Abstract: Now a days clinical data mining is used for clinicians in order to provide diagnosis, therapy and prognosis of different diseases. The accuracy of clinical-outcome prediction has been increased by using multiple measurements which are gathered from different time period and dataset. The multiple measurements are merged by using merging algorithm and the distribution of data is determined by statistical measurement. Then those data are given to the classifier for predicting the recurrence and non-recurrence of Hepatocellular Carcinoma (HCC) patients. In this study, an improved multiple time series clinical data processing is proposed. In the proposed approach, an additional measurement feature according to the frequency interval of features is included for reducing the error rate of classifier and increasing the prediction rate. The frequency based measurement feature is computed based on curvelet transform. Then, the optimal features are selected based on the Firefly optimization algorithm for reducing the classification overhead. The selected optimal features are learned by using the Support Vector Machine (SVM) classifier for predicting the patients with HCC disease and patients without HCC effectively. Finally, the experimental results prove that the proposed method has better performance than other classification methods.
P. Radha and R. Divya, 2017. A HCC Recurrence Prediction in Multiple Time Series Clinical Data with Merging Statistical Measures of Advanced Frequency Spectrum of Time Series Features. Journal of Engineering and Applied Sciences, 12: 5473-5477.