Digital intelligent management and control system for ICU medical instruments and equipment and management method of digital intelligent management and control system

Abstract

The invention discloses an ICU medical instrument equipment digital intelligent management and control system and a management method thereof, relates to the technical field of medical instrument management, and constructs a collaborative management mechanism of multi-dimensional data fusion and dynamic optimization. Equipment operation parameters, patient physiological indexes and environment interference data are collected in real time, an equipment state, physiological association and environment interference three-dimensional incidence matrix is initiatively constructed, a comprehensive credibility index is generated through dynamic coupling weights, causal chain verification and priority dynamic sorting of multi-equipment alarm events are achieved, and the reliability of the alarm events is improved. And the critical event identification accuracy is obviously improved. And forming a self-optimization system through closed-loop linkage of an equipment cooperative control instruction and a four-dimensional feedback index: when the comprehensive deviation degree exceeds a threshold value, reversely correcting a weight coefficient based on a gradient descent algorithm, so that the system continuously adapts to clinical scene changes.

Description

Technical Field

[0001] The present invention relates to the technical field of medical instrument management, and specifically to a digital intelligent management and control system for ICU medical instruments and equipment and a management method thereof. Background Art

[0002] With the rapid growth in the types and number of medical devices in intensive care units (ICUs), traditional equipment management models face severe challenges. ICU equipment management primarily relies on standalone operation and manual monitoring, which presents the following technical bottlenecks:

[0003] Current equipment monitoring systems only collect the operating parameters of a single device (such as ventilator airway pressure and ECG waveforms) and lack collaborative perception of the patient's physiological state (such as blood oxygen saturation trends) and environmental factors (such as device displacement and tubing traction). This makes it difficult to effectively verify the correlation between device alarms and the actual clinical status. Studies have shown that up to 68% of device alarms in ICU settings are clinically disruptive false alarms (such as electrode detachment misinterpreted as arrhythmia), which seriously interfere with medical decision-making. In addition, existing systems cannot dynamically evaluate the effectiveness of device collaborative control, making it difficult to optimize parameters based on actual execution feedback, resulting in inefficient equipment management. Furthermore, the lack of an intelligent priority sorting mechanism for multiple device alarms means that medical staff often need to manually process a large number of invalid alarms, increasing their workload and the risk of misoperation. To address these issues, existing technologies urgently need to be improved. Summary of the Invention

[0004] In response to the deficiencies in the prior art, the present invention provides a digital intelligent management and control system for ICU medical instruments and equipment and a management method thereof.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows:

[0006] In a first aspect, the present invention discloses a digital intelligent management method for ICU medical instruments and equipment, comprising the following steps:

[0007] Real-time collection of device layer data, physiological layer data and environmental layer data;

[0008] Constructing a three-dimensional correlation matrix, mapping the device layer data, physiological layer data, and environmental layer data to the device status dimension, physiological relevance dimension, and environmental interference dimension, respectively, calculating the dynamic coupling weights between the dimensions, and generating a comprehensive credibility index;

[0009] Based on the comprehensive credibility index, the authenticity of multiple device alarm events is verified, a dynamic priority sequence is generated according to the verification results, and device collaborative control instructions are output for events that meet the preset critical threshold;

[0010] Obtaining multi-dimensional feedback data after the execution of the device collaborative control instruction, the multi-dimensional feedback data including physiological data response curve, device status recovery time, medical staff operation time and resource consumption data;

[0011] Preprocessing the multi-dimensional feedback data, and calculating a physiological recovery efficiency index, an equipment coordination efficiency index, an operation convenience index, and a resource utilization efficiency index based on the preprocessed multi-dimensional feedback data;

[0012] Compare each evaluation indicator with the preset expected indicator value to obtain the deviation of each evaluation indicator, assign weights to each evaluation indicator according to the preset importance rules, and calculate the comprehensive deviation;

[0013] Determine whether the comprehensive deviation exceeds a preset adjustment threshold, and if so, reversely correct the weight coefficient of the three-dimensional correlation matrix according to a predefined optimization strategy;

[0014] Based on the weight coefficient of the revised three-dimensional correlation matrix, the comprehensive credibility index is recalculated, and the authenticity of multiple device alarm events is verified.

[0015] In a second aspect, the present invention discloses a digital intelligent management and control system for ICU medical instruments and equipment, which applies the above-mentioned digital intelligent management method for ICU medical instruments and equipment, including:

[0016] Multi-source data acquisition module, used to collect device layer data, physiological layer data and environmental layer data in real time;

[0017] A three-dimensional data processing and analysis module is used to construct a three-dimensional correlation matrix, map the device layer data, physiological layer data, and environmental layer data to the device status dimension, physiological relevance dimension, and environmental interference dimension, calculate the dynamic coupling weights between the dimensions, and generate a comprehensive credibility index;

[0018] A dynamic priority control and execution module, configured to verify the authenticity of multiple device alarm events based on the comprehensive credibility index, generate a dynamic priority sequence based on the verification results, and output device collaborative control instructions for events that meet a preset critical threshold;

