目的利用机器学习随机森林(random forest,RF)算法,建立一种准确识别检测小偏移的实时质控方法,并以浮动异常值之和的方法(moving sum of outlier,MovSO)作为参比方法,评价新算法效能。方法收集来自航空总医院实验室信息系统导出的2016...目的利用机器学习随机森林(random forest,RF)算法,建立一种准确识别检测小偏移的实时质控方法,并以浮动异常值之和的方法(moving sum of outlier,MovSO)作为参比方法,评价新算法效能。方法收集来自航空总医院实验室信息系统导出的2016年1月至2021年8月在罗氏化学发光E601设备检测的hs-cTnT项目检测结果,按照规定的数据清洗规则筛选出54243个结果作为无偏数据,人为引入10个不同大小的偏移,生成相应的有偏数据,每种偏移下用RF与MovSO两种算法进行实验。采用分类模型标准及临床指标对算法进行评价。结果RF算法在浮动窗口大小为10时,对10个小偏移均能检出,假阳性率(FPR)为4.0%~4.7%,受影响的患者样本数中位数(MNPed)在12以下。除了在±1 ng/L偏移时准确度为85%,其余8个偏移检出准确度均在90%以上;MovSO算法的最优浮动窗口大小为200,除了在1 ng/L时偏移无法检出,对其他9个偏移均可检出,FPR在3.5%~4.6%之间,MNPed均在100以上,仅在5 ng/L偏移时识别准确度方可达89%。RF算法总体显著优于MovSO,RF可准确识别hs-cTnT的测量小偏移。结论基于机器学习RF算法建立的质控方法可以改进类似hs-cTnT等临床对检测质量要求较高的项目的测量准确度,为实验室质控方案提供了新思路。展开更多
The molecular chain structure and crystalline structure of PB-ET copolyester made by con-densation polymerization are studied by means of <sup>13</sup>C nuclear magnetic resonance spectroscopy(<sup>1...The molecular chain structure and crystalline structure of PB-ET copolyester made by con-densation polymerization are studied by means of <sup>13</sup>C nuclear magnetic resonance spectroscopy(<sup>13</sup>C NMR),dynamic mechanical analysis （DMA）,wide angle X-ray scattering （WAXS）,depolarized light intensity （DLI） and stress-strain measurement（S-S）.It is shown that PB-ETcopolyester is a random copolymer.Because of different ratios of EG（ethylene glycol）to BD（butane diol）in the copolyester,its crystalline forms and mechanical properties vary remarkably.With the increasing amount of BD,its crystallization kinetic parameters n,k and t<sub>1/2</sub><sup>-1</sup> increaserapidly.When EG/BD equals 25/75 （mole ratio）,the copolyester has αβ transition as PBTdoes.展开更多
With the rapid development of cryptography, the strength of security protocols and encryption algorithms consumedly relies on the quality of random number. In many cryptography applications, higher speed is one of the...With the rapid development of cryptography, the strength of security protocols and encryption algorithms consumedly relies on the quality of random number. In many cryptography applications, higher speed is one of the references required. A new security random number generator architecture is presented. Its philosophy architecture is implemented with FPGA, based on the thermal noise and linear feedback shift register(LFSR). The thermal noise initializes LFSRs and is used as the disturbed source of the system to ensure the unpredictability of the produced random number and improve the security strength of the system. Parallel LFSRs can produce the pseudo-random numbers with long period and higher speed. The proposed architecture can meet the requirements of high quality and high speed in cryptography.展开更多
文摘目的利用机器学习随机森林(random forest,RF)算法,建立一种准确识别检测小偏移的实时质控方法,并以浮动异常值之和的方法(moving sum of outlier,MovSO)作为参比方法,评价新算法效能。方法收集来自航空总医院实验室信息系统导出的2016年1月至2021年8月在罗氏化学发光E601设备检测的hs-cTnT项目检测结果,按照规定的数据清洗规则筛选出54243个结果作为无偏数据,人为引入10个不同大小的偏移,生成相应的有偏数据,每种偏移下用RF与MovSO两种算法进行实验。采用分类模型标准及临床指标对算法进行评价。结果RF算法在浮动窗口大小为10时,对10个小偏移均能检出,假阳性率(FPR)为4.0%~4.7%,受影响的患者样本数中位数(MNPed)在12以下。除了在±1 ng/L偏移时准确度为85%,其余8个偏移检出准确度均在90%以上;MovSO算法的最优浮动窗口大小为200,除了在1 ng/L时偏移无法检出,对其他9个偏移均可检出,FPR在3.5%~4.6%之间,MNPed均在100以上,仅在5 ng/L偏移时识别准确度方可达89%。RF算法总体显著优于MovSO,RF可准确识别hs-cTnT的测量小偏移。结论基于机器学习RF算法建立的质控方法可以改进类似hs-cTnT等临床对检测质量要求较高的项目的测量准确度,为实验室质控方案提供了新思路。
文摘The molecular chain structure and crystalline structure of PB-ET copolyester made by con-densation polymerization are studied by means of <sup>13</sup>C nuclear magnetic resonance spectroscopy(<sup>13</sup>C NMR),dynamic mechanical analysis （DMA）,wide angle X-ray scattering （WAXS）,depolarized light intensity （DLI） and stress-strain measurement（S-S）.It is shown that PB-ETcopolyester is a random copolymer.Because of different ratios of EG（ethylene glycol）to BD（butane diol）in the copolyester,its crystalline forms and mechanical properties vary remarkably.With the increasing amount of BD,its crystallization kinetic parameters n,k and t<sub>1/2</sub><sup>-1</sup> increaserapidly.When EG/BD equals 25/75 （mole ratio）,the copolyester has αβ transition as PBTdoes.
基金National Natural Science Foundation of China(60373087 and 90104005) Foundation for Doctoral SpecialBranch by Ministry of Education of China(20020486046)
文摘With the rapid development of cryptography, the strength of security protocols and encryption algorithms consumedly relies on the quality of random number. In many cryptography applications, higher speed is one of the references required. A new security random number generator architecture is presented. Its philosophy architecture is implemented with FPGA, based on the thermal noise and linear feedback shift register(LFSR). The thermal noise initializes LFSRs and is used as the disturbed source of the system to ensure the unpredictability of the produced random number and improve the security strength of the system. Parallel LFSRs can produce the pseudo-random numbers with long period and higher speed. The proposed architecture can meet the requirements of high quality and high speed in cryptography.
