Objective This study aims to analyze the application effectiveness of a life support medical equipment sharing and borrowing system based on the Internet of Things. Methods Implement an innovative equipment management plan in the equipment department of a hospital in Shenzhen, and build a shared borrowing system for life support medical equipment. During the research process, 106 devices were selected as the research subjects, including 55 monitors, 15 invasive ventilators, 5 non-invasive ventilators, 20 infusion pumps, 3 defibrillators, 6 nutrition pumps, 1 ECMO (extracorporeal membrane oxygenation), and 1 electrocardiograph. The equipment and personnel remained unchanged before and after application, and were stored in a centralized location. Before the system application, it was in a purely manual state from June 2022 to August 2023, serving as the control group; The time after application is from September 2023 to November 2024. For the observation group, data analysis was conducted to compare the equipment borrowing time, equipment borrowing error rate, calculation and audit personnel work error rate, shift handover time, and equipment final maintenance qualified units before and after the platform application. Results The construction of equipment sharing platform and centralized management platform within the hospital has significantly improved the management level of equipment. After the application of the system, the equipment borrowing time and shift handover time were significantly reduced, the borrowing error rate, calculation and audit personnel error rate were significantly reduced, and the number of qualified equipment terminal maintenance was significantly increased (P<0.05). Conclusion Building a life support medical equipment sharing and borrowing system based on the Internet of Things is an effective way to improve equipment management level and operational efficiency. The borrowing time of devices after applying the system is significantly shortened, and the deployment error rate is significantly reduced; The automation level of equipment borrowing work has been improved, the error rate has been reduced, and work efficiency has been enhanced; The handover process for equipment borrowing work is smoother, reducing unnecessary time consumption; The standardization level of equipment borrowing work has been improved, and the final maintenance work of equipment has been better guaranteed, improving the reliability and service life of equipment. Worth promoting and using.

Objective This study aims to explore the application pathways and practical effectiveness of data intelligence technologies (such as big data analytics, machine learning, and automated identification) in the full-process hospital consumables management—including procurement, inventory, usage, and settlement—under the implementation of the volume-based procurement (VBP) policy. The goal is to provide new insights for achieving refined management in hospitals. Methods In the procurement prediction module, big data analytics were employed to deeply mine historical procurement records and clinical usage trends. A predictive model was constructed using machine learning algorithms to achieve accurate forecasting for procurement needs. Results After system implementation, the allocation time for routine medical consumables was significantly reduced from the original 5–10 minutes to under 1 minute, resulting in an efficiency improvement of 80–90%. The overall efficiency of the consumables management process increased by approximately 82%, with inventory turnover days reduced by 25% and the rate of expired and overstocked items decreased by 40%. Timeliness in consumables acquisition on the clinical side improved markedly, and departmental satisfaction rose from 78% to 93%. Conclusion The data-intelligence-based consumables VBP management system significantly enhances hospitals' operational responsiveness and resource allocation efficiency under the VBP policy. With features such as traceability, high efficiency, and low risk, it serves as a key enabler for the transformation of modern hospital consumables management and holds strong potential for wider application.

Objective To design and develop an online medical image annotation training platform based on the Django framework, addressing the challenges of annotator shortages, low efficiency, and high costs in AI-driven healthcare. Methods The platform backend was built using the Django framework, integrated with HTML, CSS, and JavaScript for frontend interaction. RESTful APIs ensured data communication, while Token-based authentication enhanced security. The assessment module utilized Python's shapely library to evaluate annotation accuracy via the Intersection over Union (IoU) algorithm. A multimodal annotation tool, developed with JavaScript Canvas API, supported 2D/3D image labeling (e.g., bounding boxes and polygons). A dynamic task allocation algorithm optimized annotation workflows. Clinical and imaging students underwent online training and assessments, after which the platform automatically assigned annotation tasks and provided feedback to task publishers and affiliated institutions.Results During the trial operation period of the platform (June 2024 May 2025), the annotation efficiency of the trainees who received training increased by 23.63% compared to the traditional training mode, and the average annotation accuracy reached 85.5%, which was 73.5% higher than the untrained group. The standardized curriculum of radiology experts has increased the clinical rationality rate by 54.16%. The annotation consistency (IoU ≥ 0.85) of the platform on different modal data reached 88.89%. The task allocation algorithm has increased the completion rate of high difficulty tasks from 68% to 89%. The standardized assessment pass rate for platform students is 90%, providing effective personnel reserves for medical AI enterprises.Conclusion The Django-based platform enhances annotator expertise, addresses the demand for large-scale medical image annotation, and delivers high-quality data support for AI-driven healthcare applications.

Objective Aiming at the problems of slow response speed, low temperature control accuracy, and inconvenient operation of traditional medical contrast agent temperature control devices, a medical contrast agent temperature control system that can be controlled through an APP is developed to achieve portable control of contrast agent temperature.Methods The system is designed based on the STM32F103C8T6 microcontroller, and uses multi-source sensor fusion technology to monitor the temperature of the contrast agent in real time. Combined with fuzzy PID control algorithm, it achieves high-precision constant temperature control of ± 0.2℃. At the software level, a Wi-Fi communication mechanism has been constructed, and an APP has been developed to achieve temperature setting, real-time monitoring, and abnormal alarm functions. Results The experimental results demonstrate that Under an ambient temperature of 25℃, the system heats 1000mL of contrast agent to the target temperature of 37℃ with an average regulation time within 90 seconds. During the steady-state phase, it demonstrates a maximum temperature fluctuation of 0.139℃ and a standard deviation of 0.014℃, meeting medical-grade temperature control standards. Conclusion The research achievement provides reliable temperature protection for medical examinations such as CT enhanced scanning and angiography, which helps to improve image contrast and reduce the risk of vascular spasm caused by cold stimulation. It has significant clinical promotion value.

Objective To study the construction and application of evaluation scale for clinical use of high-value medical consumables in order to provide a scientific method for clinical use evaluation of high-value medical consumables that have been admitted to the medical institutions. Methods Based on literature analysis and Delphi method, a three-level index pool was constructed to evaluate the clinical use of high-value medical consumables. Then, the weights of each level of indicators were determined by analytic hierarchy process, and each index was quantified in turn. Finally, the evaluation scale was constructed. A case study was carried out in a third class hospital. Results The evaluation scale consisted of 4 first-level indicators,9 second-level indicators and 24 third-level indicators. The weights and scores of each indicator were determined. The example proved that the evaluation scale was scientific and practical. Conclusion The evaluation of clinical use of high-value medical consumables based on evaluation scale was scientific and practical. This method provided scientific basis and methodological reference for the managers of medical institutions to improve the management level of high-value medical consumables, improve medical quality and control medical cost.