[0019] A feedback data acquisition module is used to obtain multi-dimensional feedback data after the device collaborative control instruction is executed. The multi-dimensional feedback data includes physiological data response curves, device status recovery time, medical staff operation time and resource consumption data;

[0020] An evaluation index calculation module is used to preprocess the multi-dimensional feedback data and calculate the physiological recovery efficiency index, the equipment coordination efficiency index, the operation convenience index and the resource utilization efficiency index based on the preprocessed multi-dimensional feedback data;

[0021] The comprehensive deviation calculation module is used to compare each evaluation indicator with the preset expected indicator value to obtain the deviation of each evaluation indicator, assign weights to each evaluation indicator according to the preset importance rules, and calculate the comprehensive deviation;

[0022] A feedback correction module is used to determine whether the comprehensive deviation exceeds a preset adjustment threshold, and if so, reversely correct the weight coefficient of the three-dimensional correlation matrix according to a predefined optimization strategy;

[0023] The correction execution module is used to recalculate the comprehensive credibility index based on the weight coefficient of the corrected three-dimensional correlation matrix and verify the authenticity of multiple device alarm events.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. By synchronously collecting equipment operating parameters, patient physiological data, and environmental interference data, a three-dimensional correlation matrix is ​​constructed to achieve cross-dimensional data fusion verification, solve the problem of false alarm identification, and eliminate the limitations of independent alarms of single devices;

[0026] 2. Based on a dynamically generated comprehensive credibility index, causal analysis is performed on conflicting alarms, raising the priority of directly life-threatening events such as ventilator obstruction to three times that of infusion pump failures;

[0027] 3. Through the closed-loop feedback mechanism, the weight coefficient is dynamically corrected to enable the device management strategy to be adaptively optimized according to the usage scenario. The time decay factor is used to prevent the model from overfitting and ensure the stability of long-term operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The disclosure of the present invention is described with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0029] Figure 1 A diagram showing the steps of the method of the present invention;

[0030] Figure 2 It is a data flow diagram of the present invention;

[0031] Figure 3 is a flow chart for calculating the comprehensive credibility index of the present invention;

[0032] Figure 4 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0033] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, a person skilled in the art can propose a variety of interchangeable structural modes and implementation modes. Therefore, the following specific embodiments and drawings are only exemplary descriptions of the technical solution of the present invention and should not be regarded as the entire invention or as a limitation or restriction of the technical solution of the present invention.

[0034] Application Overview

[0035] In traditional ICU equipment monitoring systems, standalone operation only collects discrete device parameters such as ventilator airway pressure and ECG waveforms. Dynamic physiological data such as patient oxygen saturation trends are not integrated, nor are environmental interference factors such as device displacement and tubing traction mechanical characteristics considered. Consequently, the multidimensional correlation between device alarms and the patient's true clinical status cannot be effectively verified. For example, when a ventilator triggers an abnormal airway pressure alarm, it's impossible to simultaneously verify whether the patient's spontaneous breathing trigger signal is coupled with the mechanical interference caused by environmental tubing traction, leading to a large number of clinically disruptive false alarms.

[0036] For example, in a multi-device scenario involving a ventilator, an ECG monitor, and an extracorporeal circulation machine, if the ventilator triggers an alarm due to abnormal airway pressure caused by tubing traction, the system only verifies the event based on the pressure threshold of the individual device, ignoring the patient's physiological state of maintaining normal blood oxygen saturation and no signs of respiratory distress. It also ignores data on sudden changes in tubing stress caused by device displacement. Medical staff are forced to manually troubleshoot ventilator alarms without cross-validation from multi-dimensional data, resulting in delayed clinical decision-making and reduced efficiency in device collaborative response.

[0037] If these issues are not addressed, false alarms will lead to a misalignment between device control instructions and actual clinical needs, resulting in ineffective treatments or delayed responses to critical alarms. The independent operation of devices exacerbates the data silo effect, allowing environmental interference to continuously impact device monitoring accuracy, thereby reducing the overall reliability of the intensive care system. Furthermore, the lack of a closed-loop verification mechanism for multi-dimensional feedback data will hinder the dynamic optimization of device collaboration strategies, ultimately impacting the timeliness of treatment for critically ill patients.

[0038] When faced with the above problems, this application first explores how to establish a multi-dimensional data collaborative perception mechanism to identify the real clinical status. The traditional single-machine monitoring mode cannot distinguish whether the abnormal equipment parameters are caused by physiological changes in patients or environmental interference. For this, it is necessary to build a dynamic correlation model among equipment, physiology, and environment. This application further analyzes the coupling relationship of multi-device alarm events and finds that there is a lag in the verification of single-dimensional data, and a closed-loop correction mechanism with real-time feedback needs to be introduced. In response to the problem of confusion in alarm priority sorting, this application explores a dynamic verification method based on the credibility index, and optimizes the decision-making accuracy through dynamic adjustment of weight coefficients.

