论文部分内容阅读
PURPOSE In the work, we investigate the use of support vector machines(SVM) to analyze the flow eytometry based acute myeloid leukemia (AML) minimal residual disease (MRD) automatically, objectively, and standardly.METHODS From 2010 to 2012, 36 AML patients, which have positive expression of CD7 and been detected forMRD no less than twice, are selected to export the initial disease data and MRD review data in the form of 159 Flow Cytometry Standard 3.0 files.SVM is trained by each initial disease data to set 1 as the flag, while SVM is trained by 15 normal person data to set 0 as the flag.Based on the two training groups,the parameters are optimized, and the independent predictive model is builtto analyze the MRDdata of each correspondingpatient.The automated analysis results of support vector machines are statistically compared with the conventional manual analysis results to determine the reliability.