A Hybrid Bayesian Network Model for Predicting Breast Cancer Prognosis

Abstract

Objective: Breast cancer is one of the most common cancers affecting women. Both physicians and patients have concerned about breast cancer survivability. Many researchers have studied the breast cancer survivability applying artificial nerural network model (ANN). Usually ANN model outperformed in classification of breast cancer survivability than other models such as logistic regression, Bayesian network (BN), or decision tree models. However, physicians in the fields hesitate to use ANN model, because ANN is a black-box model, and hard to explain the classification result to patients. In this study, we proposed a hybrid model with a degree of the accuracy and interpretation by combining the ANN for accuracy and BN for interpretation. Methods: We developed an artificial neural network, a Bayesian network, and a hybrid Bayesian network model to predict breast cancer prognosis. The hybrid model combined the artificial neural network and the Bayesian network to obtain a good estimation of prognosis as well as a good explanation of the results. The National Cancer Institute's SEER program public-use data were used to construct and evaluate the proposed models. Nine variables, which are clinically acceptable, were selected for input to the proposed models' nodes. A confidence value of the neural network served as an additional input node to the hybrid Bayesian network model. Ten iterations of random subsampling were performed to evaluate performance of the models. . The known risk factors are age at diagnosis, age of menarche, genetic risk, and family history [3][4][5][6] . The most widely accepted prognostic factor for breast cancer is the American Joint Commission on Cancer (AJCC) staging system based on the TNM system (T, tumor; N, node; M, metastasis) [7][8][9][10][11] . The Nottingham prognosis index is a grading system that incorporates the evaluation of tumor size, stage of disease, and tumor grade 12)13). Recently, molecular biologic markers, such as the estrogen receptor, progesterone receptor, and HER-2/neu, have become important prognostic factors [14][15][16] .Several studies have reported on breast cancer prognosis prediction using data mining techniques. Burke et al. . They used 6,787 cases of the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) breast carcinoma data set (1977)(1978)(1979)(1980)(1981)(1982) . Lee compared various predictive modeling techniques to predict the breast cancer occurance (not prognosis) by using Korean data obtained from 209 subject (109 patients and 100 controls).The AUC of the proposed Naïve Bayes model was 0.90, and it outperformed than other various Bayesian network models or regression models 19) .Lundin et al. compared logistic regression and ANN models for survival estimation in 951 breast cancer cases 20) . Eight variables were entered as input to the network, including tumor size, axillary node, histologic type, mitotic count, nuclear pleomorphism, tubule formation, tumor necrosis, and age. The area under the curve (AUC) of th...

A Survey on the Data Mining Techniques Used for Breast Cancer Predictions

“…The Data Mining Process [3] In [16], Choi et al compared the performance of an Artificial Neural Network, a Bayesian Network and a Hybrid Network used to predict breast cancer prognosis. The hybrid Network combined both ANN and Bayesian Network.…”

Analysis of Classification Techniques for Efficient Disease Prediction

“…Bayesian algorithm is the statistical classification algorithm which is based on Bayes theorem [15]. It assumes the class conditional independencies between the variables, but there exist dependencies when large health care datasets are used.…”

“…The network model is represented using Directed Acyclic Graph (DAG) [16] on a variable values and their conditional dependencies. For each variable it generates Conditional Probability Table (CPT) [17] for each variable which indicates all the possible combinational values of its parent.…”

Performance Analysis Of Various Data Mining Classification Techniques On Healthcare Data

“…Jong Pill Choi et al [22] compared the performance of an Artificial Neural Network, a Bayesian Network and a Hybrid Network used to predict breast cancer prognosis. The hybrid Network combined both ANN and Bayesian Network.…”