[0039] In this regard, Figure 1 As shown, the present application proposes a digital intelligent management method for ICU medical instruments and equipment, which is characterized by comprising the following steps:

[0040] Real-time collection of device layer data, physiological layer data, and environmental layer data; device layer data refers to the sensor signals, parameter settings, and alarm information generated by medical equipment during operation. This can be achieved by using an embedded data acquisition unit to obtain ventilator pressure waveforms and infusion pump flow rate data in real time. Physiological layer data refers to patient vital sign monitoring indicators. This can be achieved by using a multi-parameter monitor to collect ECG, blood oxygen, and blood pressure data and calculate the rate of change. Environmental layer data refers to the physical state of the device and the surrounding environmental parameters. This can be achieved by using an inertial measurement unit to obtain the device displacement trajectory and a tension sensor to detect the pulling force of the pipeline.

[0041] Construct a three-dimensional correlation matrix, map the device layer data, physiological layer data, and environmental layer data to the device status dimension, physiological relevance dimension, and environmental interference dimension respectively, calculate the dynamic coupling weights between each dimension, and generate a comprehensive credibility index; the three-dimensional correlation matrix refers to mapping the device, physiological, and environmental data to three orthogonal dimensions respectively. Specifically, this can be achieved by establishing a coordinate space of device status-physiological relevance-environmental interference and using dynamic coupling weights to calculate the interaction relationship between each dimension. The dynamic coupling weight refers to the coefficient that reflects the real-time correlation strength between the device operating status, physiological change trends, and environmental interference factors. Specifically, this can be achieved by using the covariance matrix decomposition method within a sliding time window to calculate the interaction strength between dimensions. The comprehensive credibility index refers to an evaluation indicator that quantifies the authenticity of device alarm events. Specifically, it can be calculated by weighted fusion of the device health index, physiological relevance index, and environmental interference assessment value using a linear combination formula.

[0042] Based on the comprehensive credibility index, the authenticity of multiple device alarm events is verified, and a dynamic priority sequence is generated according to the verification results. Device collaborative control instructions are output for events that meet the preset critical threshold. The dynamic priority sequence refers to the response order generated according to the authenticity verification results of the alarm event. Specifically, it can be achieved through the critical threshold filtering mechanism, combined with the basic weight of the device type and the causal relationship logic for sorting.

[0043] Obtain multi-dimensional feedback data after the execution of the device collaborative control instruction. The multi-dimensional feedback data includes physiological data response curve, device status recovery time, medical staff operation time and resource consumption data; multi-dimensional feedback data refers to the system response indicators after the execution of the device collaborative control instruction. Specifically, it can be achieved by recording the physiological data recovery curve, device restart time, medical staff intervention time and consumables consumption in the time series database.

[0044] Preprocess the multi-dimensional feedback data, and calculate the physiological recovery efficiency index, equipment coordination efficiency index, operation convenience index and resource utilization efficiency index based on the preprocessed multi-dimensional feedback data;

[0045] Each evaluation indicator is compared with the preset expected indicator value to obtain the deviation of each evaluation indicator, and a weight is assigned to each evaluation indicator according to the preset importance rule to calculate the comprehensive deviation; the comprehensive deviation refers to the overall difference between the evaluation indicator and the expected value, which can be achieved by using the weighted Euclidean distance algorithm and combining the indicator importance weight to calculate the synthetic result of the deviation value of each dimension.

[0046] Determine whether the comprehensive deviation exceeds a preset adjustment threshold, and if so, reversely correct the weight coefficient of the three-dimensional correlation matrix according to a predefined optimization strategy;

[0047] Based on the weight coefficient of the revised three-dimensional correlation matrix, the comprehensive credibility index is recalculated, and the authenticity of multiple device alarm events is verified.

[0048] The core innovation of this application lies in the multi-dimensional collaborative analysis of device operating parameters, patient physiological changes, and environmental interference factors through the construction of a three-dimensional correlation matrix. This method uses a dynamic coupling weight and feedback optimization mechanism to establish a closed-loop intelligent decision-making system. This method breaks through the limitations of traditional single-dimensional monitoring, verifies the authenticity of alarms through real-time calculation of a comprehensive credibility index, and continuously optimizes the evaluation model based on multi-source feedback data, effectively solving the technical problems of high clinical false alarm rates and difficult multi-device collaborative management.

[0049] like Figure 2 As shown in FIG, it is a data flow diagram of the present invention; as a preferred embodiment, the solution of the present application is specifically implemented as follows:

[0050] A multi-source data acquisition system is deployed in the ICU, including data interfaces of devices such as ventilators, electrocardiogram monitors, infusion pumps, patient vital sign sensors, and environmental monitoring equipment.

[0051] A three-dimensional correlation matrix is constructed, mapping parameters such as airway pressure of the ventilator to the device state dimension, mapping trends such as blood oxygen saturation to the physiological relevance dimension, and mapping device displacement trajectories to the environmental interference dimension. Through machine learning algorithms, dynamic coupling weights between dimensions are calculated to generate a comprehensive confidence index between 0 and 1.