Objective To evaluate the accuracy of dose delivery in carbon ion therapy for ocular melanoma based on offline positron emission tomography/computed tomography (PET/CT) nuclide imaging, and to validate its reliability in clinical treatment planning. Methods This study included 34 ocular melanoma patients treated with carbon ion therapy. Offline PET imaging was performed immediately after treatment to obtain β⁺ nuclide activity distribution. By comparing PET activity distribution with the dose distribution predicted by the treatment planning system (TPS), the dose conformity index (DCI) and PET conformity index (PCI) were calculated to assess three-dimensional (3D) dose delivery characteristics. The ratio of PCI to DCI (dose delivery index, DDI) was used to quantify the accuracy of dose delivery. Results At the 95% isodose level, DCI₉₅% and PCI₉₅% were 59.6%±14.7% (P<0.01) and 58.1%±14.5% (P<0.01), respectively. At the 90% isodose level, DCI₉₀% and PCI₉₀% were 43.0%±11.2% (P<0.01) and 42.0%±11.0% (P<0.01), respectively. The dose delivery accuracy indices were DDI₉₅%=99.5%±8.6% and DDI₉₀%=97.5%±11.8%, with statistically significant differences (P<0.01). The results demonstrated good consistency between PET-derived nuclide distribution and planned dose distribution within the target volume, indicating that PET activity effectively reflects relative dose distribution. Moreover, results demonstrated that both post-treatment transfer time to the PET suite and target location (left vs. right eye) significantly impacted DDI values. Conclusion Offline PET nuclide imaging can serve as an effective tool for dose verification in carbon ion therapy for ocular melanoma, particularly in assessing target dose deposition with high accuracy. This study provides important insights for online dose monitoring in carbon ion therapy and the correlation between dose and PET activity.

In order to solve the problems of insufficient data security, weak protection of patient privacy and low efficiency of cross-institutional business collaboration in traditional hospital business systems, an optimization mechanism based on hierarchical chain-cloud architecture and smart contracts is studied and proposed to improve hospital data protection and business process optimization. The optimized system integrates the advantages of blockchain and cloud databases, and realizes the automatic execution of key business processes such as medical record authorization and claims review through smart contracts. The experimental results show that the average access time of the optimized system is 0.49±0.07 seconds, and the illegal access detection rate reaches 98.67%. The response times for medical record authorization and claims review were 0.27±0.04s and 0.85±0.06s respectively, and the traceability reached 99.92% (p<0.001). The crash rate of the optimized system in high-concurrency scenarios is controlled at 4.63%, and the recovery time does not exceed 2.28 seconds. The research verified the advantages of the proposed optimization mechanism in terms of efficiency, security and scalability, providing a practical reference for the construction of the new generation of medical information systems.

Objective To propose a multivariate nonlinear prediction model for thyratron lifespan based on accelerator workload, aiming to achieve lifespan prediction of key components through dynamic workload monitoring, thereby providing a reference for preventive maintenance of linear accelerators. Methods By integrating and analyzing 4,485 sets of historical operational data from four Varian linear accelerators in Zhejiang Cancer Hospital between 2022 and 2024, a multivariate nonlinear model was developed with daily energy load and daily beam-on time as independent variables. The model was tested using 484 real-world operational datasets and evaluated using goodness-of-fit (R²) and mean absolute percentage error (MAPE). Results The model demonstrates strong model performance, with a goodness-of-fit of R²=0.913 and residuals conforming to a normal distribution. Predictions for 26.4% of test data showed errors below 10%, and the MAPE was 21.9%. Conclusion This model provides a quantitative tool for assessing thyratron failure risks, which can optimize preventive maintenance strategies for Varian linear accelerators, ensure continuity of radiotherapy, and reduce downtime losses.

Objective Utilizing a depth camera to extract and simulate the real-time movement of the surface markers on the patient's thermoplastic mask in real time, to assist in position monitoring during radiotherapy. Methods The transformation relationships among the iso coordinate system, the depth camera coordinate system, the DICOM coordinate system and the TPS coordinate system are obtained by the calibrated module, thereby achieving the unification of the multi-coordinate system. The pointcloud data collected by the depth camera is fitted in real time to identify the motion trajectory of the markers and convert it to the unified coordinate space. Singular value decomposition (SVD) registration analysis is carried out with the coordinates of the markers in the simulation CT image, and then the six-dimensional pose error is calculated. Finally, the real-time monitoring of body position during radiotherapy is achieved. In this experiment, a preset human error was introduced into the dummy model to verify the measurement accuracy of the system. The calculated error obtained from the experiment was compared with the set error, and the mean and standard deviation were statistically analyzed. The paired t-test was used to evaluate the significance of the difference between the two groups of data. Meanwhile, calculate the Pearson correlation coefficient between the two groups of errors, analyze the degree of correlation, and record the time required for the system operation.Results The errors of the system in the left-right, head-foot, and front-back directions were (0.199±0.07), (0.202±0.058), and (0.166±0.045) cm respectively; the errors in the yaw, pitch, and roll directions were (0.169±0.14), (0.117±0.095), and (0.242±0.139) ° respectively. The six-dimensional errors calculated by the system were highly correlated with the actual movement data, and the time taken for one monitoring was approximately 0.226 s. Conclusion Depth cameras can accurately simulate the markers on the patient's thermoplastic mask and register them with the marks in the simulation CT, thereby achieving accurate and real-time position monitoring during radiotherapy, which has certain clinical feasibility.

Objective The distribution and use difference of DR equipment in the medical imaging department of public hospitals in the province were investigated. Methods Electronic questionnaire investigated the medical imaging department of secondary and above public hospitals (western comprehensive and specialty), including the number of DR equipment, brand, whether it is equipped with double flat panel detectors, use, etc. Using SPSS 27.0 software, the basic situation of the provincial public western comprehensive and specialized hospitals and the details of DR equipment in the medical imaging department were analyzed differently. Results A total of 140 valid questionnaires were collected, That is, 140 public hospitals in the province, More DR equipment, more DR equipment, more complete and more versatile, Statistically significant difference (p<0.05), Not statistically significant in brand (p>0.05), DR equipment in secondary or above public general hospitals and tertiary public specialized hospitals tends to be imported brands, The DR equipment brand owned by the secondary public specialist medical imaging department, Domestic and imported are almost equivalent; Compared with the tertiary public western comprehensive medical imaging department, it prefers the suspension double-plate DR equipment with high examination efficiency and strong ease ofuse, and the difference is significant (p<0.05).Conclusion Analyze the configuration of DR equipment in secondary and above, grasp the status and demand of DR equipment, and comprehensively, clarify the advantages and disadvantages, helps the hospital improve the medical service level and efficiency, and meet theclinical demand and patient expectation; further understand the cost performance, clinical recognition and market share of domestic DR equipment providing a reference for advancing the research and development of domestic medical imaging equipment.

