|Abstract: ||近年深度學習對醫療領域有深遠的影響，深度學習結合醫療影像應用技術，且運用機器學習能力快速及精準地判斷大量的醫療影像數據，協助醫生提高疾病診斷的正確率，更進而以相關的病理資訊來預測分析疾病發生風險與機率，全球學者仍持續研究如何運用深度學習的能力應用在各項智慧醫療發展上大腦疾病複雜又難以處理，腦部疾病中的一類為失智症，失智症中最常見的為阿茲海默症(Alzheimer′s disease ,AD)，佔失智病患人數約50%-70%，目前尚無實證有效的藥物治療方式，因此目前研究方向朝向延緩病程的惡化進行。而腦中風為腦部疾病中常見的神經中樞疾病，腦中風患者在入院前的病理狀態影響手術後的存活機率，本篇論文以深度學習模型可對不同類型醫療數據資料進行執行推斷能力，來預測阿茲海默症之惡化與腦中風手術後存活率。|
;In recent years, deep learning has a far-reaching impact on the medical field, deep learning combined with medical imaging application technology, and the use of machine learning ability to quickly and accurately judge a large number of medical imaging data, to help doctors improve the correct rate of disease diagnosis, and then with relevant pathological information to predict and analyze the risk and probability of disease occurrence, global scholars continue to study how to use the ability of deep learning in the development of intelligent medical.
Brain diseases are complex and difficult to deal with, one of the brain diseases is dementia, the most common of which is Alzheimer′s disease, accounting for about 50%-70% of the number of dementia patients, there is no proven and effective drug treatment, so the current research direction towards delaying the deterioration of the course of the disease. Stroke is a common neurocreatic disease in brain diseases, and the pathological state of stroke patients before admission affects the chances of survival after surgery, this paper uses the deep learning model to make inferences about different types of medical data to predict the deterioration of Alzheimer′s disease and the survival rate after stroke surgery. In this paper, the random forest of machine learning, Gradient Boosting Trees, SVM and deep learning are used to compare the effectiveness by predicting the deterioration results of Alzheimer′s disease, and the recipient distinguishes the data set for the patient′s pathological information and the air pollution information, in an attempt to find out the risk factors for the deterioration of Alzheimer′s disease. In the same model, modules were created to evaluate the patient′s physiological indices for the pre-hospital stroke evaluation of stroke patients to see if they had an impact on post-operative survival.