With the improvement of living standards and changes in work habits caused by industrialization, the prevalence of diseases linked to lifestyle is rising. In this context, the prevention of lifestyle-related diseases (LRDs ) is extremely important. The majority of existing research exclusively concentrates on the prognosis of a particular LRD sickness, making it impossible for them to intelligently identify the important characteristics of the disease. Therefore, this study aims to propose a lifestyle-related disease prediction framework including three key components, called missing value module, a feature selection module, and a disease prediction module. The performance of the proposed framework is evaluated by using real medical data gathered during a hospital health check-up in Nanjing, China. The experiment shows that the proposed framework can automatically generate prediction ensemble models for specific LRDs diseases, and achieve good accurate performance.