[0052] When receiving alarm signals from multiple devices, the system verifies the authenticity of each alarm event based on the comprehensive confidence index. For example, when the ventilator alarms, the system synchronously analyzes the patient's blood oxygen saturation changes and ventilator displacement to determine whether the alarm is caused by environmental interference. According to the verification result, the system generates a dynamic priority sequence and issues device coordination control instructions for high-priority events.

[0053] After the execution of the control instructions, multi-dimensional feedback data such as patient physiological indicator change curves, device state recovery times, and medical staff response times are obtained. These data are standardized and pre-processed, and various evaluation indicators are calculated.

[0054] The evaluation indicators are compared with the preset expected values, and a comprehensive deviation is calculated. If the comprehensive deviation exceeds the preset threshold, the system will correct the weight coefficients of the three-dimensional correlation matrix according to the predefined gradient descent algorithm. Subsequently, the authenticity of the alarm event is re-verified based on the new weight coefficients, forming a closed-loop optimization.

[0055] Through the above scheme, the accuracy of ICU medical device alarms can be effectively improved, and the interference of false alarms on medical staff can be reduced. The collaborative analysis of multi-dimensional data can more comprehensively reflect the true state of the patient, avoiding false judgments caused by single device parameter abnormalities. The dynamic priority sorting mechanism can help medical staff respond to critical alarms more quickly, improving rescue efficiency. The closed-loop feedback optimization mechanism can continuously improve the decision-making accuracy of the system, making device management more intelligent and personalized. In addition, this scheme can also reduce the data silo effect between devices, improving the reliability and collaborative efficiency of the entire ICU monitoring system.

[0056] The correction formula for the weight coefficient is further proposed in the application:

[0057] wherein, is the weight coefficient before correction, is the weight coefficient after correction, is the preset learning rate.

[0058] The correction formula uses the gradient descent principle to determine the parameter adjustment direction by calculating the partial derivative of the comprehensive deviation with respect to the current weight. As a step size adjustment factor, it controls the update amplitude of the weight coefficient. The partial derivative reflects the impact of the change in the weight coefficient on the comprehensive evaluation index. When the deviation increases, the weight coefficient is iteratively corrected in the opposite direction of the gradient.

[0059] Specifically, after each calculation of the comprehensive deviation, the system uses the backpropagation algorithm to obtain the partial derivative of the evaluation index with respect to the weight coefficient. The preset learning rate is multiplied by the partial derivative to obtain the weight adjustment amount, which can change dynamically according to the error sensitivity of the current system. For example, when the weight coefficient has a significant impact on the comprehensive deviation, the partial derivative value increases, and the product of the learning rate and the partial derivative prompts a larger adjustment of the weight coefficient. The preset learning rate initial value can be set to 0.05 and adjusted according to the exponential decay rule during the optimization cycle. In this way, the correction process of the weight coefficient not only maintains the accuracy of the adjustment direction, but also balances the convergence speed and stability through the learning rate parameter, ensuring that the dynamic coupling weight of the three-dimensional correlation matrix is ​​always in the optimal state.

[0060] For example, the preset learning rate can be set to 0.01. The comprehensive deviation is calculated based on the physiological recovery efficiency index, the equipment coordination efficiency index, the operational convenience index, and the resource utilization efficiency index. During the weight coefficient correction process, the partial derivative of the comprehensive deviation with respect to each weight coefficient is first calculated, and then iteratively updated according to the above formula. For example, assuming that a weight coefficient before correction is 0.6, and the partial derivative of the comprehensive deviation with respect to the weight coefficient is 0.2, then the corrected weight coefficient = 0.6-0.01*0.2=0.598. Through multiple iterations, the weight coefficient will gradually converge to the optimal value.

[0061] Through the above technical solution, the present application realizes the adaptive adjustment of the weight coefficient of the three-dimensional correlation matrix. As a result, the system can dynamically optimize the weight configuration according to the actual operation effect, thereby improving the accuracy of the comprehensive credibility index. Furthermore, this adaptive mechanism enables the system to have the ability of continuous learning and self-improvement, and can better adapt to the complex and changing medical environment of the ICU. Specifically, through the dynamic correction of the weight coefficient, the system can more accurately identify real alarm events, reduce the false alarm rate, and improve the work efficiency of medical staff.

[0062] This application further proposes:

[0063] Obtain historical operation success rate data and construct a feedback weight adjustment factor based on the historical operation success rate data. When the success rate is lower than the preset threshold in N consecutive operations, increase the preset learning rate to (1 + weight adjustment factor) times the original value, where the weight adjustment factor is the result of a logarithmic function calculation based on the number of historical failures.

