform. 2011;44(3):402–412.Liang H, Tsui BY, Ni H, et al. Evaluation and accurate diagnoses of pediatric diseases using artificial intelligence. Nat Med. 2019;25:433–438.

• Publisher: BMJ Publishing Group LTD
• Subject: Algorithms / Artificial intelligence / Case studies / Chatbots / Client Server Architecture / Comparative Analysis / Comparative Education / Comparative studies / Control Groups / Critical Thinking / Decision making / Feedback / Hypotheses / Hypothesis Testing / Inferences / Medical Education / Medical errors / Medical innovations / Pediatrics / Resource allocation / Thinking Skills / Trainees / Training