Large Language Model–Assisted Qualitative Analysis: A Case Study of Medical Students with Academic Difficulties

Abstract

 Large language models (LLMs) are increasingly used to support qualitative research, yet practical accounts of how they can be incorporated without displacing human interpretation remain limited. This study presents a researcher-supervised workflow for processing and analysing semi-structured interviews with medical students experiencing academic difficulties. Thirty interviews were transcribed, checked against recordings, de-identified, and analysed through three stages: event/theme identification, element-level examination of focal events, and cross-event interpretive construction. The workflow identified recurring difficulties involving unclear learning goals, limited learning resources, professional-identity tension, and family expectations. LLMs supported the organisation of long transcripts and the comparison of related experiences; researchers verified outputs, retained contextual qualifications, and developed the final interpretation. The findings illustrate that LLMs can make qualitative analysis more traceable and efficient when their outputs are treated as provisional analytic material rather than authoritative conclusions. We propose practical safeguards for transcription, privacy, validation, and documentation. This case study offers a concise, reproducible pathway for using LLMs to assist—not replace—qualitative inquiry.