Objective The oropharyngeal cavity is not a regular spatial structure. In the face of difficult airways, how to quickly deliver the guide strip from the incisors to the vicinity of the glottis, accurately perceive its position, and then quickly bypass the epiglottis and pass through the glottis to enter the trachea remains a difficult pain point to solve. Using the oropharyngeal airway guidance designed in this paper and the magnetic positioning navigation technology, the aim is to accurately and quickly complete the tracheal intubation operation while reducing soft tissue injury. Methods This magnetic positioning and navigation tracheal intubation system consists of a flexible oropharyngeal airway, a flexible magnetic positioning device, a flexible magnetic guidance strip and a navigation magnet. First, insert the soft material oropharyngeal airway into the oral cavity, adjust its position, and fully expose the epiglotis and larynx. The flexible magnetic guide strip is inserted along the soft material oropharyngeal airway, so that the guide strip will not bend or turn around from the incisors to the larynx during blind insertion. The magnetic positioning module can track and monitor the tip position of the flexible magnetic guide strip outside the body. When the magnetic guide strip reaches the vocal cavity area, the magnetic force of the external navigation magnet attracts and navigates the magnet at the tip of the magnetic guide strip, causing it to bend in a directional manner and enter the trachea. Then, insert the tracheal intubation along the magnetic guide strip and remove the magnetic guide strip. The soft oropharyngeal airway can remain in the oral cavity as a dental pad and a fixed catheter. On the dummy , two students without a medical background conducted blind tracheal intubation experiments 50 times each. The learning time, the success rate of one intubation, and the time spent on intubation were counted to evaluate the advantages of this system during the intubation process. Results With the guidance of the oropharyngeal airway and the assistance of the magnetic positioning navigation system, the average first-attempt success rate of blind tracheal intubation on a manikin reached 94%. Among the participants, 86% achieved a success rate of 90% or higher, and 98% achieved 80% or higher, demonstrating good consistency and reliability. In terms of intubation time, all participants were able to complete the procedure within 37 seconds. Among the total of 470 successful intubations, 89.1% were completed within 30 seconds, and 51.4% were completed within 25 seconds. Conclusion Based on the oropharyngeal airway guidance and magnetic positioning navigation technology showing advantages in tracheal intubation, it is expected to effectively increase the one-time success rate of tracheal intubation in difficult airways, shorten the operation time, and have low dependence on the operator. It has good universality and application prospects, providing useful inspirations for the development of intelligent medical robots for tracheal intubation and is worthy of further research and clinical promotion and application.

Cardiac arrest remains a major global public health challenge. The core mechanisms underlying its high mortality and disability rates involve prolonged cerebral exposure to a "no-flow" state and ischemia-reperfusion injury triggered by hypoperfusion during resuscitation. Real-time acquisition of cerebral perfusion and oxygenation parameters during resuscitation is therefore critical for improving outcomes. Regional cerebral oxygen saturation, a non-invasive and continuous monitoring indicator of cerebral tissue oxygenation, enables bedside dynamic assessment through near-infrared spectroscopy technology. This approach has been widely adopted in critical care, rehabilitation medicine, and surgical settings. In the context of cardiopulmonary resuscitation, rSO₂ monitoring provides real-time visualization of cerebral oxygen supply-demand balance, offering crucial evidence for optimizing chest compression quality, predicting return of spontaneous circulation, evaluating cerebral blood flow metabolism, guiding targeted temperature management, assessing neurological prognosis, and informing extracorporeal cardiopulmonary resuscitation decisions. These applications underscore its unique clinical value in CPR. This review systematically summarizes recent technological advancements and multi-scenario applications of rSO₂ monitoring in the CPR field, aiming to provide evidence-based guidance for optimizing resuscitation strategies and improving patient prognoses.

Objective Analttic Hierarchy Process (AHP) was used to construct an index system based on the performance requirements of ECG monitor accessories in different medical scenarios, so as to guide clinical selection of the most appropriate type of accessories. Methods Six departments of pediatrics, internal medicine, surgery, anesthesia surgery, emergency department and critical care medicine in our hospital were selected as the research objects. The index system was initially established through expert interviews and focus group discussion, and then the weights of each index were determined by AHP and Likert scale. Finally, the evaluation scores of departments before and after the use of attachments were verified through experiments. Results The accuracy and stability of the Mindray original SpO2 probe (A=9.714, A'=5.714) were higher than that of Xiaocai brand (B3=10, B3'=5). The convenience, flexibility and maintainability of the original SpO2 probe of the medical staff were significantly better than that of Xiaocai brand (t=3.458, 4.614, 5.308, P < 0.05), and the comfort feeling of the original SpO2 probe of the patients was also significantly higher than that of Xiaocai brand (t=11.484, P < 0.05). Combined with the index system, it is concluded that the original SpO2 probe (comprehensive score 7.867) is more suitable for internal medicine departments than Xiaocai brand (comprehensive score 6.078). Conclusion The performance requirement index system of monitor accessories based on the characteristics of different medical scenarios was constructed to provide decision-making basis for clinical departments to select the most suitable monitor accessories.

Objective To construct an evaluation index system for centralized procurement of electrophysiological consumables in Diagnosis Related Groups ( DRG ) based on the influence of policy effect on medical treatment, medical insurance and medicine, so as to provide reference for subsequent policy formulation.Methods The initial framework of the index system was developed through literature research and semi-structured interviews. The Delphi method was used to conduct two rounds of consultations with 40 experts, and the direct scoring method of experts was used to determine the weight of indicators at all levels.Results The effective recovery rates of the two rounds of expert consultation questionnaires were 88.89% and 100.00%, respectively. The expert authority coefficient was expressed by familiarity and judgment coefficient. The expert authority coefficient in this survey was 0.85 (Cs = 0.90, Ca = 0.80), indicating that experts had high authority in this field. The Kendall's coordination coefficients (Kendall's W) were 0.2080 and 0.2350, respectively. Finally, the DRG electrophysiological consumables centralized procurement evaluation index system was determined to include 3 first-level indicators, 7 second-level indicators, and 24 third-level indicators. Conclusion The evaluation index system of centralized purchase of DRG electrophysiological consumables constructed in this study is scientific, reliable and operable, which can provide reference for relevant decision-making departments.

