Disease Prediction using Convolution Neural Network with Radial Basis Function

Breast cancer is the second most occurring cancer in women, therefore detection of this particular type of cancer at an early stage is important. A small increase in breast cancer will risk in people with Parkinson's Disease (PD) and therefore Breast cancer screening is recommended to the people who are suffering from PD. The occurrence of PD in India is expected to double by 2030 and the main objective is to predict the PD and breast cancer diseases by using the available technological advancements to significantly improve the patient's survivability. The present research collects the relevant dataset from the standard repository database called UCI machine learning for an effective prediction process. The existing methods were not able to handle more complex datasets as they include noisy and imbalanced data. The proposed Convolution Neural Network with Radial Basis Function (CNN-RBF) is used for the present research to handle the issues of complex noisy data and imbalance data. The results attained from the proposed CNN-RBF showed robust as CNN extracts the prominent features for disease prediction and RBF fits the feature values in the dimension space thereby overcoming the problem of imbalance and noisy data. The proposed CNN-RBF obtained better accuracy of 98.97 % for the PD prediction and 97.98 % of accuracy for the breast cancer prediction when compared to the existing Deep CNN and RBF models.