Objective To explore an artificial intelligence (AI)-based method for automated hand hygiene monitoring and to compare the effectiveness of three algorithms (UniFormerV2, TDN, C3D) in recognizing hand hygiene steps in surgical settings, thereby aiding hospital infection control. Methods From April to October 2024, we non-invasively collected 641 video recordings of healthcare staff performing hand hygiene at four-bay scrub sinks in two tertiary hospitals using overhead HD cameras. The dataset was annotated by five trained experts for model training and validation. Results Following training on 385 samples, internal validation (n=119) showed the C3D model achieved 81% accuracy, 87% recall, and an 83% F1-score. The TDN model achieved 93%, 91%, and 92% for the same metrics. The UniFormerV2 model outperformed both, with an accuracy, recall, and F1-score of 93%—an improvement of over 10 percentage points compared to traditional CNNs (TDN, C3D). It also achieved an 84% accuracy in external validation, demonstrating strong generalization. Conclusion The UniFormerV2 model is more accurate than CNN-based models for hand hygiene step recognition and shows robust performance in external validation. It presents a viable tool for healthcare facilities to enhance hand hygiene management, ultimately improving medical quality and patient safety.

Objective To construct large medical model named by "Huaxi HongYi"and explore its application effectiveness in assisting medical record generation. Methods By the way of a full-chain medical large model construction paradigm of "data annotation - model training - scenario incubation", through strategies such as multimodal data fusion, domain adaptation training, and localization of hardware adaptation, "Huaxi HongYi" with 72 billion parameters was constructed. Combined with technologies such as speech recognition, knowledge graphs, and reinforcement learning, an application system for assisting in the generation of medical records was developed. Results Taking the assisted generation of discharge records as an example, in the pilot department, after using the application system, the average completion times of writing a medical records shortened (21 min vs. 5 min) with efficiency increased by 3.2 time, the accuracy rate of the model output reached 92.4%. Conclusion It is feasible for medical institutions to build independently controllable medical large models and incubate various applications based on these models, providing a reference pathway for artificial intelligence development in similar institutions.

At present, robotic surgery system, as a cutting-edge technique in minimally invasive surgery, has become a crucial component in the training programs for outstanding surgeons. Given the relatively late start of this technique and immature faculty resources, there is an urgent need to optimize and improve the teaching philosophies and methodologies for robotic skill training in China, and Sino-foreign cooperation has emerged as a significant pathway to promote this reform. Taking the "International Robotic Surgery Training Center (IRSTC)" jointly established by The First Affiliated Hospital of Sun Yat-sen University and University of California as an example, this article provides a comprehensive analysis of collaboration background, construction of the training system, curriculum design, cooperation channels, and localization practices, in order to provide a reference and valuable experiences for Sino-foreign cooperation in clinical skill training.

Neovascular fundus diseases mainly include neovascular age-related macular degeneration(nAMD)and dia-betic retinopathy(DR).Pathological neovascular leakage and the subsequent retinal detachment are the main causes of se-vere visual impairment.Anti-vascular endothelial growth factor(VEGF)is the first-line treatment for neovascular fundus diseases,but it has shortcomings,such as the need for frequent intravitreal injections and poor patient compliance.With the annually increasing incidence of acquired neovascular fundus diseases like nAMD and DR,there is an urgent need for safer and more long-lasting treatment options.In recent years,the field of gene therapy has advanced rapidly,with thera-peutic strategies mainly involving gene supplementation and editing.The mechanism underlying gene therapy can be suc-cinctly described as the correction of pathological alterations induced by defective genes.This is achieved either by the in-troduction of exogenous functional genes to restore normal cellular processes or by directly editing aberrant genes at the ge-nomic level.Extensive basic and clinical research has demonstrated that gene therapy is both safe and effective.There are dozens of clinical trials on retinal gene therapy being carried out currently,focusing not only on inherited retinal diseases but also on neovascular fundus diseases.In this article,the application of the gene supplementation,clustered regularly in-terspaced short palindromic repeats(CRISPR)and CRISPR-associated protein 9(Cas9)system in the treatment of nAMD and DR is summarized.