Objective  To address inefficiencies in traditional medical consumables management—such as poor information tracking and inadequate supervision—we developed an SPD (Supply-Processing-Distribution) management platform based on the Unique Device Identifier (UDI) system. This aims to enhance management efficiency, healthcare quality, and safety.Methods  Core requirements were identified through user research. The platform employs a UDI-centered architecture with three layers: external data interfacing, internal UDI-based materials management, and internal system integration. A layered technical framework and six functional modules enable closed-loop, full-process management. Performance metrics from 20 clinical departments were compared pre- and post-implementation.Results  After implementation, weekly average time consumption for requisitioning, collection, inventory counting, and billing decreased by 49.6%, 46.2%, 46.3%, and 48.0%, respectively (all P<0.001). The billing error rate for high-value consumables dropped from 3.3% to 1.3% (P<0.001), indicating significant improvements in efficiency and accuracy.Conclusion  By deeply integrating technological innovation with workflow optimization, this platform resolves key challenges in traditional management and provides a replicable solution for standardizing and digitizing medical supplies management.

Objective Taking municipal hospitals in Beijing as an example, an in-depth analysis is conducted on the procurement, operation, maintenance costs, and usage data of both domestic and imported large-scale medical equipment. This analysis aims to provide references and suggestions for equipment evaluation and the healthy development of domestic medical equipment. Methods Data on CT, MRI, and X-ray equipment in use in 22 municipal hospitals in Beijing were collected through research,the chi-square test was used to analyze the cost differences from three aspects: "comparison of procurement prices between domestic and imported products", "comparison of procurement prices between domestic products of domestic and imported brands", and "comparison of equipment maintenance prices", and the actual usage and purchase ratio of imported and domestic equipment products are compared, so as to analyze the difference in equipment availability.. Results The procurement prices of domestic products for "64-slice (inclusive) - 128-slice (exclusive) CT", "1.5T and below MRI", "1.5T and above MRI", "DSA", "DR", and "C-arm" were lower than those of imported products (P < 0.05), with a reduction range of 12.97% - 50.8%. Among "64-slice and below CT", "64-slice (inclusive) - 128-slice (exclusive) CT", "1.5T and below MRI", "DR", and "C-arm", only the procurement prices of domestic brand and imported brand domestic products for C-arm had no significant difference (P = 0.711), while the procurement prices of domestic brand products for the other categories were all lower than those of imported brand domestic products (P < 0.05), with a reduction range of 19.68% - 31.91%. For "64-slice (inclusive) - 128-slice (exclusive) CT", "1.5T and below MRI", "DR", and "C-arm", the full-machine maintenance rates after the warranty period for domestic and imported products were basically the same (P > 0.05). In 2020, the proportion of domestic equipment purchases exceeded that of imported equipment for the first time, and remained stable in the past five years.There was no significant difference in the average usage of domestically produced and imported CT and MRI equipment (P=0.260 and 0.055). Conclusion Overall, the procurement prices and maintenance costs of domestic products are lower than those of imported products, The availability and market acceptance of domestic products are not lower than those of imported products. Medical institutions should strengthen their research and demonstration to reduce equipment procurement and operating costs while meeting clinical needs.

Objective Given the variations in key performance of negative pressure isolation chambers across multiple scenarios, this study presents a research methodology to examine how actual operating conditions affect their critical performance. The proposed approach aims to provide a reference for evaluating the practical operational performance of such equipment. Methods Based on the actual operational scenarios and the technical specifications of YY 1885–2023 Negative-Pressure Isolation Capsules for Transporting Infectious Patients, this study designs test methods for the key performance of isolation capsules under three representative scenarios: Continuous use of portable oxygen concentrators by infectious patients inside the capsule, Temperature fluctuations during multi-scenario transportation, Variations in power supply modes and states. Devices including voltage stabilizers, micro-differential pressure gauges, and anemometers are employed to test and analyze the effects of the above scenarios on the negative pressure, air exchange rate, airflow velocity, and filter integrity of the capsules. Results The key performance of negative-pressure isolation capsules for transporting infectious patients is influenced by the three scenarios: When the oxygen concentrator trigger frequency is low or flow rate is high, the negative pressure fluctuates by −5% to 2.5%, and the airflow velocity changes by −3.8% to 2.3%. Temperature variations affect the negative pressure by −34.4% to 12.5%, airflow velocity by −10.7% to −1.8%, and air exchange rate by −9.5% to 4.8%. Changes in power supply modes/states impact the negative pressure by 2–17 Pa, airflow velocity by 0.03–0.05 m/s, and air exchange rate by 1–5 times/h. Conclusion The proposed research methods are feasible and generate valid test results, aiming to provide references for the revision and upgrading of standards for such equipment.

Objective To address the challenge of intelligent fault diagnosis for the main control board of soft tissue therapy instruments in the absence of circuit diagrams, an efficient and accurate diagnostic method is proposed. Methods Electrical signal data were collected from 45 pins across 7 ports of the main control board, and combined with symptom information including panel displays, button status, and output waveforms to construct a 198-dimensional multimodal feature set. Support Vector Machine Recursive Feature Elimination (SVM-RFE) was applied to select key features, retaining the optimal 60 features to enhance model generalization. An improved Sparrow Search Algorithm (ISSA) incorporating adaptive inertia weight adjustment and hybrid mutation strategy was designed to optimize the SVM kernel parameter γ and penalty factor C, thereby establishing the ISSA-SVM fault diagnosis model. Results Experimental results showed that the ISSA-SVM model outperformed PSO-SVM, GA-SVM, and original SSA-SVM models in classification accuracy (97.62%) and precision (97.68%). The model achieved an average convergence iteration of 34 and a runtime of 28.5 seconds, demonstrating superior optimization efficiency. The recognition accuracy for all seven types of control board states exceeded 95%, significantly improving the precision and robustness of fault identification. Conclusion The proposed ISSA-SVM fault diagnosis method effectively addresses the intelligent diagnosis problem of main control boards in medical equipment without circuit diagrams, offering strong practical value and broad application prospects.

To address the issues of insufficient concurrent processing capacity of traditional electronic medical record systems, as well as inefficiency and security flaws in data access control methods, the research designs an electronic medical record system based on a cloud platform architecture and employs a ciphertext-policy attribute-based encryption algorithm to optimize data access. The effectiveness of the system is evaluated by testing the response time and throughput of the system under different concurrent scenarios, as well as the key generation time, encryption and decryption durations of the ciphertext-policy attribute-based encryption algorithm with different numbers of attributes. The results show that the system has a response time of less than 1200ms and a throughput of 2700 under the 5000-concurrent scenario. At 12K concurrency, the maximum throughput is 3924, with all response times being less than 3000ms. For the ciphertext-policy attribute-based encryption algorithm, the key generation time is only 1247ms when the number of attributes is 25, and the encryption and decryption durations are only 824ms and 901ms respectively when the number of attributes is 20. This indicates that the cloud platform-based electronic medical record system and data access optimization method designed in the research have strong feasibility and practical application potential, effectively solving the drawbacks of traditional systems, providing strong support for the efficient and secure operation of electronic medical record systems, and being of great significance to the informatization construction of the medical industry.

