목적: Oral cavity cancer recurrence is a significant concern in clinical oncology, necessitating effective predictive tools for better patient management. Traditional machine learning (ML) techniques have been employed in this context, but the advent of Automated Machine Learning (AutoML) offers a potentially more efficient alternative. This study aims to assess the efficacy of AutoML frameworks compared to single ML modules in predicting the recurrence of oral cavity cancer. 방법:The study utilized a dataset of 597 patients from the Asan Medical Center Cancer Registry who underwent surgical treatment for oral cavity cancer. The dataset comprised 22 features, including demographic data (age, sex, BMI), clinical parameters (TNM stage, depth of invasion [DOI]), and others. The research compared the performance of six traditional ML models including logistic regression, random forest, and XGBoost) with five different AutoML frameworks including mljar-suprvized AutoML, Autogluon, AutoKeras, FLAML and TPOT. Methodological steps involved data preprocessing (missing value imputation, standardization, binarization), oversampling to address class imbalance, and 5-fold cross-validation for model assessment. 결과:In evaluating the models, two primary metrics were used: accuracy and the Receiver Operating Characteristic Area Under the Curve (ROC-AUC) score. The findings revealed that AutoML frameworks outperformed the single ML modules in predicting cancer recurrence, with the best- performing AutoML model achieving an accuracy of 72% and a ROC-AUC score of 0.77. This superior performance is attributed to AutoML's capabilities in efficient experimentation, iteration, and internal ensemble model formation, which reduces the need for extensive hyperparameter tuning. Additionally, the study identified key predictive factors for oral cavity cancer recurrence, including DOI, adjuvant therapy, overall stage, N stage, and extranodal extension (ENE). 결론:The research highlights the potential of AutoML as a more effective tool for predicting oral cavity cancer recurrence. The use of AutoML in clinical settings could enhance predictive accuracy, thereby improving clinical decision-making and patient management in oncology. The identification of significant predictive factors further contributes to understanding cancer recurrence, offering valuable insights for future research and clinical practice.