[0064] Historical operation success rate data is generated by recording the execution results of device collaborative control instructions. The feedback weight adjustment factor is defined as the natural logarithm of the number of failures, specifically ln(number of failures + 1). The system periodically counts the execution results of device collaborative control instructions. When three consecutive operation failures are detected, the number of historical failures within that time window is automatically calculated. The number of failures is mapped to a weight adjustment factor using a natural logarithm function to avoid parameter mutations caused by linear growth. When the learning rate is increased exponentially, the gradient descent of the weight coefficients accelerates, prompting the three-dimensional correlation matrix to converge to its optimal state more quickly. For example, when the original learning rate is 0.1 and there are five consecutive operation failures, the adjusted learning rate becomes 0.1×(1+ln5)=0.1×2.609=0.2609, increasing the weight adjustment by 2.6 times and effectively shortening the model training cycle. This mechanism, combined with the backpropagation algorithm, significantly improves the efficiency of parameter optimization in low success rate scenarios while maintaining system stability.

[0065] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0066] Historical operation success rate data is obtained and used to construct a feedback weight adjustment factor. The system sets the threshold for consecutive operations (N) to 5 times and the success rate to 80%. If the success rate falls below 80% over 5 consecutive operations, the preset learning rate is increased from 0.01 to (1 + weight adjustment factor). The weight adjustment factor is calculated based on the logarithm of the number of historical failures. The specific calculation formula is: Weight adjustment factor = log2(number of historical failures + 1).

[0067] For example, suppose the system detects success rates of 75%, 70%, 78%, 72%, and 76% for five consecutive operations, respectively, with an average success rate of 74.2%, which is lower than the preset threshold of 80%. At this point, the cumulative number of historical failures is 13. Substituting this into the formula, we obtain the weight adjustment factor = log2(13+1)≈3.81. Therefore, the preset learning rate is increased from 0.01 to 0.01*(1+3.81)=0.0481.

[0068] Through the above technical solution, this application achieves adaptive adjustment of the learning rate. When the system performs poorly after multiple consecutive operations, increasing the learning rate speeds up weight correction, effectively improving the system's responsiveness and adaptability to abnormal situations. Furthermore, using a logarithmic function to calculate the weight adjustment factor can smooth the increase when there are a large number of historical failures, avoiding excessive fluctuations in the learning rate and ensuring system stability.

[0069] The present application further proposes: setting an independent time decay factor λ for each weight coefficient, and automatically restoring the weight coefficient when no correction is triggered after exceeding a preset optimization period.

[0070] The time decay factor λ is used to control the recovery rate of the weight coefficient. The weight coefficients corresponding to different dimensions can be configured with different λ values ​​to match their dynamic characteristics. The preset optimization cycle triggers the automatic recovery mechanism through a preset time threshold. The cycle is set according to the stability requirements of the clinical scenario. The automatic recovery operation realizes the gradual adjustment of the weight coefficient through an exponential function, and the calculation formula is:

[0071] ,in, is the weight coefficient after automatic recovery, The time interval since the last correction.

[0072] This process gradually reduces the impact of historical correction results through an exponential decay model, avoiding the problem of weight solidification caused by long-term non-correction. By independently setting the time decay factor, the weights of different dimensions can be adjusted differently according to their data update frequency. For example, the physiological relevance dimension is configured with a higher In actual application, the preset optimization period can be set to 24 hours. The accuracy of the time interval is ensured by calculating the difference between timestamps. This mechanism effectively prevents model rigidity caused by long-term uncorrected changes while maintaining the dynamic adaptability of the weights, thus ensuring the calculation accuracy of the comprehensive credibility index.

[0073] For example, if the preset optimization period is 30 days, the time decay factor of a certain weight coefficient is If the weight coefficient does not trigger the correction within 45 days, = 45-30 = 15 days. Assuming the corrected weight coefficient is 0.8, the weight coefficient after automatic recovery is calculated as follows:

[0074] Weight coefficient after automatic recovery = ;

[0075] As a result, the weight coefficient was automatically adjusted to 0.78, achieving dynamic recovery of the weight.

[0076] Through the above technical solution, this application implements an automatic recovery mechanism for the weight coefficient. When no correction is triggered for a long time, the weight coefficient will gradually return to its initial state, avoiding the accumulation of weight deviations caused by historical corrections. This mechanism improves the adaptability and stability of the system, allowing the weight coefficient to better reflect the actual situation of the current ICU environment.

[0077] like Figure 3The figure shows a flow chart for calculating the comprehensive credibility index. This application further proposes that the calculation process of the comprehensive credibility index is as follows:

[0078] Normalize the sensor offset in the device layer data to generate the device health index;

[0079] Perform trend correlation analysis on the vital signs change rate in physiological layer data and the equipment alarm parameters to generate a physiological correlation index;

[0080] Perform weighted calculation on the equipment movement trajectory data and pipeline mechanics data in the environmental layer data to generate an environmental interference assessment value;

[0081] Comprehensive credibility index = device health index * weight coefficient of device layer data + physiological correlation index * weight coefficient of physiological layer data + environmental interference assessment value * weight coefficient of environmental layer data.

