This paper conducts a comparative analysis between several kernel functions of support vector machine learning classifier for (Surveillance), to determine which kernel function is better in determining the best one that can be used on medical diagnosis datasets for the best accuracy and performance. Its goal is to find the best kernel for the best accuracy and performance for medical diagnosis datasets and doing it through comparing 3 different kernels which are: ―Radial Basis Function‖, ―Linear Function‖, ―Polynomial Function‖. We used Support Vector Machine (SVM) algorithm in our training and testing while K folding (Cross-Validation) in our research to find the best accuracy. And we tested our kernel functions on a dataset called ―lungcancer.csv‖ from Kaggle and experimented on the dataset achieving good results in performance measures: accuracy, precision, sensitivity (Recall) and specificity. The dataset briefly talks about categories of COVID-19 surveillance case like Person without Symptoms (PWS) refers to people who show no symptoms, Person in Monitoring (PIM) refers to people under observation (suspected person) and Patient under Supervision (PUS) refers to patients under surveillance.