<![CDATA[Accurate early detection is crucial for reducing mortality rates from cervical cancer. However, the application of machine learning to medical data is often hindered by class imbalance, causing prediction results to be biased toward the majority class. On the other hand, the process of parameter search using conventional methods such as GridSearchCV requires significant computational time. Therefore, this study proposes the application of the ADASYN (Adaptive Synthetic Sampling) method and Bayesian optimization to the Random Forest algorithm. In its implementation, ADASYN is used to adaptively synthesize minority data samples to rebalance their distribution. Meanwhile, Bayesian optimization serves to determine the optimal hyperparameter combination through a faster probabilistic approach. Model evaluation was conducted across four testing scenarios with training-to-test data splits of 90:10, 80:20, and 70:30. Findings from this study indicate that the standard Random Forest algorithm still produces biased predictions. However, classification performance improved significantly after the model was combined with ADASYN and Bayesian Optimization. The optimal results were achieved at a 70:30 ratio, recording accuracy of 98.06%, precision of 97.03%, recall of 99.13%, and an F1-score of 98.07%, with a computation time of 32.66 seconds. Overall, the proposed model successfully addresses data imbalance while reducing optimization time, enabling it to predict biopsy diagnoses with high precision.]]>