Echocardiography is a core non-invasive tool for diagnosing cardiovascular diseases. However, its dynamic imaging characteristics make it difficult to standardize image acquisition and the analysis process highly dependent on the operator's experience, which leads to significant differences in diagnoses among different physicians and limits the diagnostic efficiency and objectivity. In recent years, the breakthrough progress of artificial intelligence (Artificial intelligence, AI) especially deep learning (Deep learning, DL) technology has provided new opportunities for optimizing the ultrasound workflow. This article mainly reviews the latest research progress and application value of AI technology based on DL in echocardiography image analysis, focusing on the application of AI technology in image segmentation, cardiac function assessment, and cardiovascular disease diagnosis, etc. At the same time, this article also deeply analyzes the current challenges and looks forward to future development directions, aiming to provide a reference basis for clinical physicians to understand the application value of AI tools and achieve more efficient and accurate heart disease diagnosis.

Objective Exploring the Feasibility of Knowledge Graph-Enhanced Deep Learning for AI-Driven Quality Control in Mammography.Methods A retrospective analysis was conducted on 6,961 mammographic studies. A multi-tiered knowledge graph was constructed by synthesizing empirical evidence and domain expertise. Ten board-certified radiologists performed three rounds of annotation guided by this knowledge graph. The annotated data were utilized to train an AI-based quality control (AI-QC) model. An expert adjudication panel established the reference standard for evaluating AI-QC performance. To evaluate the application effectiveness of AI-based quality control, the weighted average precision was adopted, and a paired samples t-test was conducted to compare the precision differences between AI-based and manual quality control. Results  The AI-QC system achieved mean AUC values ≥0.940 across all four standard mammographic views. The right craniocaudal (RCC) view attained the highest AUC of 0.963. Following expert adjudication, the WAP exceeded 0.970 for all views, demonstrating statistically superior performance compared to manual QC. Conclusion The knowledge graph-deep learning integrated approach for mammographic QC demonstrates superior performance to manual methods, providing Objective and effective image quality evaluation with promising clinical applicability.

Objective Based on remote photoplethysmography and 3D convolutional neural networks, an intelligent heart rate monitoring method is constructed to solve the problem of inaccurate detection results in the existing vital sign detection methods. Methods The study proposes to combine remote photoelectric volumetric tracing with 3D convolutional neural network, and introduce singular spectrum analysis and adaptive filtering algorithm to optimize it. The metrics selected for the validity assessment experiments include the squared absolute error between the predicted value and the true value, the root-mean-square error, the Pearson's correlation coefficient, the consistency assessment plot, and the confidence interval. Results The ablation experiments revealed that the proposed algorithm can effectively reduce the mean absolute error and root mean square error between the predicted and true values, and the lowest of the two can reach 0.46 bpm and 1.92 bpm.The confidence intervals of the proposed method are [-2.5, 2.5], and the fit between the predicted value curves and the true heart rate curves is also high. Conclusion he vital sign detection method based on remote photoelectric volumetric tracing with 3D convolutional neural network can accurately predict heart rate values and has significant advantages in long-term vital sign monitoring.

In order to improve the accuracy of predicting the residual risk of urinary stones after ureteroscopic lithotripsy and reduce the postoperative intervention rate, a post lithotripsy stone residual risk prediction model based on optimized Extreme Gradient Boosting (XGBoost) was proposed. This model combines an improved sparrow search algorithm to optimize the XGBoost model, and improves the quality of hyperparameter search by introducing Circle chaotic mapping initialization population, nonlinear decreasing inertia weight, and Levy flight strategy. A comparative experiment was conducted based on the CICIDS 2018 dataset and a real postoperative follow-up dataset from a certain hospital. The experimental results show that the optimized XGBoost model has an AUC value of 0.94 on both the training and testing sets, with a classification accuracy of 92.49%. In normal traffic scenarios, the AUC is as high as 0.96, and the F1 score and recall rate are both 0.91, significantly better than the comparison model. The results indicate that the model can stably predict the risk of residual stones, providing reliable support for preoperative risk assessment and personalized treatment.

Objective To address three major challenges in hysteroscopic image lesion detection—high computational complexity, low recall rate for small lesions (<5mm), and difficulties in deploying models on primary care devices—this study proposes a lightweight real-time detection model. Methods A Multi-Scale Feature Pyramid (MSFP) architecture was constructed to enhance small lesion feature extraction through cross-layer feature fusion. A Depthwise Separable Self-Attention (DSSA) module was designed to decompose feature interactions into intra-channel self-attention and pointwise convolution, significantly reducing parameter count. A Dynamic Region Attention (DRA) mechanism was introduced to dynamically adjust weights in 4×4 grid regions based on lesion distribution probability matrices, improving localization robustness. Gradient-aware pruning and 8-bit integer quantization were combined to achieve efficient model compression. Results On the HS-CMU dataset (3,385 hysteroscopic images), the model achieved an AP@0.5 of 82.1% with only 2.1M parameters and an inference speed of 48 FPS. The recall rate for small lesions (<5mm) improved to 76.3%, and under Gaussian noise (σ=0.1) and motion blur (kernel=15) interference, the recall rate fluctuation remained below 3.5%, significantly outperforming mainstream models. Conclusion The MSFP-DSSA-DRA co-design achieves a balance between accuracy and efficiency, providing a low-cost, highly robust real-time lesion localization solution for primary care settings. This model can assist physicians in improving disease detection rates and holds significant clinical value for widespread adoption.

Objective Design an integrated system based on the open-ended coaxial probe method for real-time measurement of the dielectric properties of biological tissues. Analyze the accuracy of the system in measuring tissue dielectric properties and study the impact of probe pressure on the measurement of dielectric properties. Methods Based on components such as a microcontroller, pressure sensor, and signal processing circuit, construct a device for measuring reflection coefficient and probe pressure; develop host computer software with calibration and tissue dielectric property calculation functions to display the probe pressure and the dielectric properties of the measured tissue in real-time. After building the measurement system, the dielectric properties of methanol, n-propanol, liver phantom, glioma and normal brain tissue were measured, and the dielectric properties of the system were verified according to the standard parameters such as dispersion coefficient, average percentage error and relative difference. Results The average dispersion coefficients of the relative dielectric constant and conductivity of methanol and n-propanol in the frequency sweep range are 0.0026 and 0.0827, respectively. The average percentage errors were 0.85 % and 16.01 %, respectively. The average dispersion coefficients of the relative dielectric constant and conductivity of isopropyl alcohol in the frequency sweep range are 0.0028 and 0.0078, respectively. The average error percentages were 2.43 % and 14.38 %, respectively. The relative differences of dielectric constants between glioma tissues and normal brain tissues at 64MHz, 128MHz and 298MHz frequencies were 57.28 %, 48.97 % and 48.85 %, respectively. The relative differences of electrical conductivity were 73.61 %, 69.9 % and 66.26 %, respectively. Conclusion This system can be used for the real-time measurement of the dielectric properties of biological tissues in vitro, while also providing a reference for studying the impact of probe pressure on dielectric property measurement. It holds promise to become an auxiliary tool for clinical diagnosis and treatment in the future.

