CAJ | 학술논문

经验医学向循证医学( Evidence Based Medicine，EBM)的转变是临床研究与实践的一大趋势。循证医学采用大样本随机对照试验(Randomized Controlled Trail，RCT)和Meta分析技术，在较大程度上保证了疾病治疗方案有效安全可靠。但是，RCT的基础是随机采样理论，试图用最少的数据获得最有用的信息，这可能导致证据偏倚及治疗方案失信。随着医疗大数据时代的到来，循证医学的上述局限与不足可望得到根本改进，并朝证据采集、制作与评估自动化的方向发展。面向大数据的深度学习是一种自动化的特征提取技术，能从海量高维数据自动提取病因并建立高精度的决策分析模型，从而大大提高循证医学证据分析与决策支持的效率，促进循证医学在医学各领域更加广泛地应用和推广。
경험의학향순증의학( Evidence Based Medicine，EBM)적전변시림상연구여실천적일대추세。순증의학채용대양본수궤대조시험(Randomized Controlled Trail，RCT)화Meta분석기술，재교대정도상보증료질병치료방안유효안전가고。단시，RCT적기출시수궤채양이론，시도용최소적수거획득최유용적신식，저가능도치증거편의급치료방안실신。수착의료대수거시대적도래，순증의학적상술국한여불족가망득도근본개진，병조증거채집、제작여평고자동화적방향발전。면향대수거적심도학습시일충자동화적특정제취기술，능종해량고유수거자동제취병인병건립고정도적결책분석모형，종이대대제고순증의학증거분석여결책지지적효솔，촉진순증의학재의학각영역경가엄범지응용화추엄。
The transforming from empirical medicine (EM) to evidence based medicine (EBM) becomes a big trend in recent years in clinical research and practice. In EBM, the randomized controlled trail (RCT) and Meta analysis in a great degree ensure the effectiveness, safety and reliability of therapeutic regimens. However, the RCT is based on random sampling theory, it tries to gain most useful information from least data, this may cause the problems of bias evidences and suspicious regimens. With the coming of big data era, the above defects of EBM are expected to be thoroughly remedied, and EBM will continue its development towards automatic evidence collection, making and evaluation. The big data oriented deep learning is a kind of automatic feature extraction technique, which can automatically extract disease features from high dimension large scale data and build high accuracy decision making models, thus it can greatly enhance the efifciency of evidence analysis and decision making for EBM, and promote the application and popularization of EBM into all medicine ifelds more far and wide.