| Diabetes is one of the most important chronic diseases that threaten the global human health.In 2017,there were 114 million people in China who were suffering from diabetes,ranking first in the world,and the undiagnosed rate was as high as53.6%.China will certainly face the increasing pressure for diagnosing and treating diabetes.Strengthening prevention,timely diagnosis,early treatment,construction of a diabetes research system,and improvement of diabetes diagnosis and treatment technology are important issues that China needs to pay attention to for a long time.The accuracy of traditional diagnosis of diseases is obviously affected by individual factors of doctors.The diagnosis of diabetes is complex and diverse.It will inevitably lead to some misdiagnosis,which will cost optimal treatment time and trigger a series of irreversible consequences.The rapid development of computer technology has provided a basis for changing this traditional situation.Several researches have shown that computer technology can effectively assist in the diagnosis of diseases.Through extensive literature analysis,it is found that the existing papers have explored disease diagnosis using basic machine learning algorithms such as decision trees,but there are still many shortcomings.Therefore,this paper introduces a fully-connected neural network and a convolutional neural network based on deep learning for diabetes diagnosis system.This research collected some existing diabetes diagnostic data and cleaned it by exploratory data analysis and data preprocessing methods.Then settled the features for diagnosis combined with the common sense of diabetes diagnosis and generated labeled samples for model training.The training set,test set,and validation set are divided according to a ratio of 7: 2: 1.The two models above were all trained and verified through these labeled samples with early stop and dropout methods that could make sure the models trained were not over fitting.The test results were pretty good,which further verified the positive role of deep learning in disease diagnosis. |