This paper presents a comparative study of the registration requirements for in vitro diagnostics (IVD) in China and ASEAN. Since the implementation of the "Belt and Road" initiative, economic cooperation between China and ASEAN has become increasingly close, particularly in the field of medical devices. By analyzing the similarities and differences in regulations, product classification, registration filing materials, and registration paths for IVD in China and ASEAN, the paper proposes suggestions for promoting cooperation in IVD registration between the two regions. The study indicates that while China and ASEAN share common regulatory objectives, there are significant differences in the details of implementation. Strengthening cooperation in product registration standardization and regulatory mutual recognition between the two regions will help facilitate market access, enhance product safety and efficacy, and promote the development of the health industry in both regions.

Objective Based on real clinical data research, systematically analyze the failures of digestive endoscopy equipment and propose targeted preventive maintenance strategies, aiming to improve equipment utilization efficiency, reduce maintenance costs, and ensure medical safety. Methods A retrospective analysis was conducted, selecting 183 cases of endoscope failure data from the digestive endoscopy center of a tertiary hospital between January 2021 and December 2024. The chi-square test was used to comprehensively analyze failure rates, failure phenomena, causes, locations, and maintenance levels. Results The overall failure rate of the digestive endoscope was 0.16%, with significant differences in distribution (P＜0.001). The most frequent failure phenomena were angle issues (0.11%) and component wear/damage issues (0.11%); among the causes of failure, aging accounted for the highest proportion (38.96%); high-frequency failure locations were concentrated in the insertion section (35.65%) and operating section (33.18%), and the location distribution showed statistical differences (P＜0.05); maintenance levels were predominantly in class D (62.30%), with significant differences in maintenance level distribution (P＜0.001); high-frequency failures of digestive endoscopes primarily occurred in the 1-3 years and 7-9 years usage periods. Conclusion Based on the analysis of real data on endoscope failure maintenance, a comprehensive optimization plan is proposed, covering standardized training, multi-level maintenance systems, targeted maintenance for high-frequency failures, information management, improvement of endoscope management systems, and extending service life, providing a theoretical basis for reducing the clinical endoscope failure rate.

Objective High-frequency irreversible electroporation (HF-IRE), an emerging minimally invasive and non-thermal ablation technology, demonstrates potential in neuro-oncology due to its precise targeted ablation and neuroprotective advantages. However, standardized treatment parameters remain lacking. This study aims to evaluate the safety and partial efficacy of HF-IRE in ablating orthotopic brain tumors in rats, providing critical parameter references for clinical translation. Methods Orthotopic brain tumor models were established in rats, and HF-IRE treatment was administered at three electric field intensities (900, 1200, and 1500 V/cm). Tumor volume was dynamically monitored via magnetic resonance imaging (MRI), while hematoxylin-eosin (H&E) staining and TdT-mediated dUTP nick-end labeling (TUNEL) were employed to assess histopathological damage and apoptosis in brain tissues, systematically evaluating safety and efficacy. Results Electric field intensity positively correlated with tumor inhibition rates (82%, 95.6%, and 95.8%, respectively). However, the 1500 V/cm group exhibited significant histological damage, including cerebral edema and neuronal vacuolization. The 1200 V/cm group achieved a 95.6% tumor inhibition rate with only reversible mild edema and scattered apoptotic bodies, showing significantly reduced neural damage compared to the high-intensity group. Conclusion 1200 V/cm represents the optimal balanced parameter for HF-IRE in brain tumor treatment, achieving a significant tumor suppression rate (>90%) while minimizing adverse effects and neuroinjury risks associated with HF-IRE.

To address the complexity of fault types and low efficiency of manual diagnosis in medical equipment power supplies, an intelligent fault identification method based on an improved Convolutional Neural Network (CNN) is proposed. A total of 30,000 time-series signal samples covering four types (normal, switching transistor fault, PWM chip fault, and MCU fault) were collected to construct a multi-channel dataset. The CNN model incorporates batch normalization, L2 regularization, a squeeze-and-excitation (SE) attention module, and the AdamW optimizer to enhance feature extraction efficiency and robustness. Model performance was evaluated using accuracy, recall, and F1-score, and robustness was tested under noise interference and data loss conditions. Experimental results show that the improved CNN achieved an accuracy of 99.2%, a recall of 98.9%, and an F1-score of 99.0% on the test set, outperforming SVM and conventional CNNs by 2.0% to 7.7%. Under Gaussian noise with variance σ² = 0.1 and up to 30% feature loss, the model maintained an accuracy of approximately 93%. The proposed method demonstrates high precision, strong robustness, and deployability, offering technical support for intelligent maintenance of medical equipment.

Objective To construct the prediction model of chronic rotator cuff injury degree based on MR Water-lipid separation quantitative index to provide reference for the subsequent diagnosis and treatment of rotator cuff injury. Methods A retrospective study was conducted on 169 patients with chronic rotator cuff injury who underwent MR examination in our hospital from January 2021 to January 2024. According to the severity of the injury, they were divided into a partial tear group (123 cases) and a complete tear group (46 cases); Compare the clinical characteristics of the tear group and the severe tear group. Multivariate analysis of predictive factors for the degree of chronic rotator cuff injury. Clinical efficacy analysis of MR related parameters for predicting the degree of chronic rotator cuff injury. Results There were statistically significant differences (P<0.05) in the maximum diameter, acromion type, and fat fractions of the supraspinatus and infraspinatus muscles of the humeral tuberosity between the two groups. The results of logistic multivariate analysis showed that the maximum diameter of cystic changes in the greater tuberosity of the humerus, as well as the fat fractions of the supraspinatus and infraspinatus muscles, were independent predictors of the degree of chronic rotator cuff injury (P<0.05). A column chart model was constructed using the maximum diameter of the humeral tuberosity cystic change, the fat fraction of the supraspinatus muscle, and the fat fraction of the infraspinatus muscle. The ROC curve was used to predict the degree of chronic rotator cuff injury using the maximum diameter of the humeral tuberosity cystic change, the fat fraction of the supraspinatus muscle, the fat fraction of the infraspinatus muscle, and the column chart. The areas under the curves were 0.725, 0.734, 0.729, and 0.864, respectively. Conclusion The quantitative indexes of fat fraction of supraspinatus muscle and infraspinatus muscle showed better efficacy than traditional MR Indexes in predicting of chronic rotator cuff injury degree. The column chart prediction model constructed using the above factors shows good performance in predicting the severity of tearing in patients, and is worth further analysis in the work.

