Utilizing natural language processing to analyze student narrative reflections for medical curriculum improvement

Area of Science:

• Medical Education
• Natural Language Processing
• Curriculum Development

Background:

• Medical curricula require continuous improvement for relevance and effective student preparation.
• Traditional end-of-year evaluations often yield low response rates and insufficient feedback.
• Student reflections offer a potential source of rich feedback but are challenging to analyze manually.

Purpose of the Study:

• To develop and implement a Natural Language Processing (NLP) pipeline to analyze medical student reflective writings.
• To retrospectively mine student reflections for actionable feedback to inform curriculum adjustments.
• To identify common themes and topics within fourth-year medical students' reflections on their Sub-Internship experiences.

Main Methods:

• A Natural Language Processing (NLP) pipeline was developed to process reflective writings from medical students.
• The dataset comprised student responses to a faculty-issued question regarding challenges during their fourth-year Sub-Internship (August 2016 - July 2018).
• The NLP pipeline was used to automatically identify recurring themes and topics within the narrative data.

Main Results:

• Eleven distinct themes and topics were successfully identified from the student reflections.
• Several of the identified themes directly informed subsequent curriculum revisions and improvements.
• The NLP approach provided an efficient and scalable method for analyzing qualitative student feedback.

Conclusions:

• Natural Language Processing (NLP) is an effective tool for analyzing qualitative data in medical education.
• Utilizing student reflections via NLP can uncover critical areas for curriculum enhancement.
• This method offers a practical solution for leveraging student feedback to improve medical training programs.