(ID: 284) Developing pharmacy students’ consultation skills using artificial intelligence

Abstract Introduction Pharmacists are increasingly expected to engage in patient-facing roles, necessitating strong consultation skills. However, many pharmacy graduates report inadequate training in communication during their undergraduate education. Traditional teaching methods, such as lectures and role-playing, often lack sufficient feedback and realism. With the rise of Artificial Intelligence (AI), tools like ChatGPT offer new opportunities to simulate patient interactions and provide feedback [1]. This study investigates whether ChatGPT can support the development of consultation skills in pharmacy students by offering realistic, low-stress practice environments and useful feedback. Aim The study aimed to evaluate the effectiveness of ChatGPT-4o in enhancing the consultation skills of undergraduate pharmacy students. Specifically, it assessed the impact of AI consultations on students’ confidence and anxiety, the usefulness and accuracy of AI-generated feedback, and the fidelity of ChatGPT in simulating real patient interactions. Methodology A cross-sectional study was conducted with 32 MPharm students (15 Year 3, 17 Year 4) at the University of Bath. Participants engaged in a 10-minute voice-based consultation with ChatGPT, simulating a patient case involving a hypertensive patient experiencing a dry cough. Pre- and post-consultation surveys measured changes in students’ confidence and anxiety using Likert-scale items. Statistical analysis employed Wilcoxon signed-rank and Mann-Whitney U tests. Thematic analysis of student feedback and faculty-reviewed consultation transcripts assessed the perceived realism (fidelity) and feedback quality of the AI tool. Results Post-consultation survey results showed statistically significant increases in students’ confidence (Item A: p = 0.035) and reduced anxiety (Item B: p = 0.003). Thematic analysis revealed five positive themes, including ChatGPT as a useful revision tool and a less stressful alternative to traditional methods. Analysis confirmed that ChatGPT provided generally accurate and helpful feedback, particularly in areas like empathy and professionalism, though it lacked detail in basic consultation elements (e.g., allergies, patient ID). Year 3 students found the feedback more useful than Year 4 students (p < 0.001). Regarding fidelity, some students described the AI interaction as realistic, particularly resembling a phone consultation, though issues such as premature information disclosure and interruptions reduced realism. Discussion We conclude that ChatGPT can serve as a valuable, accessible tool for pharmacy students to practice consultation skills in a low-pressure environment. It effectively boosts confidence and provides useful feedback, especially for earlier-year students. However, limitations include insufficient detail in feedback and occasional disruptions in conversation flow. Differences in perception between Year 3 and Year 4 students suggest the need for tailoring AI simulations to students’ experience levels. Future research should explore long-term impacts, refine AI prompts using mark schemes, and assess ChatGPT’s emotional intelligence to enhance realism and feedback quality. Despite its limitations, ChatGPT holds promise as a supplementary tool in pharmacy education, potentially improving students’ readiness for real-world consultations and patient care.