Objective This study aimed to investigate the regulatory effect of electroacupuncture(EA)intervention on the protein kinase B(Akt)/mammalian target of rapamycin(mTOR)/hypoxia-inducible factor 1α(HIF-1α)signaling pathway in the hippocampal tissue of Alzheimer's disease(AD)model mice and its effect on astrocytic glycolytic function,further exploring how EA ameliorates AD-related cognitive impairment.Methods Eighteen APP/PS1 mice were randomly divided into model,EA,and sham EA groups(n=6)using the random number table method.Six wild-type C57BL/6J mice served as the control group.The EA group received EA stimulation at acupoints"Shenshu"(BL23),"Baihui"(GV20),and"Zusanli"(ST36)(administered every other day,20 min per session,for 4 weeks).The sham EA group received identical needle insertions at the same acupoints without electrical stimulation.The control and model groups were only restrained.Cognitive function was assessed using the Morris water maze and Y-maze spontaneous alternation tests.Hippocampal morphology was observed via hematoxylin and eosin staining.Hippocampal β-amyloid peptide 1-42(Aβ1-42)deposition was detected using immunohistochemistry.HIF-1α protein expression,the p-Akt/Akt,and p-mTOR/mTOR ratios were measured using Western blotting.Pyruvate kinase M2(PKM2)and lactate dehydrogenase A(LDHA)activities were quantified using enzyme-linked immunosorbent assay.Hexokinase(HK)activity and L-lactate content were determined using a colorimetric assay.Co-localization of LDHA with the astrocyte marker glial fibrillary acidic protein was quantitatively analyzed using immunofluorescence double-labeling combined with Pearson's correlation coefficient.Results Compared with the control group,the model group mice exhibited cognitive decline,as shown by prolonged escape latency(P<0.01),reduced number of platform crossings,lower time spent in the target quadrant,and decreased spontaneous alternation accuracy(P<0.01).The hippocampal neurons showed cell body swelling,deeper nuclear staining,enlarged intercellular spaces,and increased average optical density of Aβ1-42(P<0.01).The p-Akt/Akt and p-mTOR/mTOR ratios,as well as HIF-1α protein expression,were elevated(P<0.01).PKM2,LDHA,HK,and L-lactic acid levels were significantly increased(P<0.01),and the co-localization coefficient of LDHA with astrocytes was enhanced.Compared to the model group,the EA group of mice showed improved cognitive function.The hippocampal neurons had more intact structures,with a more uniform cell distribution.The average optical density of Aβ1-42 decreased(P<0.01),and the p-Akt/Akt and p-mTOR/mTOR ratios,as well as HIF-1α protein expression,decreased(P<0.01).PKM2,LDHA,HK,and L-lactic acid levels decreased(P<0.05),and the co-localization coefficient of LDHA with astrocytes significantly decreased(P<0.01).No significant improvement was observed in any of the indicators in the sham EA group compared with the EA group.Conclusion EA at"Shenshu"(BL23),"Baihui"(GV20),and"Zusanli"(ST36)ameliorates cognitive dysfunction in AD model mice.The underlying mechanism may involve suppressing the overactivation of the hippocampal Akt/mTOR/HIF-1α signaling pathway,thereby downregulating key glycolytic enzyme activities and reducing abnormal lactate accumulation.Furthermore,the astrocytic glycolytic metabolic pathway may constitute a key therapeutic target for this intervention.

