With 3, 3’5, 5’-tetramethylbenzidine（TMB） as the detection substrate, a reliable and highly selective method was established and optimized for the determination of Lactoperoxidase（LP） activity in raw milk. The m...With 3, 3’5, 5’-tetramethylbenzidine（TMB） as the detection substrate, a reliable and highly selective method was established and optimized for the determination of Lactoperoxidase（LP） activity in raw milk. The method was based on the enzymatic reaction principle, where hydrogen peroxide oxidated TMB in the presence of LP. The optimized conditions of this assay system were obtained, consisting of 20 mmol · L-1 TMB solution, 0.6 mmol · L-1 hydrogen peroxide and 0.1 mol · L-1 Citric Acid（CA）/0.2 mol · L-1 disodium hydrogen phosphate（Na P） buffer（pH 4.8）. TMB detection method was applied to the analysis of LP in milk samples with a practical working concentration range from 2 to 14 mg · L-1. The intra-and inter-batch variation coefficients were all below 5%, indicating a good repeatability. Confirmation test between TMB method and 2, 2-azinobi（3-ethylbenzothiazoline-6-sulphonate） diammonium salt（ABTS） method was carried out, and the results of TMB assay were in accordance with that of ABTS method.展开更多
基于股指成分股基本面和技术面数据构建了时序股票关联网络,然后利用深度图神经网络学习股票关联网络层次化表征,以端到端的方式获得候选预测信号.在此基础上,提出了一种考虑动作评估反馈的深度强化学习方法(Action Evaluation Feedback...基于股指成分股基本面和技术面数据构建了时序股票关联网络,然后利用深度图神经网络学习股票关联网络层次化表征,以端到端的方式获得候选预测信号.在此基础上,提出了一种考虑动作评估反馈的深度强化学习方法(Action Evaluation Feedback based Deep Q-Learn-ing,AEF-DQN),旨在将不同的候选预测信号融入智能体的动作空间,并基于股票关联网络层次化表征、股票市场整体运行状态和历史动作评估反馈学习环境状态;借鉴前景理论中的参照依赖特性估计奖励值函数,从而建立状态、动作与奖励值之间的映射关系.最后,采用沪深300指数、标普500指数、英国富时100指数和日经225指数的成分股历史数据,构造了股指期货交易模拟器,在投资胜率、最大回撤率、阿尔法比率和夏普比率4个回测指标上对股指预测模型展开实证分析.研究结果表明:1)通过层次化聚合股票关联网络的节点属性信息可以动态捕捉不同行业对股指价格波动的影响,进而可提升预测方法的准确率;2)考虑动作评估反馈的深度强化学习结构可智能化选择适用于当前股票市场环境的最优模型结构,进而可提升预测方法的鲁棒性.展开更多
Objective:To investigate the prognostic value of radiomics features based on ^(18)F-FDG PET/CT imaging for advanced non-small cell lung cancer(NSCLC)treated with chemotherapy.Methods:A sample of 146 NSCLC patients sta...Objective:To investigate the prognostic value of radiomics features based on ^(18)F-FDG PET/CT imaging for advanced non-small cell lung cancer(NSCLC)treated with chemotherapy.Methods:A sample of 146 NSCLC patients stagedⅢor stageⅣwere included in this retrospective study who received ^(18)F-FDG PET/CT before treatment.All patients were treated with standardized chemotherapy after PET/CT examination and were divided into training group and validation group in an 8:2 ratio randomly.Radiomics features were extracted.In the training group,the minimum absolute contraction and selection operator(LASSO)algorithm and Cox risk proportional regression model were used to screen radiomics and clinical prognostic factors of progression-free survival(PFS).The radiomic model,clinical model and complex model were established respectively.The corresponding scores were calculated,then verified in the validation group.Results:The LASSO algorithm finally screened four radiomics features.ROC results showed that in the training group,the AUC of PFS predicted by the radiomics model was 0.746,and that in the verification group was 0.622.COX multivariate analysis finally included three clinical features related to PFS in NSCLC patients,namely pathological type,clinical stage and MTV30.The AUC for predicting PFS by clinical model,radiomics model and composite model were 0.746,0.753 and 0.716,respectively.The radiomics model had the highest diagnostic efficacy,and its sensitivity and specificity were 0.663 and 0.833,respectively.Delong test verified that there was no statistical difference in the predictive efficacy between the radiomics model and the composite model(Z=1.777,P=0.076)and the clinical imaging model(Z=0.323,P=0.747).Conclusion:The radiomics model based on PET/CT has a good predictive value for the prognosis of advanced NSCLC treated with chemotherapy,but it needs further validation before it can be widely used in clinical practice.展开更多
基金Supported by Project for Research and Development of Harbin Aapplication Technology(2016RAQXJ046)the National "Twelfth Five-year" Plan for Science and Technology Support Program of China(2013BAD18B06)
文摘With 3, 3’5, 5’-tetramethylbenzidine（TMB） as the detection substrate, a reliable and highly selective method was established and optimized for the determination of Lactoperoxidase（LP） activity in raw milk. The method was based on the enzymatic reaction principle, where hydrogen peroxide oxidated TMB in the presence of LP. The optimized conditions of this assay system were obtained, consisting of 20 mmol · L-1 TMB solution, 0.6 mmol · L-1 hydrogen peroxide and 0.1 mol · L-1 Citric Acid（CA）/0.2 mol · L-1 disodium hydrogen phosphate（Na P） buffer（pH 4.8）. TMB detection method was applied to the analysis of LP in milk samples with a practical working concentration range from 2 to 14 mg · L-1. The intra-and inter-batch variation coefficients were all below 5%, indicating a good repeatability. Confirmation test between TMB method and 2, 2-azinobi（3-ethylbenzothiazoline-6-sulphonate） diammonium salt（ABTS） method was carried out, and the results of TMB assay were in accordance with that of ABTS method.
