科研之友
微信
新浪微博
Facebook
分享链接
摘要
Deep learning, which is a promising end-to-end learning method for accurate health diagnosis, can extract effective representations directly from high-dimensional sensory inputs without manually feature engineering. However, it is still not enough for the end-to-end dynamic process of healthcare, which needs to derive effective representations of human body from miscellaneous observations and generalize past experience to new situations. This paper proposes an end-to-end framework that emulates the process of healthcare. There are two major modules in our framework. The first one is the deep recognition module to diagnose the health states based on the deep neural networks (DNNs). The second module is the action evaluation module based on the Bayesian inference graphs. A simulation environment is designed in order to evaluate the framework. It includes a body simulator to produce the body instance that can receive treatments, and the latent health state of the simulated patient will be changed by different interventions. It also includes a deep recognition module and an action evaluation module to nurse the body. The experiments demonstrate that the healthcare process under our framework can become more and more efficient as the time goes on with the increasing statistic data.
- 出版日期2020-11
- 单位广州大学
