Random Forest-Based Prediction of Academic Misconduct Among Medical Professional Master's Student

作者：Li Zhang, Shuo Wang, Shuai Xue · 年份：2026 · DOI：10.1145/3802607.3802668 · 被引用次数：1 · 研究领域：Academic integrity and plagiarism、E-Learning and COVID-19、Statistical Methods in Epidemiology

Academic integrity issues among medical professional master's students are now common under the dual-track integration training model in China. Data-driven approaches may help with prediction and intervention. This study applies the Random Forest algorithm to predict academic misconduct behavior based on survey data from 333 medical professional master's students. The anonymous questionnaire captured information on misconduct frequency, self-evaluation of academic integrity, perceived causes, and attitudes toward preventive measures. Five classification algorithms were compared: Random Forest, Support Vector Machine, Logistic Regression, Decision Tree, and K-Nearest Neighbors. The Random Forest model reached a classification accuracy of 84.7%, the best among tested models, with precision of 83.1%, recall of 82.5%, F1-score of 82.8%, and AUC of 0.89. Feature importance analysis showed publication pressure (0.186), lack of supervision (0.172), and negative social atmosphere (0.165) as the top three predictors, together making up about 52% of the model's predictive power. Survey results show that 89.2% of respondents supported implementing a dual-mentor system that combines clinical guidance and research supervision. These findings offer some suggestions for educational reform and dual-mentor systems to reduce academic misconduct in medical graduate education.