文摘基于股指成分股基本面和技术面数据构建了时序股票关联网络,然后利用深度图神经网络学习股票关联网络层次化表征,以端到端的方式获得候选预测信号.在此基础上,提出了一种考虑动作评估反馈的深度强化学习方法(Action Evaluation Feedback based Deep Q-Learn-ing,AEF-DQN),旨在将不同的候选预测信号融入智能体的动作空间,并基于股票关联网络层次化表征、股票市场整体运行状态和历史动作评估反馈学习环境状态;借鉴前景理论中的参照依赖特性估计奖励值函数,从而建立状态、动作与奖励值之间的映射关系.最后,采用沪深300指数、标普500指数、英国富时100指数和日经225指数的成分股历史数据,构造了股指期货交易模拟器,在投资胜率、最大回撤率、阿尔法比率和夏普比率4个回测指标上对股指预测模型展开实证分析.研究结果表明:1)通过层次化聚合股票关联网络的节点属性信息可以动态捕捉不同行业对股指价格波动的影响,进而可提升预测方法的准确率;2)考虑动作评估反馈的深度强化学习结构可智能化选择适用于当前股票市场环境的最优模型结构,进而可提升预测方法的鲁棒性.
基金Research and Cultivation Foundation of Hainan Medical College(HYPY2020022)Hainan Natural Science Foundation Youth fund(822QN482)+1 种基金Doctoral Research Fund project of Hainan Cancer Hospital(2022BS04)Key R&D projects in Hainan Province(ZDYF2021SHFZ244)。
文摘Objective:To investigate the prognostic value of radiomics features based on ^(18)F-FDG PET/CT imaging for advanced non-small cell lung cancer(NSCLC)treated with chemotherapy.Methods:A sample of 146 NSCLC patients stagedⅢor stageⅣwere included in this retrospective study who received ^(18)F-FDG PET/CT before treatment.All patients were treated with standardized chemotherapy after PET/CT examination and were divided into training group and validation group in an 8:2 ratio randomly.Radiomics features were extracted.In the training group,the minimum absolute contraction and selection operator(LASSO)algorithm and Cox risk proportional regression model were used to screen radiomics and clinical prognostic factors of progression-free survival(PFS).The radiomic model,clinical model and complex model were established respectively.The corresponding scores were calculated,then verified in the validation group.Results:The LASSO algorithm finally screened four radiomics features.ROC results showed that in the training group,the AUC of PFS predicted by the radiomics model was 0.746,and that in the verification group was 0.622.COX multivariate analysis finally included three clinical features related to PFS in NSCLC patients,namely pathological type,clinical stage and MTV30.The AUC for predicting PFS by clinical model,radiomics model and composite model were 0.746,0.753 and 0.716,respectively.The radiomics model had the highest diagnostic efficacy,and its sensitivity and specificity were 0.663 and 0.833,respectively.Delong test verified that there was no statistical difference in the predictive efficacy between the radiomics model and the composite model(Z=1.777,P=0.076)and the clinical imaging model(Z=0.323,P=0.747).Conclusion:The radiomics model based on PET/CT has a good predictive value for the prognosis of advanced NSCLC treated with chemotherapy,but it needs further validation before it can be widely used in clinical practice.
