Integration of PBL-Informed Medical Cases into First-Year Chemistry Laboratories in a Traditional Medical Curriculum: Perceived Educational Outcomes from NLP-Based Sentiment Analysis

Purpose: To examine first-year medical students’ perceptions of integrating brief, Problem-Based Learning (PBL)–informed medical cases into General and Organic Chemistry laboratories within a traditional 6-year curriculum, and to evaluate Natural Language Processing (NLP)–based sentiment analysis for analysing free-text feedback. Methods: First-year medical students participated across two consecutive semesters. Concise, patient-oriented cases were embedded within General Chemistry (fall) and Organic Chemistry (spring) laboratory sessions. After each semester, students completed an anonymous questionnaire with Likert-type items and three open-ended questions addressing challenges, suggested improvements, and overall experience. Quantitative responses were summarised descriptively, with group comparisons by prior experience conducted using chi-square tests. Free-text responses were analysed using NLP-based sentiment analysis in Python. A stratified sample of comments underwent independent sentiment coding by five blinded human raters, and agreement was assessed using Krippendorff’s alpha. Results: Response rates were 49% in the fall (106/218) and 52% in the spring (112/215). Overall satisfaction was high across both semesters. Most students agreed that integrated medical cases enhanced understanding of medical concepts, increased engagement and interactivity, and was perceived to support early clinical reasoning. Attitudes did not differ significantly by prior experience with integrated labs. Of 380 open-ended responses, 62.6% were classified by NLP as positive and 12.1% as negative; negative comments focused mainly on time constraints, language barriers, and diagnostic difficulty. Inter-rater agreement among human coders was high (Krippendorff’s alpha 0.77), and NLP classifications agreed with the human majority decision in 87% of sampled subset of comments. Conclusion: PBL-informed case integration into first-year chemistry laboratories was associated with high student satisfaction and perceived enhancement of engagement, conceptual understanding, and early clinical reasoning, supporting the value of clinically contextualised laboratory teaching for evidence-based curriculum development in early medical education. In addition, this study demonstrates the potential of combining NLP-based sentiment analysis with targeted human verification as an innovative and scalable methodology for evaluating qualitative student feedback, contributing to the advancement of research and practice in medical education and educational technology. Plain Language Summary: Medical students often find it difficult to see how basic science subjects relate to clinical practice, particularly in the early years of training. In this study, we examined whether adding short, clinically focused medical cases to first-year chemistry laboratory sessions could improve students’ learning experience in a traditional medical curriculum. First-year medical students participated in chemistry laboratories where brief patient cases were used to highlight the medical relevance of the experiments. After completing the sessions, students provided feedback through questionnaires that included both rating scales and open-ended written comments. We analysed these responses using standard statistical methods and computer-based text analysis. Most students reported high satisfaction with the integrated laboratory sessions and felt that the medical cases made the labs more engaging, interactive, and relevant to future clinical practice. Many students also reported that the approach encouraged early clinical reasoning. We found that computer-based analysis of students’ written feedback could reliably summarise overall positive and negative sentiments, although human review was still important for comments suggesting improvements. These findings suggest that simple changes to laboratory teaching can help connect basic science with clinical practice early in medical education and that automated text analysis can support educators in efficiently evaluating student feedback in large courses.