Online Poster Portal

Objective

Objective structured clinical examinations (OSCEs) are an important part of medical education, allowing students to apply clinical knowledge and communication skills in simulated patient encounters. However, traditional OSCEs require significant personnel, time, and infrastructure, which can make them difficult to scale in resource-limited settings. Artificial intelligence (AI)-based OSCE platforms may offer a standardized alternative, particularly in resource-limited settings. Female genital mutilation/cutting (FGM/C) affects more than 230 million women and girls globally.1 In The Gambia, where approximately 73% of women aged 15-49 have undergone FGM/C, graduating health professionals are likely to encounter survivors requiring informed and culturally sensitive care.2 We evaluated the feasibility of implementing AI-based OSCE simulations at the University of The Gambia and explored students performance in FGM-related counseling domains.

Study Design

We conducted a prospective pilot educational study in collaboration with the University of The Gambia during March 2026. Nursing and midwifery students completed 2 AI-simulated reproductive health encounters using the SimChat platform: (1) recurrent urinary symptoms in the setting of prior FGM/C and (2) preconception/family planning counseling. Encounters were completed using text or audio formats. Data collected included encounter transcripts, automated checklist scores, student reflections, and post-session surveys. Feasibility outcomes included connectivity issues, transcription quality, troubleshooting burden, and overall usability.

Results

Thirty-one students participated. Audio-based encounters were frequently limited by poor transcription quality and were excluded from primary transcript analysis. Text-based encounters were subsequently analyzed across both cases. Most students demonstrated baseline communication skills and focused history taking. However, applied counseling and contextual reasoning skills were limited. Direct discussion of FGM/C was rare, occurring in only 3 encounters involving 2 students despite embedded opportunities to do so. Implementation was feasible but required frequent troubleshooting. Major barriers included unstable internet connectivity, prolonged session times related to limited familiarity with laptop-based assessments, and speech-recognition limitations for audio encounters.

Conclusion

AI-based OSCEs were feasible to pilot in The Gambia but required substantial local adaptation. While students demonstrated foundational communication skills, the study identified gaps in FGM-informed counseling, autonomy-centered reproductive counseling, and clinical reasoning. These findings suggest that clinically relevant FGM/C education warrants greater attention within health professions curricula. With further refinement and improved technological infrastructure, AI OSCEs may become a scalable tool to support health professions education in resource-limited settings.