Objective To investigate the application of pediatric early warning score (PEWS) in neurology department.Methods The clinical data and PEWS of 1,158 children with neurologic diseases (764 febrile seizures,259 viral encephalitis and 135 bacterial meningitis) admitted into neurology department from August 2013 to November 2016 were analyzed retrospectively.PEWS was compared among the three different diseases.According to the disease severity,cases were categorized into no-monitor-required group (n =996,697 febrile seizures,203 viral encephalitis and 96 bacterial meningitis),monitor-required group (n =138,67 febrile seizures,41 viral encephalitis and 30 bacterial meningitis) and PICU group (n =24,0 febrile seizures,15 viral encephalitis and 9 bacterial meningitis).PEWS was assessed at admission in both no-monitor-required group and monitorrequired group.PEWS was recorded 12 hours before PICU transfer in PICU group.Then PEWS was compared among all groups in different diseases respectively.Results The median (minimum ～ maximum) of PEWS in febrile seizures,viral encephalitis and bacterial meningitis was 0 (0 ～ 3),2(0 ～ 8) and 2(0 ～ 8),respectively and there was significant difference among three diseases (P ＜ 0.01).No patient of febrile seizures was transferred to PICU.PEWS of febrile seizures in monitor-required group was higher than that in no-monitor-required group (P ＜ 0.01).For both viral encephalitis and bacterial meningitis,PEWS in monitor-required group was higher than that in no-monitorrequired group,and the score in PICU group was higher than that in monitor-required group.The difference was significant (P ＜ 0.01).Conclusion PEWS is effective for the assessment of illness severity of hospitalized children in neurology department,and can be used for the prediction PICU transfer as an excellent screening tool.

Objective To study the effects of diterpene ginkgolides meglumine injection ( DGMI) on memory impairment, activation of microglia and astrocytes and inflammatory cytokines in aged mice. Methods Twenty aged mice (22 months old) were randomly divided into two groups:aged mouse group(n=10) and DGMI group(n=10). Another 10 mice (2 months old) were selected as young mouse control group. The mice in DGMI group were received 5 mg/kg DGMI per day by tail vain injection for 4 weeks. The mice in the other two groups were received the same amount normal saline for 4 weeks. The Morris water maze was used to evaluate the function of spacial learning and memory after administration of drugs. The ex-pression of CD11b,GFAP,IL-1β,IL-6,TNF-α and NFκB in mice brain hippocampus were detected by West-ern blot. Results (1) The escape latency time of aged mouse group was significantly longer than that of young mouse control group from the 2nd day to the 7th day(P<0. 01). The times of platform crossing,time and distance in target quadrant of aged mouse group were significantly shorter than those of young mouse group (all P<0. 01). Compared with aged mouse group,DGMI significantly reduced the escape latency time of DGMI group (P<0. 01). DGMI increased the times of platform crossing,time and distance in target quad-rant of aged mouse group (P<0. 01). (2) The expressions of CD11b,GFAP in young mouse control group, aged mouse group and DGMI group were as follows respectively:CD11b:(1. 036±0. 023),(1. 757±0. 046), (1. 214±0. 024);GFAP:(1. 022±0. 071),(1. 344±0. 021),(1. 086±0. 073). DGMI reduced the expres-sion of CD11b and GFAP in hippocampus compared with aged mouse group ( t=5. 556,P<0. 01;t=5. 484, P<0. 01). (3) The expressions of IL-1β,IL-6,TNF-α and NFκB in young mouse control group,aged mouse group and DGMI group were as follows respectively:IL-1β:( 1. 003 ± 0. 057),( 2. 062± 0. 105),( 1. 182± 0. 084);IL-6:(1. 018±0. 024),(1. 583± 0. 052),( 1. 152± 0. 031); TNF-α:( 1. 021± 0. 054),(1. 449± 0. 053),(1. 211±0. 036);p-NFκB:(1. 052±0. 034),(1. 782± 0. 113),( 1. 158± 0. 066). DGMI reduced the expression of p-NFκB(t=6. 547,P<0. 01) and pro-inflammatory cytokines including IL-1β(t=8. 513,P<0. 01),IL-6(t=3. 421,P<0. 01) and TNF-α( t=5. 562,P<0. 01) in hippocampus compared with aged mouse group. Conclusion DGMI can improve the ability of learning and memory in aged mice. The mecha-nism may be related with inhibiting activity of microgliosis,astrocytosis,NFκB and neuroinflammaton.

Background: This study aimed to develop a diabetic kidney disease (DKD) prediction model using long short term memory (LSTM) neural network and evaluate its performance using accuracy, precision, recall, and area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Methods: The study identified DKD risk factors through literature review and physician focus group, and collected 7 years of data from 6,040 type 2 diabetes mellitus patients based on the risk factors. Pytorch was used to build the LSTM neural network, with 70% of the data used for training and the other 30% for testing. Three models were established to examine the impact of glycosylated hemoglobin (HbA1c), systolic blood pressure (SBP), and pulse pressure (PP) variabilities on the model’s performance. Results: The developed model achieved an accuracy of 83% and an AUC of 0.83. When the risk factor of HbA1c variability, SBP variability, or PP variability was removed one by one, the accuracy of each model was significantly lower than that of the optimal model, with an accuracy of 78% (P<0.001), 79% (P<0.001), and 81% (P<0.001), respectively. The AUC of ROC was also significantly lower for each model, with values of 0.72 (P<0.001), 0.75 (P<0.001), and 0.77 (P<0.05). Conclusion The developed DKD risk predictive model using LSTM neural networks demonstrated high accuracy and AUC value. When HbA1c, SBP, and PP variabilities were added to the model as featured characteristics, the model’s performance was greatly improved.

