Artificial Intelligence (AI) to predict dental student academic performance based on pre-university results

• Main Authors: Ahmad Amin, Afifah Munirah, Abdullah, Adilah Syahirah, Lestari, Widya, Sukotjo, Cortino, Utomo, Chandra Prasetyo, Ismail, Azlini
• Format: Conference or Workshop Item
• Language: English
• Published: 2022
• Subjects: QA Mathematics; QA276 Mathematical Statistics; QA297 Numerical Analysis; QA300 Analysis; RK Dentistry
• Online Access: http://irep.iium.edu.my/98509/1/98509_Artificial%20Intelligence%20%28AI%29%20to%20predict%20dental%20student%20academic.pdf; http://irep.iium.edu.my/98509/

Background: The dental school admission process involves establishing criteria for evaluating applicants, weighing the various admission criteria, and then comparing applicants based on selected criteria and weighting. In the Kulliyyah of Dentistry, International Islamic University Malaysia (IIUM), admission is mainly based on matriculation cumulative grade point assessment (CGPA) results. Successful graduation from the Bachelor of Dental Surgery program is assessed through four Professional Exams. Objective: This study aims to predict the academic performance of dental students based on their admission results using Artificial Intelligence. Methods: Various Machine Learning (ML) algorithms were applied using academic result samples of graduates of the Kulliyyah of Dentistry, IIUM from 2012-2017. Classifiers employed for the system include Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM) techniques. The dataset input variables included age during admission, pre-university results, and Professional Exams results. The predictions focused on the number of failures in Year 1 and Year 3. Due to imbalanced data, resampling using the Synthetic Minority Oversampling (SMOTE) method was employed. Results: The number of failures or repeat papers in Year 1 and Year 3 were identified. Logistic Regression (LR) is the most effective algorithm for forecasting student success in Year 1 with accuracy 0.88 and Decision Tree (DT) in Year 3 with accuracy 0.9. Support Vector Machine (SVM) was the least efficient, with accuracy of 0.69 for Year 1 and 0.88 for Year 3. Conclusion: The results obtained show that machine learning technology is efficient and relevant for predicting dental students’ performance based on their pre-admission results. Stakeholders can use the identified effective factors to improve education outcomes.