The acute effect of acupuncture on Alzheimer＇s disease,i.e.,on brain activation during treatment,has been reported.However,the effect of long-term acupuncture on brain activation in Alzheimer＇s disease is unclear.Th...The acute effect of acupuncture on Alzheimer＇s disease,i.e.,on brain activation during treatment,has been reported.However,the effect of long-term acupuncture on brain activation in Alzheimer＇s disease is unclear.Therefore,in this study,we performed long-term needling at Zusanli（ST36）or a sham point（1.5 mm lateral to ST36）in a rat Alzheimer＇s disease model,for 30 minutes,once per day,for 30 days.The rats underwent 18F-fluorodeoxyglucose positron emission tomography scanning.Positron emission tomography images were processed with SPM2.The brain areas activated after needling at ST36 included the left hippocampus,the left orbital cortex,the left infralimbic cortex,the left olfactory cortex,the left cerebellum and the left pons.In the sham-point group,the activated regions were similar to those in the ST36 group.However,the ST36 group showed greater activation in the cerebellum and pons than the sham-point group.These findings suggest that long-term acupuncture treatment has targeted regulatory effects on multiple brain regions in rats with Alzheimer＇s disease.展开更多
为更准确地预测远程会诊需求量,提高远程会诊资源配置效率,文中引入多元回归分析(Multiple Linear Regression)和注意力机制来优化长短期记忆网络(LSTM)。首先,根据远程会诊需求中存在的假期效应生成假期指标,通过多元回归分析选取显著...为更准确地预测远程会诊需求量,提高远程会诊资源配置效率,文中引入多元回归分析(Multiple Linear Regression)和注意力机制来优化长短期记忆网络(LSTM)。首先,根据远程会诊需求中存在的假期效应生成假期指标,通过多元回归分析选取显著性高的指标作为模型输入,然后根据长短期记忆网络学习输入指标的内部复杂映射关系,利用注意力机制对指标分配不同权重,最后根据权重和LSTM隐藏层输入预测结果。基于国家远程医疗中心(NTCC)的实际历史会诊数据,研究MLR-Attention-LSTM的预测性能,并比较其与整合移动平均自回归模型、支持向量机、K近邻、BP神经网络和LSTM神经网络5种模型的预测效果。结果表明,优化后的LSTM模型预测精度最高。进一步地,探究假期指标对模型性能的影响,结果表明假期指标的输入可以进一步提高模型的预测精度,验证了MLR-Attention-LSTM和假期相关变量输入在远程会诊需求预测领域的可行性与适用性,为远程医学中心实际应用提供了理论支撑和实践指导。展开更多
基金supported by the National Basic Research Program of China(973 Program),No.2006CB504505,2012CB518504the National Natural Science Foundation of China,No.90709027+1 种基金the Student's Platform for Innovation and Entrepreneurship Training Program of Southern Medical University of China,No.201512121165the Doctoral Foundation of Guangdong Medical University of China,No.2XB13058
文摘The acute effect of acupuncture on Alzheimer＇s disease,i.e.,on brain activation during treatment,has been reported.However,the effect of long-term acupuncture on brain activation in Alzheimer＇s disease is unclear.Therefore,in this study,we performed long-term needling at Zusanli（ST36）or a sham point（1.5 mm lateral to ST36）in a rat Alzheimer＇s disease model,for 30 minutes,once per day,for 30 days.The rats underwent 18F-fluorodeoxyglucose positron emission tomography scanning.Positron emission tomography images were processed with SPM2.The brain areas activated after needling at ST36 included the left hippocampus,the left orbital cortex,the left infralimbic cortex,the left olfactory cortex,the left cerebellum and the left pons.In the sham-point group,the activated regions were similar to those in the ST36 group.However,the ST36 group showed greater activation in the cerebellum and pons than the sham-point group.These findings suggest that long-term acupuncture treatment has targeted regulatory effects on multiple brain regions in rats with Alzheimer＇s disease.
文摘为更准确地预测远程会诊需求量,提高远程会诊资源配置效率,文中引入多元回归分析(Multiple Linear Regression)和注意力机制来优化长短期记忆网络(LSTM)。首先,根据远程会诊需求中存在的假期效应生成假期指标,通过多元回归分析选取显著性高的指标作为模型输入,然后根据长短期记忆网络学习输入指标的内部复杂映射关系,利用注意力机制对指标分配不同权重,最后根据权重和LSTM隐藏层输入预测结果。基于国家远程医疗中心(NTCC)的实际历史会诊数据,研究MLR-Attention-LSTM的预测性能,并比较其与整合移动平均自回归模型、支持向量机、K近邻、BP神经网络和LSTM神经网络5种模型的预测效果。结果表明,优化后的LSTM模型预测精度最高。进一步地,探究假期指标对模型性能的影响,结果表明假期指标的输入可以进一步提高模型的预测精度,验证了MLR-Attention-LSTM和假期相关变量输入在远程会诊需求预测领域的可行性与适用性,为远程医学中心实际应用提供了理论支撑和实践指导。