Background: This study aimed to develop a diabetic kidney disease (DKD) prediction model using long short term memory (LSTM) neural network and evaluate its performance using accuracy, precision, recall, and area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Methods: The study identified DKD risk factors through literature review and physician focus group, and collected 7 years of data from 6,040 type 2 diabetes mellitus patients based on the risk factors. Pytorch was used to build the LSTM neural network, with 70% of the data used for training and the other 30% for testing. Three models were established to examine the impact of glycosylated hemoglobin (HbA1c), systolic blood pressure (SBP), and pulse pressure (PP) variabilities on the model’s performance. Results: The developed model achieved an accuracy of 83% and an AUC of 0.83. When the risk factor of HbA1c variability, SBP variability, or PP variability was removed one by one, the accuracy of each model was significantly lower than that of the optimal model, with an accuracy of 78% (P<0.001), 79% (P<0.001), and 81% (P<0.001), respectively. The AUC of ROC was also significantly lower for each model, with values of 0.72 (P<0.001), 0.75 (P<0.001), and 0.77 (P<0.05). Conclusion The developed DKD risk predictive model using LSTM neural networks demonstrated high accuracy and AUC value. When HbA1c, SBP, and PP variabilities were added to the model as featured characteristics, the model’s performance was greatly improved.

Background: This study aimed to develop a diabetic kidney disease (DKD) prediction model using long short term memory (LSTM) neural network and evaluate its performance using accuracy, precision, recall, and area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Methods: The study identified DKD risk factors through literature review and physician focus group, and collected 7 years of data from 6,040 type 2 diabetes mellitus patients based on the risk factors. Pytorch was used to build the LSTM neural network, with 70% of the data used for training and the other 30% for testing. Three models were established to examine the impact of glycosylated hemoglobin (HbA1c), systolic blood pressure (SBP), and pulse pressure (PP) variabilities on the model’s performance. Results: The developed model achieved an accuracy of 83% and an AUC of 0.83. When the risk factor of HbA1c variability, SBP variability, or PP variability was removed one by one, the accuracy of each model was significantly lower than that of the optimal model, with an accuracy of 78% (P<0.001), 79% (P<0.001), and 81% (P<0.001), respectively. The AUC of ROC was also significantly lower for each model, with values of 0.72 (P<0.001), 0.75 (P<0.001), and 0.77 (P<0.05). Conclusion The developed DKD risk predictive model using LSTM neural networks demonstrated high accuracy and AUC value. When HbA1c, SBP, and PP variabilities were added to the model as featured characteristics, the model’s performance was greatly improved.

Objective    To systematically review the efficacy and safety of transfemoral transcatheter aortic valve replacement (TFTAVR) under local anesthesia (LA) and general anesthesia (GA). Methods    Electronic databases including PubMed, EMbase, The Cochrane Library, Web of Science, CNKI, WanFang and CBM were searched to collect randomized controlled trial and cohort studies on clinical outcomes of TFTAVR under LA and GA from inception to September 2020. Two authors independently screened literature, extracted data and assessed the quality of studies, and a meta-analysis was performed by using Stata 16.0 software. Results    A total of 30 studies involving 52 087 patients were included in this study. There were 18 719 patients in the LA group and 33 368 patients in the GA group. The results of meta-analysis showed that the in-hospital all-cause mortality rate [RR=0.65, 95%CI (0.45, 0.94), P=0.021], 30-day all-cause mortality rate [RR=0.73, 95%CI (0.62, 0.86), P<0.001], 30-day stroke [RR=0.82, 95%CI (0.68, 0.98), P=0.025], cardiac arrest [RR=0.50, 95%CI (0.34, 0.73), P<0.001], ICU stay time [RR=−6.86, 95%CI (−12.31, −1.42), P=0.013], and total hospital stay  time [RR=−2.02, 95%CI (−2.59, −1.45), P<0.001] in the LA group were all better than those in the GA group. There was no significant difference in the in-hospital stroke [RR=0.83, 95%CI (0.69, 1.00), P=0.053], in-hospital myocardial infarction (MI) [RR=1.74, 95%CI (0.43, 7.00), P=0.434], or 30-day MI [RR=0.77, 95%CI (0.42, 1.42), P=0.404] between the two groups. Conclusion    LA provides a safe and effective way to induce sedation without intubation, and may be a good alternative to GA for TFTAVR.

Objective    To investigate effectiveness and safety of transcatheter aortic valve replacement in the treatment of aortic regurgitation. Methods     PubMed, EMbase, The Cochrane Library, Web of Science, CNKI, Wanfang Data and VIP were searched from inception to August 2021. According to the criteria of inclusion and exclusion, two reviewers independently screened the literature, extracted the data and evaluated the quality of the included studies. Then, Stata 16.0 software was used for meta-analysis. Subgroup meta-analysis of valve type used and study type was performed. Results    Twenty-five studies (12 cohort studies and 13 single-arm studies) were included with 4 370 patients. Meta-analysis results showed that an incidence of device success was 87% (95%CI 0.81-0.92). The success rate of the new generation valve subgroup was 93% (95%CI 0.89-0.96), and the early generation valve subgroup was 66% (95%CI 0.56-0.75). In addition, the 30-day all-cause mortality was 7% (95%CI 0.05-0.10), the 30-day cardiac mortality was 4% (95%CI 0.01-0.07), the incidence of pacemaker implantation was 10% (95%CI 0.08-0.13), and the incidence of conversion to thoraco-tomy was 2% (95%CI 0.01-0.04). The incidence of moderate or higher paravalvular aortic regurgitation was 6% (95%CI  0.03-0.09). Conclusion     Transcatheter aortic valve replacement for aortic regurgitation is safe and yields good results, but some limitations can not be overcome. Therefore, multicenter randomized controlled trials are needed to confirm our results.