Prognostic Systems for NPC: A comparison of the Neural Network Model and The Cox Proportional Hazards Model

• Main Authors: Abdul-Kareem, S., Baba, S., Zubairi, Y.Z., Prasad, U., Wahid, M.I.A.
• Format: Article
• Published: 2002
• Subjects: T Technology (General)
• Online Access: http://eprints.um.edu.my/5694/; http://www.wseas.us/e-library/conferences/switzerland2002/papers/409.pdf
• Institution: Universiti Malaya

Statistical methods such as the life-table, the Kaplan-Meier method and regression models, such as the Cox Proportional Hazard are usually used to model and predict survival data. Neural networks have been used in medicine for more than two decades, first, as an aid to diagnosis and treatment and then, recently, as a tool to study medical prognosis of aids, coronary heart disease, and a variety of cancer types. The use of neural networks to study the prognosis of nasopharyngeal carcinoma is, however, relatively new. In this paper we describe our research in the use of neural network to predict the prognosis of nasopharyngeal carcinoma. Two prognostic models for nasopharyngeal carcinoma were developed, namely the neural network model and the Cox model and their performance compared.