Survival Prediction of Cervical Cancer Patients using Genetic Algorithm-Based Data Value Metric and Recurrent Neural Network
Ojie, D.V1, Akazue M2, Omede E.U3, Oboh E.O4, Imianvan A5
1Ojie Deborah Voke, Department of Software Engineering, University Delta, C Agbor, Nigeria.
2Dr. Akazue M, Department of Computer Science, Delta State University, Abraka, Nigeria.
3Dr. Omede E. U, Department of Computer Science, Delta State University, Abraka, Nigeria.
4Dr. Oboh E.O, Department of Radiotherapy/ Clinical Oncology, University of Benin Teaching Hospital, Edo State.
5Prof. Imianvan A., Department of Computer Science, University of Benin, Benin, Edo Nigeria.
Manuscript received on 12 April 2023 | Revised Manuscript received on 20 April 2023 | Manuscript Accepted on 15 May 2023 | Manuscript published on 30 May 2023 | PP: 29-41 | Volume-13 Issue-2, May 2023 | Retrieval Number: 100.1/ijsce.B36080513223 | DOI: 10.35940/ijsce.B3608.0513223
Open Access | Editorial and Publishing Policies | Cite | Zenodo | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Survival analysis and machine learning are an indispensable aspect of disease management, as they enable practitioners to understand and prioritise treatment, particularly in terminal diseases. Cervical cancer is the most common malignant tumour of the female reproductive organs worldwide. Survival analysis, a time–to–event analysis for survival prediction, is therefore necessary for patients with cervical cancer. The Data Value Metric (DVM) is an information-theoretic measure that utilises the concept of mutual information and is a good metric for quantifying the quality and utility of data, as well as feature selection. This study proposed a hybrid of a Genetic Algorithm and a Data Value Metric for feature selection. At the same time, a Recurrent Neural Network and a Cox Proportional Hazard ratio were used to build the survival prediction model for managing cervical cancer patients. A dataset of 107 patients with cervical cancer was collected from the University of Benin Teaching Hospital in Benin, Edo State, and used to build the proposed model (RNN+GA-DVM). The proposed system outperforms the existing system, achieving an accuracy of 70% and an ROC score of 0.6041. In contrast, the proposed model yielded an accuracy of 75.16% and an ROC score of 0.7120, respectively. From this study, it was observed that the variables highly associated with cervical cancer mortality, as identified using the GA_DVM feature selection, are age_at_diagnosis, Chemotherapy, Chemoradiation, Histology, Comorbidity, Menopause, and MENO_Post. Thus, with early diagnosis and proper health management of cervical cancer, the age of survival of cervical cancer patients can be prolonged.
Keywords: Cervical cancer, Cox Proportional Hazard, Machine Learning, Survival Model.
Scope of the Article: Machine Learning