This study evaluates the effect of using the Synthetic Minority Over-sampling Technique (SMOTE) on the performance of Support Vector Machine (SVM) classification models in the diagnosis of appendicitis in children. Class imbalance in medical data is often a significant challenge that can reduce the accuracy of predictive models. To address this, K-Nearest Neighbors (KNN) imputation is used to handle missing data in the dataset. An SVM model with a polynomial kernel was chosen for its ability to capture the non-linear relationship between clinical features and diagnosis. The polynomial kernel parameters were set with d = 3 and c = 1 to balance the model complexity and risk of overfitting and the use of SmOTE to address class imbalance. The results showed that the use of SMOTE increased the model precision from 87.00% to 98.65% and AUC-ROC from 85.96% to 88.04%. However, there was a decrease in recall from 92.55% to 77.66% and F1-Score from 89.69% to 86.90%. This suggests a trade-off between the model's increased ability to distinguish between positive and negative classes and its decreased ability to detect all positive instances. This study makes an essential contribution to medical informatics by showing that while SMOTE can improve some of the model's performance metrics, there are significant trade-offs that must be considered. These findings can assist medical professionals in making better decisions based on more accurate and representative data analysis, particularly in the diagnosis of appendicitis in children.