Objective To analyze the current status and development trends of biomaterials in rehabilitation. Methods Literatures with the theme of biomaterials in rehabilitation were retrieved from the Web of Science core database. The R language was used to visualize analysis, including publication time, cited references, authors, keywords, institutions, and countries/regions. Results A total of 1965 articles, 9946 authors, 5629 keywords, and 76259 references were extracted. Research on biomaterials in rehabilitation has shown a fluctuating upward trend, with the United States and China taking the lead. Global scientific research focused on biomaterials, tissue engineering, bone regeneration, biocompatibility, hydrogels, etc. Conclusion China's research in this field is developing rapidly, and more results will be produced in the future. This study uses R language analysis tools to examine biomaterials research in rehabilitation from a metrological perspective, providing information to inform future decisions regarding the development of this field.

To enhance the objectivity and accuracy of syndrome differentiation in traditional Chinese medicine and address the issues of reliance on expert experience and strong subjectivity, the research focuses on constructing an intelligent diagnosis and treatment model based on multimodal clinical data such as tongue images, pulse images, and symptom texts. Firstly, an improved convolutional recurrent neural network is designed to extract the spatial features of the tongue image, process the temporal features of the pulse image and the symptom text sequence. Introduce the spatial reconstruction unit and adaptively capture the irregular pathological features such as tongue cracks and ecchymosis through the dynamic deformable convolution kernel; Design the channel reconstruction unit and screen the key semantic channels by combining depth-separable convolution and compression-incentive mechanisms; A three-layer cross-modal attention mechanism is adopted to dynamically integrate multi-modal features, and a three-level syndrome type priority matching algorithm is integrated to simulate the priority syndrome differentiation process of the main symptoms in traditional Chinese medicine. The results show that the differentiation accuracy rate reaches 88.69%, which is superior to the support vector machine and recurrent neural network. In the tongue image classification task, the AUC of the mirror tongue reached 0.9663, and the generalization accuracy of the test set remained stable at 68.48%. The receiver operation characteristic curve and the precision-recall curve verified the performance balance of the model. The conclusion proves that this model effectively solves the problems of extracting and fusing multimodal features in traditional Chinese medicine, significantly improves the level of objective syndrome differentiation, and provides a reliable tool for intelligent clinical diagnosis and treatment.

Objective To investigate the impact of magnetic resonance imaging (MRI) combined with musculoskeletal ultrasound on the assessment and correlation of musculoskeletal function adjacent to the dorsal root ganglion (DRG) in patients with zoster-associated pain (ZAP). Methods Included patients with ZAP from June 2020 to July 2023 in our hospital, divided into postherpetic neuralgia group (n=45) and herpes zoster neuralgia group (n=64), comparing the general data, MRI, and musculoskeletal ultrasound parameters between the two groups. Compared the Oswestry Disability Index (ODI) and Visual Analog Scale (VAS) scores of both groups. Conducted multiple linear regression analysis of MRI parameters and adjacent musculoskeletal functional indicators of the DRG. Used multivariate logistic regression analysis to identify independent risk factors affecting postherpetic neuralgia in patients. Constructed a restricted cubic spline model to analyze the dose-response relationship between MRI, musculoskeletal ultrasound parameters, adjacent musculoskeletal function of the DRG, and postherpetic neuralgia. Plotted the receiver operating characteristic curve (ROC) and calibration curve to assess the predictive accuracy of the model. Results The differences between the two groups of patients in terms of hypertension history and smoking are statistically significant (P<0.05). Patients with postherpetic neuralgia had significantly higher values in the surface volume ratio of the DRG, maximum 2D diameter pillar, median, large area high grayscale emphasized parameters, musculoskeletal ultrasound score (MUS), pain numerical rating scale score (NRS), ODI, and VAS compared to patients with herpes zoster neuralgia (P<0.001). The results of the multifactor logistic regression analysis showed that hypertension, smoking, surface volume ratio, maximum 2D diameter pillar, median, short-range low grayscale emphasis, large area high grayscale emphasis, MUS, and NRS are independent influencing factors for postherpetic neuralgia in patients (P<0.05). The results of the restricted cubic spline model analysis showed that there is a nonlinear dose-response relationship between MRI, musculoskeletal ultrasound parameters, adjacent muscle and bone function of the DRG, and the risk of developing postherpetic neuralgia. The ROC curve results showed an AUC of 0.859 (95% CI: 0.813~0.902), and the calibration curve Brier score was 0.087, indicating a high predictive ability of the model. Conclusion MRI combined with musculoskeletal ultrasound imaging features can assess the musculoskeletal function adjacent to the DRG in ZAP patients. The MRI and musculoskeletal ultrasound parameters are correlated with the functional parameters of the adjacent musculoskeletal structures to the DRG, and the model shows high discriminative ability and accuracy.

ObjectiveThis study aims to systematically analyze the yield fluctuations of the Fastlab 2 automatic synthesis system in the actual production of [18F]FDG for the first time through the Statistical Process Control (SPC) method, identify the key influencing factors, and put forward specific and operational quality optimization suggestions. Methods A retrospective analysis was conducted on the production data of 43 batches of [18F]FDG in our center within the past 3 months. The X-bar and R control charts were used to evaluate the process control of the yield, and the coefficient of variation (RSD) was combined to evaluate the volatility. Parameter traceability (such as vacuum pressure, hydrolysis temperature, and operational compliance) was conducted for 3 failed batches (with a yield of 0%) and 3 low-yield batches (<40%). Results The success rate of the Fastlab 2 system was 93.02%. The yield of successful batches (excluding 3 failures) ranged from 31% to 91%, with an average yield of 81.6% and an RSD of 18.64%. The X-bar chart and R chart show that most batches are within the control limit, and the overall production process is stable. However, some local batches have abnormal vacuum pressure (such as 190 mbar vs.) Fluctuations in yield were caused by excessive hydrolysis temperature (standard 600 mbar), high hydrolysis temperature (93℃ vs. standard 85-88℃), and operational deviations (improper installation of ferrules and failure to conduct water bag checks). The failed batches were mainly related to vacuum faults of the equipment (2 times) and operational errors (1 time). Conclusion: The Fastlab 2 system has a good ability of automated synthesis, but the fluctuation of yield still needs attention. By optimizing the startup process confirmation of the equipment, improving the installation process of reagent cartridges, strengthening the training of operators and standardizing the management of the experimental environment, the consistency of the system and the production success rate can be effectively enhanced, providing a referenceable optimization path for similar radiopharmaceutical production units.