[0082] The sensor offset is normalized using the range method or Z-score standardization method to eliminate dimensional differences between different devices and make the device health index horizontally comparable. Trend correlation analysis between the rate of change of vital signs and device alarm parameters is achieved using the Pearson correlation coefficient or dynamic time warping algorithm of the data sequence within a sliding window to capture the temporal correlation between physiological parameter fluctuations and device alarm triggering. The weighted calculation of the environmental interference assessment value uses the hierarchical analysis method to determine the primary and secondary weight ratios of the device movement trajectory data and the pipeline mechanical data. The pipeline tension data is collected by strain sensors and then processed by cubic spline interpolation to form a continuous assessment curve.

[0083] Specifically, during the generation of the device health index, the raw offset of the ventilator airway pressure sensor is converted to a normalized value within the interval [0, 1]. An exponential decay correction is triggered when the offset exceeds the tolerance range defined by the device manufacturer. The physiological correlation index is calculated based on the RR interval variability of the ECG monitor. Dynamic trend matching is performed with the ventilator minute ventilation alarm event within a 30-second window. A strong correlation is determined when the absolute value of the correlation coefficient exceeds 0.7. The environmental interference assessment value is obtained by weighted summing the bed displacement accelerometer data and the intravenous infusion line tension data, with the line tension weight set to 0.6 and the displacement acceleration weight set to 0.4. Both are noise-reduced using a Kalman filter before being input into the assessment model. The weight coefficients for the three dimensions are initially set to 0.4 for the device layer, 0.35 for the physiological layer, and 0.25 for the environmental layer. Backpropagation optimization is performed based on the deviation between the dynamic calculation results of the comprehensive credibility index and the actual clinical status.

[0084] By the technical solution, the application realizes multi-dimensional evaluation of the running state of the ICU medical equipment. By fusing the device itself state, patient physiological indicators and environmental factors, a comprehensive credibility index is generated, providing a reliable basis for verifying the authenticity of device alarms. This method overcomes the limitations of traditional single device monitoring, improves the accuracy and credibility of device alarms, and reduces the interference of false alarms on medical staff decision-making. At the same time, the dynamic weight adjustment mechanism of the method enables the system to adapt to different ICU scenarios and patient individual differences, further improving the intelligent level of device management.

[0085] The application further proposes a scheme for verifying the authenticity of multi-device alarm events, including triggering a reverse verification mechanism when the single-device alarm parameter in the device layer data conflicts with the physiological layer data trend, and obtaining device hardware state parameters for secondary judgment.

[0086] Among them, the trigger condition of the reverse verification mechanism is that there is a statistically significant difference between the device alarm parameter and the physiological data trend, for example, the electrocardiogram monitor displays a fibrillation alarm, but the blood oxygen saturation waveform does not correspondingly decrease. At this time, the reverse verification mechanism reads the electrode impedance, signal noise ratio, and power stability parameters through the device self-checking interface, and generates a hardware state evaluation report in combination with the device historical fault record. The hardware state evaluation report is logically matched with the original alarm parameter, if the impedance abnormally rises and the noise ratio exceeds the limit, it is determined that the false alarm is caused by poor device contact; if the hardware parameters are normal and the physiological data trend is continuously abnormal, an upgraded alarm is triggered.

[0087] For example: when the ventilator pressure alarm parameter shows airway high pressure, but the patient's chest movement waveform and blood oxygen saturation are not synchronously abnormal, the system automatically retrieves the ventilator pipeline pressure sensor calibration data, flow sensor offset, and near-end carbon dioxide waveform data. If the calibration data deviates from the reference value by more than 5% and the flow sensor offset continuously increases, it is determined that the sensor drift causes a false alarm, and the alarm is suspended and calibration instructions are sent; if the calibration data is normal and the near-end carbon dioxide waveform disappears, it is determined that it is a real airway obstruction event, and the emergency treatment process is activated. By introducing secondary judgment of device hardware state parameters, device itself faults and real clinical events are effectively distinguished, reducing interference alarms caused by hardware abnormalities.

[0088] Through the above technical solution, the application can effectively improve the accuracy of ICU device alarms. By introducing a reverse verification mechanism, single-device alarms are cross-verified with patient physiological data and device hardware state in multiple dimensions, greatly reducing the false alarm rate. This not only reduces the workload of medical staff, but also improves the timely response capability to patient critical conditions, thereby significantly improving the medical quality and patient safety of the ICU.

[0089] This application further proposes that the weights of evaluation indicators are allocated according to the ICU scenario, and a scenario-based comprehensive evaluation indicator system is constructed. When the scenario switches, the weights of the evaluation indicators of the corresponding scenario are automatically called to calculate the comprehensive deviation and correct the weight coefficient.

[0090] The scenario-based comprehensive evaluation index system includes multiple predefined ICU scenario classifications and corresponding weighting schemes. Each scenario classification corresponds to a set of evaluation index weights. Scenario switching is triggered by monitoring the device's operating mode, patient medical record tags, or medical order information. The built-in weight database is automatically called to store the preset weight ratios for physiological recovery efficiency indicators, equipment coordination efficiency indicators, operational convenience indicators, and resource utilization efficiency indicators in different scenarios.

