Comparison of feed forward, Cascade forward and Layer Recurrent Algorithm model for breast cancer prediction

Abstract

The number of cases of Breast cancer seems to be constantly increasing worldwide in the recent years. According to 2012 world cancer report, the breast cancer incidence rate in Asia appears to be 37.5 and the mortality rate is 13.21. As per the recent statistics report of 2018, it has been estimated that the number of new cases of breast cancer is 14% in India. The percentage is higher when compared to lung cancer, oral cancer and other types of cancers. It has been identified that it occurs mostly in the age group between 41-50 years and the distribution of this is found to be 42%, whereas it is 18%, 24% for the age groups of 31-40 and 51-60 respectively. The five stages of survival include the survival rate of 100%, 98%, 88%, 52%, 16% for each stage respectively. Diagnosis in the early stage can reduce the death rate. Even during the stage 3 if the cancer is predicted, it can be treated. Machine learning using Artificial Neural Network(ANN) techniques can be effectively utilised for the prediction of cancer. In the proposed method, networks designed using various ANN Algorithm are used to predict the breast cancer if it is benign or malignant. Three different networks such as feedforward back prop, Cascade forward and layer recurrent has been implemented. The performance measures such as Accuracy, error, specificity, sensitivity, positive predictive value and negative predictive value are obtained. From the table 2, it is inferred that among these algorithms, feed-forward algorithm has better performance compared to other two algorithms.ANN has been a powerful tool for analyzing the data when there are non-linear interactions between the input and the output to be predicted. The results show that the accuracy of ANN for the prediction was better than other approaches. The Wisconsin breast cancer database has been used for the analysis. The results are very competitive and can be used for diagnosis, prognosis, and treatment.