ObjectiveThis study based on the "structure-process-outcome" three-dimensional quality model, aims to establish a sensitive indicator system applicable to the management quality assessment of hemodialysis machines, providing a quantitative and objective monitoring basis for improving the management quality of hemodialysis machines. MethodsIn the research process, first, the framework of the indicator system was initially constructed by combining literature research and qualitative interviews; then, two rounds of expert consultations were carried out using the Delphi method; finally, the weights of various indicators were determined through the Analytic Hierarchy Process. ResultsThe constructed indicator system covers 3 first-level indicators, 6 second-level indicators, and 21 third-level indicators. Among them, the sensitive indicators with the top three weights are the intact rate of hemodialysis machine equipment (0.1001), the implementation rate of regular maintenance of hemodialysis machines (0.1000), and the failure rate of hemodialysis machine equipment (0.0430). The effective recovery rates of the two-round questionnaires both reached 100%. The expert authority coefficients were 0.836 and 0.862 respectively, and Kendall's W coordination coefficients were 0.542 and 0.568 respectively (P < 0.001). ConclusionThe indicator system constructed based on the three - dimensional quality model can provide strong support for the standardized management of hemodialysis machines. It can be popularized and applied in the hemodialysis rooms and intensive care units of hospitals at all levels in China that use hemodialysis machines, promoting the standardization process of hemodialysis machine management.

The linear accelerator Vital Beam occupies a central position in tumor radiotherapy, and its imaging accuracy is crucial for the effectiveness of radiotherapy. This article systematically elaborates on the standardized operation process of image calibration for the Varian Vital Beam linear accelerator, with a focus on the key steps of kilovolt KV mode calibration, Air Norm and HU calibration for cone beam CT. Based on long-term quality control data (January 2023 and July 2024), the impact of strict execution of the calibration process for 10 months before and after on parameters such as image quality and HU value consistency was quantitatively evaluated. The study used the Catphan 504 phantom for detection. The results showed that the spatial resolution of the calibrated image remained at a high level with a slight increase of 14.29, and the overall low contrast resolution improved by 18.34%; The standard deviation (SD) of HU value was improved by 3.03% for the overall noise of the water model before calibration, 4.58% for the density accuracy of key materials such as polytetrafluoroethylene (Teflon), 2.79% for local noise, and 53.94% for uniformity, significantly improving image accuracy. In addition, the corrective effect of image calibration was analyzed based on typical fault cases. Strictly following the calibration process is the key to ensuring high-precision implementation of image-guided radiotherapy. This study provides comprehensive technical reference and basis for radiotherapy workers, which is of great significance for improving the quality and safety of radiotherapy.

Aiming at the core issues in pediatric emergency triage, such as strong disease dynamics, high insidiousness of symptoms, high rate of missed diagnosis of severe cases due to high concurrent resources, unbalanced resource allocation and low clinical trust. The research utilizes the dual-actor module to optimize the proximal strategy optimization algorithm. By dynamically adjusting the triage strategy, the processing ability of time series data is enhanced. Combined with the non-dominated ranking genetic algorithm, an efficient and accurate intelligent triage classification model is constructed. The research results show that the average return rate of the improved proximal strategy optimization algorithm with the introduction of the dual-actor module can reach 88.5%, and the number of environmental interactions that achieve the highest return rate is only 50 times. In addition, the missed diagnosis rate and over-triage rate of the intelligent triage and classification model for pediatric emergency proposed in the study were at least 2.8% and 1.5% respectively. After adding noise, the area under the precision recall curve of this triage model only decreased by 0.041. The result data show that this model, through the collaboration of dual algorithms, significantly improves the efficiency and accuracy of pediatric emergency triage, effectively solves the problems of missed diagnosis of severe cases and misallocation of resources, and provides a practical solution for the intelligent emergency system.

ObjectiveTo explore the influence of maintenance strategies for micro-injection pumps on the operating time of the equipment and the maintenance nodes. Methods A retrospective analysis was conducted on the fault data of 50 randomly selected microinjection pumps in our hospital in 2023. They were divided into two groups according to the maintenance strategies: the control group adopted traditional passive maintenance, and the experimental group adopted active preventive maintenance. Taking the operating time and status of two groups of equipment as events, the Kaplan-Meier method was used to conduct survival analysis on the fault data under different maintenance strategies, and the survival rate and survival time of the equipment under each strategy were presented through charts. Meanwhile, three chi-square test methods were adopted to test the statistical differences in the survival functions of the two groups of samples (P), and the survival function curves of different maintenance strategies were compared. ResultsThe median survival time for equipment in the experimental group was 49 weeks, while it was only 27 weeks for the control group, indicating that the maintenance strategy adopted by the experimental group significantly improved the survival rate, extended the service life, and reduced the risk of failure. Additionally, through analyzing the survival function curve, it is evident that the performance status of equipment in the experimental group significantly declined after 33 weeks, providing a scientific basis for preventive maintenance of medical equipment. ConclusionAnalyzing the fault status of micro-infusion pumps under different maintenance strategies using the Kaplan-Meier method is feasible. Proactive maintenance strategies can effectively enhance equipment performance and lifespan, ensuring good operational conditions. Combined with the survival function curve, key time points can be clearly identified, allowing for preventive maintenance measures to be taken, thereby improving the reliability and safety of injection pumps.

Muscle loss has been shown to be closely associated with an increased risk of post-transplant complications in transplant recipients. Imaging diagnosis is currently a common method for the diagnosis of sarcopenia through the use of a series of imaging methods, including dual-energy X-ray absorptiometry, DXA, computed tomography (CT), magnetic resonance imaging (MRI), ultrasonic (US), and AI-assisted diagnostic techniques can accurately assess muscle mass, Thus diagnosing sarcopenia. This article will summarize and analyze the research progress on the correlation between sarcopenia and poor prognosis after transplantation assessed by different imaging methods, aiming to provide a reliable reference for clinicians to select the most appropriate diagnostic method.

With the continuous development of the smart healthcare system, artificial intelligence has been widely applied in the field of medical care. As an important department for surgical treatment and rescue, the risk management in the operating room is particularly crucial. At present, there are few reports on the relevant research of artificial intelligence in the risk management of operating rooms. Therefore, this study aims to summarize the current application status of artificial intelligence in the risk management of operating rooms at home and abroad, and analyze five high-risk factors including the handover of surgical patients, surgical safety verification, intraoperative pressure ulcers, perioperative hypothermia and pathological specimen management. To clarify the current research status and limitations at home and abroad, explore and look forward to the development direction of intelligent technologies in the above-mentioned risk management in China in the future, with the aim of providing a reference for the further development of artificial intelligence technology in the field of operating room risk management.