Objective To carry out a dynamic simulation analysis on the surgical process of the surgical robot for pedicle screw placement so as to enhance the safety of the procedure.Methods Firstly,the process of pedicle screw placement were analyzed to determine the three typical force conditions during pedicle screw track drilling including no-load condition,bone layer switching condition and spine dynamic displacement condition.Secondly,a virtual protype model of the surgical robot for pedicle screw placement was constructed with the automated dynamic analysis of mechanical systems(ADAMS).Finally,the dynamic characteristics of the surgical robot were simulated and analyzed with considerations on the three typical force conditions.Results The driving torque of the robot joints was sensitive to the load applied to the end of the opener mechanism under a wide range of operating conditions.Conclusion The surgical robot meets the requirements for pedicle screw placment,and a new idea is provided for enhancing the accuracy of pedicle screw placement.[Chinese Medical Equipment Journal,2025,46(8):32-37]

AIM To establish the near-infrared spectroscopy quantitative models for moisture,23-acetylalismol B and 23-acetylalismol C in Alismatis Rhizoma decoction pieces.METHODS The near-infrared spectroscopy(NIRS)data were collected in 95 batches of decoction pieces,after which drying method was adopted in the content determination of moisture,HPLC was applied to determining the contents of 23-acetylalismol B and 23-acetylalismol C,the quantitative models were established by partial least squares method combined with feature extraction algorithms.RESULTS The model training determination coefficients were 0.952 6,0.958 1 and 0.920 8,along with the prediction determination coefficients of 0.930 0,0.905 2 and 0.906 4,the residual prediction deviations(PRD)of 4.00,3.58 and 3.46,and the root mean square error ratios of prediction values to calibration values(RMSEP/RMSEC)of 1.15,1.11 and 1.06,respectively.CONCLUSION The quantitative models based on NIRS exhibit good prediction effects,which can be used for the rapid quality detection of Alismatis Rhizoma decoction pieces.

Objective This study aimed to investigate the regulatory effect of electroacupuncture(EA)intervention on the protein kinase B(Akt)/mammalian target of rapamycin(mTOR)/hypoxia-inducible factor 1α(HIF-1α)signaling pathway in the hippocampal tissue of Alzheimer's disease(AD)model mice and its effect on astrocytic glycolytic function,further exploring how EA ameliorates AD-related cognitive impairment.Methods Eighteen APP/PS1 mice were randomly divided into model,EA,and sham EA groups(n=6)using the random number table method.Six wild-type C57BL/6J mice served as the control group.The EA group received EA stimulation at acupoints"Shenshu"(BL23),"Baihui"(GV20),and"Zusanli"(ST36)(administered every other day,20 min per session,for 4 weeks).The sham EA group received identical needle insertions at the same acupoints without electrical stimulation.The control and model groups were only restrained.Cognitive function was assessed using the Morris water maze and Y-maze spontaneous alternation tests.Hippocampal morphology was observed via hematoxylin and eosin staining.Hippocampal β-amyloid peptide 1-42(Aβ1-42)deposition was detected using immunohistochemistry.HIF-1α protein expression,the p-Akt/Akt,and p-mTOR/mTOR ratios were measured using Western blotting.Pyruvate kinase M2(PKM2)and lactate dehydrogenase A(LDHA)activities were quantified using enzyme-linked immunosorbent assay.Hexokinase(HK)activity and L-lactate content were determined using a colorimetric assay.Co-localization of LDHA with the astrocyte marker glial fibrillary acidic protein was quantitatively analyzed using immunofluorescence double-labeling combined with Pearson's correlation coefficient.Results Compared with the control group,the model group mice exhibited cognitive decline,as shown by prolonged escape latency(P<0.01),reduced number of platform crossings,lower time spent in the target quadrant,and decreased spontaneous alternation accuracy(P<0.01).The hippocampal neurons showed cell body swelling,deeper nuclear staining,enlarged intercellular spaces,and increased average optical density of Aβ1-42(P<0.01).The p-Akt/Akt and p-mTOR/mTOR ratios,as well as HIF-1α protein expression,were elevated(P<0.01).PKM2,LDHA,HK,and L-lactic acid levels were significantly increased(P<0.01),and the co-localization coefficient of LDHA with astrocytes was enhanced.Compared to the model group,the EA group of mice showed improved cognitive function.The hippocampal neurons had more intact structures,with a more uniform cell distribution.The average optical density of Aβ1-42 decreased(P<0.01),and the p-Akt/Akt and p-mTOR/mTOR ratios,as well as HIF-1α protein expression,decreased(P<0.01).PKM2,LDHA,HK,and L-lactic acid levels decreased(P<0.05),and the co-localization coefficient of LDHA with astrocytes significantly decreased(P<0.01).No significant improvement was observed in any of the indicators in the sham EA group compared with the EA group.Conclusion EA at"Shenshu"(BL23),"Baihui"(GV20),and"Zusanli"(ST36)ameliorates cognitive dysfunction in AD model mice.The underlying mechanism may involve suppressing the overactivation of the hippocampal Akt/mTOR/HIF-1α signaling pathway,thereby downregulating key glycolytic enzyme activities and reducing abnormal lactate accumulation.Furthermore,the astrocytic glycolytic metabolic pathway may constitute a key therapeutic target for this intervention.