[0091] For example, we first define multiple typical ICU scenarios, such as cardiac monitoring, respiratory monitoring, neurological monitoring, etc. For each scenario, we preset the weights of different evaluation indicators.

[0092] When the system recognizes that the current scenario has switched to postoperative monitoring, it automatically extracts the weights of the four evaluation indicators corresponding to this scenario from the weight database. The weight of the physiological recovery efficiency indicator is set to 0.4, the device collaboration efficiency indicator is set to 0.3, the operational convenience indicator is set to 0.2, and the resource utilization efficiency indicator is set to 0.1. This weight distribution reflects the higher requirements for the recovery of patients' vital signs in the postoperative period. A weighted sum is performed based on the adjusted weights to produce an evaluation result that matches the scenario.

[0093] When the scenario switches to infection control, the system automatically adjusts the weight of the resource utilization efficiency indicator to 0.35 and reduces the weight of the equipment collaboration efficiency indicator to 0.25, prioritizing the optimization of disinfection equipment resource scheduling efficiency during the weight coefficient correction process. The system continuously monitors changes in ventilator usage patterns. When high-frequency oscillation ventilation mode is activated, it determines that the acute respiratory distress syndrome treatment scenario has been entered, triggering the loading of the corresponding weight configuration.

[0094] Through the above technical solutions, this application achieves scenario-based adaptation of ICU equipment management, improving the pertinence and accuracy of comprehensive evaluation. Key indicators in different scenarios are highlighted, making device collaborative control more in line with actual clinical needs. At the same time, automated scenario recognition and weight adjustment reduce the operational burden on medical staff and improve the system's usability.

[0095] This application further generates a dynamic priority sequence based on the verification results, including:

[0096] It is proposed to set basic priority weights according to the device type for verified real alarm events, and when there is a causal relationship between multiple device alarms, the priority of the upstream device alarm is increased to 1.5 times that of the downstream device alarm.

[0097] The basic priority weight is preset based on the clinical importance of the device type. The basic weight of a ventilator is higher than that of an infusion pump, and the basic weight of an ECG monitoring device is higher than that of a temperature monitoring device. Causality is determined through time series analysis of device alarm parameters and directional verification of physiological parameter changes. When a ventilator abnormality triggers a drop in blood oxygen, the ventilator is identified as an upstream device. The priority escalation mechanism uses a linear weighting method: upstream device priority = basic weight × 1.5 + downstream device priority × 0.5. The processing order is formed by superimposing the weights.

[0098] For example, when a ventilator alarm parameter indicates abnormal airway pressure and a blood oxygen monitoring device alarm indicates a decrease in blood oxygen saturation, the system first verifies the authenticity of both alarms. After confirming that the abnormal ventilator pressure occurs earlier than the decrease in blood oxygen saturation and that the rate of change in blood oxygen saturation is negatively correlated with the abnormal pressure parameter, the system determines that the ventilator is the upstream alarm source. The ventilator's basic priority weight is preset to level 8, and the blood oxygen monitoring device's basic weight is level 6. After the execution priority is increased, the ventilator's priority is adjusted to 8 × 1.5 = level 12, while the blood oxygen monitoring device remains at level 6. Based on this, the system generates a processing sequence, prioritizing the execution of ventilator abnormality handling instructions over blood oxygen monitoring alarms. This mechanism ensures that the root cause of the chain reaction is handled first. By controlling the ventilator parameters to restore airway pressure, blood oxygen data returns to the normal range within 35 seconds, avoiding the redundant consumption of resources caused by separately handling downstream alarms.

[0099] Through the above technical solution, this application effectively solves the technical problem of misjudging the priority of causally related events in multi-device alarm scenarios. By establishing a priority adjustment mechanism based on the causal chain of device operation, it avoids the situation where the priority processing of downstream device alarms causes delays in handling upstream device failures. This solution can ensure that the execution order of device control instructions is consistent with the development logic of clinical pathology, thereby shortening the negative impact time of device failures on patients' physiological status and improving the efficiency of handling critical events in the ICU.

[0100] This application further proposes real-time monitoring of the pre-processed multi-dimensional feedback data. When a data point deviates from the historical data distribution by more than a preset abnormality threshold, it is marked as abnormal data; the cause of the abnormal data is analyzed. If it is determined that the abnormality is caused by equipment failure or human error, the weight coefficient correction is suspended and a corresponding warning message is issued.

[0101] In the real-time monitoring, the historical data mean and standard deviation are calculated by using a sliding time window, and the preset abnormal threshold is set as a range deviating from the mean by ±2 standard deviations; the system has a built-in device fault code library and an operation specification database, and the cause classification is realized by comparing the abnormal data pattern with the typical fault mode in the database; during the weight coefficient correction suspension period, the system maintains the current three-dimensional correlation matrix parameters unchanged, and simultaneously outputs warning information through a sound-light alarm.

