Predicting Medical Expenditure and Survivability of the Comorbidities Patients before Diagnosed with Lung Cancer by Bayesian Network

• Main Authors: Ting-Yang Su, 蘇庭揚
• Other Authors: Kung-Jeng Wang
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/mcvw44

碩士 === 國立臺灣科技大學 === 工業管理系 === 105 === Lung cancer is already considered to be the leading cause of tumor-related death in the world. It is also the most common cancer death in Taiwan and medical cost of lung cancer is the highest among top ten cancer in 2015. This study collected cases of which patients were diagnosed with comorbidities before diagnosed lung cancer from 1996 to 2010 in Taiwan National Health Insurance Research Database, and 2,875 cases were selected as experimental data. In addition, conditional Gaussian Bayesian network was proposed to evaluate survival time and medical cost of experimental data. Risk factors were used to construct Bayesian network model and Kaplan–Meier estimate was used to adjust censored data. The R2 of result of survival time prediction is 62.37% and the R2 of result of medical expenditure prediction is 23.84%. The proposed model is not only useful to predict the survival probability and medical expenditure of patients with lung cancer, but it can also calculate the posterior probabilities of variety of medical-related query.