Heart disease has become one of the most dangerous diseases all over the world because it can't be discovered at an early time due to the lack of medical resources in rural areas. Recently, deep learning technologies offer ways to diagnose heart diseases automatically which may benefit millions of people in rural areas.
In this talk, we will introduce current ways to diagnose heart diseases and its pros and cons. We will then elaborate on our heart diseases diagnosis platform in details. This section includes the requirements form the data annotation and storage to train the models, the infrastructure we use and detailed design of our coarse-grain diagnosis model at the edge side and fine-grained diagnosis model in the cloud. Next, we will demo our heart disease diagnoses platform and conclude our experience in building a deep learning platform which is suitable for the remote healthcare.