| By the end of 2014, the population over the age of 60 has reached 212 million in our country, accounting for 15.5 percent of the total population, the population over the age of 65 years has reached 138 million, accounting for 10.1 percent of the total population, and keeping sustained growth trend. With the rapid growth of aging population all over the world, the health requirement of old people has caused widespread concerns. As the elderly getting older, the physiological function in old people decrease gradually, physical problems come to light gradually, the incidence of common diseases in the elderly population is increasing. The health assessment of old people not only can help them understand their own health status and take early prevention for some diseases, at the same time can also provide scientific guidance for formulating corresponding health policies and improving the quality of old people’s life. Health assessment for old people refers to many aspects, for example, physical,mental, psychological, self-care ability and so on. Then we can valuate health status of old people objectively and effectively. The evaluation results can be used to identify some important problems including the state of body, psychology[1] and self-care ability to some extent. In order to explore more and better health assessment methods for old people, based on understanding of other assessment methods, this paper considers optimizing the prior probability of Bayesian method by using artificial neural network that can increase accuracy of posterior probability to realize the purpose of judging the influence degree of each factor on some disease of the elderly. The main work of this paper is depicted as follows:First, this paper has a clear understanding of the elderly health assessment and disease diagnosis methods, and introduces the research situation at home and abroad in detail to learn the evaluation process and advantages and disadvantages of each method.The combination of neural network and Bayesian approach is proposed to evaluate the health of old people(for a certain disease), and its feasibility will be predicted and verified. According to the experimental simulation results, we will evaluate advantages and shortages of the new method, and whether it can be extended to use in other similar diseases.Second, The development of artificial neural network and Bayesian network,basic concept, principle, algorithm and its application in health assessment and disease diagnosis are summarized, then a suitable network model is built on the understanding of the relevant theory according to actual situation.Third, the selection of BP artificial neural network model is introduced in detail in this papers, including the source of data, the pretreatment of data, the setting of training parameters, the process and results of training. And how to convert the weight vector to weights and priori probabilities by formulas is introduced. Then we analysis the rationality of the posterior probability and speculate the biggest factor in a certain disease for elderly people. The model has the features of convenient construction,strong adaptability,simple training process and excellent performance. |
