Using Predictive Analytics for Cancer Identification

Abstract

Malignant pleural mesothelioma (MPM) or malignant mesothelioma (MM) is an atypical, aggressive tumor that matures into cancer in the pleura, a stratum of tissue bordering the lungs. MPM accounts for about 75% of all mesothelioma diagnosed yearly in the United States. Given the difficulty of diagnosing MPM, early identification is crucial for patient survival. Our study implements predictive analytics and recommends the best fit model for early diagnosis and prognosis of MPM. We retrieved medical reports generated by Dicle University (Turkey) and implemented the following predictive analytics methods for the early stage diagnoses of MPM: the Stochastic Gradient Descent, Adaptive Boosting, Kernel Logistic Regression, Multi-layer Perceptron Classifier, Voted Perceptron, Hoeffding Tree, and the Clojure Classifier. The performance of these predictive methods were evaluated using the following measure: accuracy, f -measure, recall, precision, root mean squared error, receiver operating characteristic, and precision- recall curve. Among the predictive methods tested, Adaptive Boosting had the highest accuracy with 71.29%. Results indicate improved MPM diagnosis through the use of predictive analytics methods with the accuracy of testing and training dependent upon the specific application.

Keywords: mesothelioma, predictive modeling, cancer detection