Deep Learning Algorithm for Prediction of Brain Diseases Using Iris Image


With the rapid development of ICT technology along with the 4th industrial revolution, in the field of Traditional Korean medicine. we also analyze data on obesity and complications for preventive management of yoyo phenomenon, custom medical treatment such as improvement of complications, and medical examination. Various ICT technologies have been introduced and studied. From the viewpoint of Traditional Korean medicine, iris is an institution that expresses the singularity of the human body. Iris is directly connected to the brain, and the biomarkers displayed on the body are reflected in the iris. In this study, we try to classify the brain disease-related biomarkers indicated by the iris by learning the iris image-based map of the causes of dementia induction in brain diseases. As a result of the experiment using the proposed algorithm, the learning time is about 26 minutes, the learning accuracy is 99%, the loss rate is 1%, the accuracy of the test data is 91%, and the loss rate is 17.4%


Based on Danhak, a classification algorithm for brain disease, one of the causes of dementia, was studied. For the classification algorithm, CNN, one of the deep learning classification techniques, was used, and the number of filters, kernel size, batch normalization, activation function, and loss were used. The experiment was carried out by adjusting the parameter values ​​such as functions. As a result, the learning accuracy was 99%, the loss rate was 1%, the accuracy was 91% for the test data, and the loss rate was 17.4%. The learning time was about 26 minutes. As a result of evaluation using the classification evaluation index, the accuracy was 91%, the precision was 99%, the recall was 83%, and the F1 score was 90%. Through the above results, a model with high loss rate but high accuracy can be created and classified, and the reliability of the model is secured through precision, recall, and F1 Score. The purpose of this study is to analyze iris biomarkers for diabetes, old age, external factors, etc., and develop a diagnostic assistance algorithm through biomarkers displayed on the iris. If future research continues and biomarkers for the causes of dementia are sorted out, it is expected that dementia can be prevented by estimating the possibility of its occurrence at an early stage.

Jin-Beom Seo1· Young-Bok Cho2*1bachelor’s course, Department of information Security, Daejeon University, Daejeon, 34520 Korea2Professor, Department of information Security, Daejeon University, Daejeon, 34520 Kore

Read Full Abstract (Korean Language):