Background: Gestational diabetes mellitus (GDM) is a metabolic disorder posing significant risks to maternal and infant health, with a lack of effective early screening markers. Therefore, identifying early screening biomarkers for GDM with higher sensitivity and specificity is urgently needed. Methods: High-throughput sequencing was employed to screen for key circular RNAs (circRNAs), which were then evaluated using reverse transcription quantitative polymerase chain reaction. Logistic regression analysis was conducted to examine the relationship between clinical characteristics, circRNA expression, and adverse pregnancy outcomes. The diagnostic accuracy of circRNAs for early and mid-pregnancy GDM was assessed using receiver operating characteristic curves. Pearson correlation analysis was utilized to explore the relationship between circRNA levels and oral glucose tolerance test results. A predictive model for early GDM was established using logistic regression. Results: Significant alterations in circRNA expression profiles were detected in GDM patients, with hsa_circ_0031560 and hsa_ circ_0000793 notably upregulated during the first and second trimesters. These circRNAs were associated with adverse pregnancy outcomes and effectively differentiated GDM patients, with second trimester cohorts achieving an area under the curve (AUC) of 0.836. In first trimester cohorts, these circRNAs identified potential GDM patients with AUCs of 0.832 and 0.765, respectively. The early GDM prediction model achieved an AUC of 0.904, validated in two independent cohorts. Conclusion Hsa_circ_0031560, hsa_circ_0000793, and the developed model serve as biomarkers for early prediction or midterm diagnosis of GDM, offering clinical tools for early GDM screening.

Background: Gestational diabetes mellitus (GDM) is a metabolic disorder posing significant risks to maternal and infant health, with a lack of effective early screening markers. Therefore, identifying early screening biomarkers for GDM with higher sensitivity and specificity is urgently needed. Methods: High-throughput sequencing was employed to screen for key circular RNAs (circRNAs), which were then evaluated using reverse transcription quantitative polymerase chain reaction. Logistic regression analysis was conducted to examine the relationship between clinical characteristics, circRNA expression, and adverse pregnancy outcomes. The diagnostic accuracy of circRNAs for early and mid-pregnancy GDM was assessed using receiver operating characteristic curves. Pearson correlation analysis was utilized to explore the relationship between circRNA levels and oral glucose tolerance test results. A predictive model for early GDM was established using logistic regression. Results: Significant alterations in circRNA expression profiles were detected in GDM patients, with hsa_circ_0031560 and hsa_ circ_0000793 notably upregulated during the first and second trimesters. These circRNAs were associated with adverse pregnancy outcomes and effectively differentiated GDM patients, with second trimester cohorts achieving an area under the curve (AUC) of 0.836. In first trimester cohorts, these circRNAs identified potential GDM patients with AUCs of 0.832 and 0.765, respectively. The early GDM prediction model achieved an AUC of 0.904, validated in two independent cohorts. Conclusion Hsa_circ_0031560, hsa_circ_0000793, and the developed model serve as biomarkers for early prediction or midterm diagnosis of GDM, offering clinical tools for early GDM screening.