Prediction of Alzheimer's disease using deep learning. Lead Investigator: Takuya Sakurai Institution : Fuji Xerox Co., Ltd. E-Mail : sakurai.takuya@fujixerox.co.jp Proposal ID : 1354 Proposal Description: Recently, dementia is very important problem in Japan. Because the aging population of Japan is increasing at an alarming rate and the patients of dementia will probably continue to grow more and more in the future. So we are trying to predict the state of diagnosis. We will apply our multimodal approach which we have developed for visual question and answering task to this one using both longitudinal image and blood test information. According to past papers, they almost used the volumes of the brain and the hippocampus from MR image. But, we think MR image have more powerful information and there is room for improvement to extract feature from MR image. We are planning to improve the prediction precision by using deep learning, which will need a big data set for training and validation. NACC data will be very helpful for our research.