Abstract: PURPOSE: To perceive the dental undergraduate's policy of coping with online learning and their decision-making laws during the COVID-19 pandemic. METHODS: For dental undergraduate students from the 2016 grade to 2018 grade of Lishui University, two prospective questionnaire surveys were conducted before the online course starting and four weeks later. SPSS Modeler18.0 software was used to screen, review, and analyze the data. TAN (tree augmented naive) Bayesian network models were utilized to analyze and predict variables. Indicators like the overall prediction accuracy, receiver operating characteristic curve (ROC curve), and area under the ROC curve(AUC value) were applied to evaluate the model's predicting performances. RESULTS: The case score of each survey was 422 and 382, and the Cronbach's α coefficients of internal consistency were 0.91 and 0.82. Among the decision-making variables in the aspect of "whether to preview online learning materials", the top-two variables were "looking forward to the semester beginning" and "the validity of the network materials". In speaking of "whether the online courses meet the offline course standards", the top-three variables were "the rhythm of lecturing on live or in recorded videos", "how many online tasks', and" the data frame and organization". The overall prediction accuracy of each constructed TAN Bayesian network model was 89.42% and 87.82%, and their AUC values were 0.75 and 0.93, respectively. CONCLUSIONS: To truly make online courses comparable to the off-line curriculum, teachers should fully understand how the students cope with their online learning at first. Then, only by perceiving and recognizing the students' expectations for education, by efficiently preparing and organizing online materials with all-round, clearly-structured, vivid, comprehensible contents and moderate difficult tasks, by well interacting with students through different websites and social media, can we truly achieve " ongoing learning with suspended class".

Key words: Online learning, TAN Bayesian network, COVID-19 pandemic