| Abstrakt: |
Lung cancer has a substantially worse five-year survival rate than many other malignancies and is the most common cause of cancer-related deaths in both men and women worldwide. For better disease detection and medical management, accurate survival analysis is urgently required. In the literature, few works are reviewed for survival analysis, but it fails to achieve optimal outcomes. Hence, in this paper, Cox Proportional Hazard Analysis Based Deep Learning Model (CPHA-DLM) is developed for risk identification in lung cancer detection. The proposed method is proceeding with two stages such as lung cancer detection and risk identification of patients with the basis of survival rate. At first, the databases are collected from the SEER program. The main motive of the research is survival analysis which is achieved by considering Cox Proportional Hazard Analysis. Initially, lung cancer is detected by considering the deep learning model. The databases are sent to the deep learning model of the Hybrid Convolutional Neural Network (HCNN). The deep learning model is a grouping of a Convolutional Neural Network (CNN) and Cat and Mouse based Optimizer (CMO). In CNN, the hyperparameter is optimized with the consideration of the CMO. After that, the survival rate of the patients is analyzed with hazard analysis. To compute the predictive power of the survival model, two measures are considered as concordance index and Kaplan Meier Estimate. The proposed method is validated by considering the conventional approaches. According to this study, the patient has a low risk after 20 years. The patient has a medium risk at 8 years and a high risk after 5 years, respectively. Experimental results show that the proposed method attained the maximum Precision of 96.29%, recall of 96.10%, and F-Measure of 96.16%. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink