Construction genetic algorithm prediction model in breast cancer / liver cancer

• Main Authors: Wei-Pin Chang, 張偉斌
• Other Authors: Der-Ming Liou
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/39719888564623775416

博士 === 國立陽明大學 === 公共衛生研究所 === 96 === In recent years, Data Mining attracts great concern from information industries, its main reason is that a large amount of extant materials can be used extensively, and there are urgent demands to be changed these materials into useful information and knowledge. The information and knowledge obtained are admissible to improve and promote efficiency, the field used includes very much, and the application case that Data Mining in the medical field increases gradually. According to records from Department of Health, Breast cancer and Liver cancer were major manifestations among Taiwanese population leading to deaths of top ten causes in Taiwan. These two indications had some characteristics in common as increasing risk with increasing age and sharing the same pool of risk factors in our living environment. The central role of data mining uses artificial intelligence and statistical methods to extract meaningful information from puzzles of variables and data. The present study focused on the investigation of the application of artificial intelligence and data mining techniques to the prediction models of breast cancer and liver cancer. The artificial neural network, decision tree, logistic regression, and genetic algorithm were used for the comparative studies and the accuracy and positive predictive value of each algorithm were used as the evaluation indicators. 699 records acquired from the breast cancer patients, 729 records acquire from the liver cancer patient. In breast cancer data, 9 predictor variables, and 1 outcome variable were incorporated for the data analysis followed by the 10-fold cross-validation. The results revealed that the accuracies of logistic regression model were 0.9637 (sensitivity 0.9716 and specificity 0.9482), the decision tree model 0.9435 (sensitivity 0.9615, specificity 0.9105), the neural network model 0.9502 (sensitivity 0.9628, specificity 0.9273), the genetic algorithm model 0.9878 (sensitivity 1, specificity 0.9802). The accuracy of the genetic algorithm was significantly higher than the average predicted accuracy of 0.9612. The predicted outcome of the logistic regression model was higher than that of the neural network model but no significant difference was observed. The average predicted accuracy of the decision tree model was 0.9435 which was the lowest of all 4 predictive models. The standard deviation of the 10-fold cross-validation was rather unreliable. On other hand, liver cancer data include 12 predictor variables, and 1 outcome variable were incorporated for the data analysis followed by the 10-fold cross-validation. The results revealed that the accuracies of logistic regression model were 0.7658 (sensitivity 0.7682 and specificity 0.7630, the decision tree model 0.7636 (sensitivity 0.7497, specificity 0.7793), the neural network model 0.7760 (sensitivity 0.7875, specificity 0.7679), the genetic algorithm model 0.8072 (sensitivity 0.8444, specificity 0.0.763). The accuracy of the genetic algorithm was significantly higher than the average predicted accuracy of 0.7684. The predicted outcome of the neural network model was higher than that of the logistic regression model and decision model but no significant difference was observed. The present study indicated that the genetic algorithm model yielded better results than other data mining models for the analysis of the data of breast cancer and liver cancer patient in terms of the overall accuracy of the patient classification, the expression and complexity of the classification rule. The results showed that the genetic algorithm described in the present study was able to produce accurate results in the classification of breast cancer data/liver cancer data and the classification rule identified was more acceptable and comprehensible.