[0102] For example, when the physiological data response curve appears a mutation, the abnormal mark is triggered after detecting that it deviates from the historical fluctuation range. By calling the device operation log, it is found that the ventilator pressure sensor has 3 consecutive self-check failure records, and it is determined that the device hardware fails. The system automatically freezes the weight coefficient correction process to avoid the influence of the matrix parameters by the false data, and controls the bed head warning light to flash a red signal. During this period, the three-dimensional correlation matrix parameters before freezing are continued to be used to maintain the operation until the maintenance personnel manually restore the correction function after completing the device repair.

[0103] As a preferred embodiment, the scheme of the application is implemented as follows: in the scenario where the ventilator and the electrocardiograph monitor work cooperatively, multi-dimensional feedback data after data cleaning and standardization processing are received in real time. The statistical deviation of the current data point from the historical data distribution is calculated by using a sliding time window method, and an abnormality detection model based on the 3σ principle is adopted. When the standard deviation of the oxygen saturation recovery rate exceeds 2.5 times of the historical mean, the abnormal data mark mechanism is triggered. By comparing the device self-check log with the operation record, it is found that there are ventilator pipeline air leakage alarms and emergency intubation operations of nursing personnel in the period corresponding to the abnormal data point. According to the dual factor determination results of the device fault and the manual operation, the weight coefficient correction process of the three-dimensional correlation matrix is automatically frozen, and warning information including the abnormal timestamp, data characteristics and preliminary diagnosis conclusion is pushed to the central monitoring station.

[0104] Through the above technical scheme, the application effectively solves the problem of false weight correction caused by sudden device failure or operation error in the device cooperative control process. By establishing the data abnormality identification and correction suspension mechanism, the pollution of the system decision model by the false feedback data is avoided, and the reliability of the three-dimensional correlation matrix weight adjustment is ensured. The association analysis function of the abnormal data and the operation event provides a clear technical path for quickly locating the system abnormal source, thereby improving the anti-interference ability and operation safety of the ICU device cooperative control system.

[0105] As shown in Figure 4 The application further proposes an ICU medical instrument device digital intelligent management and control system, which includes a multi-source data acquisition module, a three-dimensional data processing and analysis module, a dynamic priority control and execution module, a feedback data acquisition module, an evaluation index calculation module, a comprehensive deviation calculation module, a feedback correction module and a correction execution module.

[0106] The multi-source data acquisition module is configured to synchronously acquire ventilator operating parameters, ECG monitoring waveform data, blood oxygen saturation time series data, device displacement sensor signals and pipeline traction mechanical data through wired and wireless transmission protocols.

[0107] The three-dimensional data processing and analysis module has a built-in correlation matrix generation algorithm, which maps device layer data with device health index, associates physiological layer data with trend correlation analysis results, and binds environmental layer data with weighted interference assessment values.

[0108] The dynamic priority control and execution module integrates an event verification algorithm. When the alarm parameters of a single device conflict with the physiological trend, the reverse verification mechanism is activated to retrieve the device self-test log for secondary judgment.

[0109] The feedback data acquisition module deploys a timestamp synchronization mechanism to accurately record the equipment status recovery time and the interval between medical staff operations. Resource consumption data is collected in real time through IoT metering devices.

[0110] The evaluation index calculation module has a built-in standardized conversion unit to convert the physiological data response curve into a quantitative value of recovery efficiency. The equipment collaborative efficiency index is generated by calculating the event processing time difference.

[0111] The comprehensive deviation calculation module is equipped with a scene recognition unit, which automatically switches the evaluation weight distribution scheme according to the current treatment stage of the ICU.

[0112] The feedback correction module embeds the gradient descent algorithm to generate weight adjustment according to the comprehensive deviation; the correction execution module triggers system recalibration after starting the matrix update process.

[0113] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0114] The multi-source data acquisition module is configured to simultaneously acquire the ventilator tidal volume parameters, the ST segment waveform data of the electrocardiogram monitor, and the infusion pump flow rate monitoring value as device layer data, synchronously acquire the patient's arterial blood pressure waveform and central venous pressure trend line as physiological layer data, and acquire the ventilator tube tension value and the infusion tube droop angle as environmental layer data through the pressure sensor array.

[0115] The three-dimensional data processing and analysis module maps the ventilator operating parameters to the device status dimension, maps the phase difference between the arterial blood pressure and the ventilator pressure waveform to the physiological correlation dimension, converts the pipeline tension value into an interference assessment value through a finite element mechanics model and maps it to the environmental interference dimension, and uses a gradient descent algorithm to dynamically adjust the weight coefficient of each dimension.

[0116] When the ventilator high-pressure alarm and the arterial blood pressure drop event occur simultaneously, the dynamic priority control and execution module identifies the real critical event by calculating the temporal correlation between the two and generates a coordinated instruction sequence including ventilator parameter adjustment and pressor drug infusion.

[0117] The feedback data acquisition module records the rising slope of the patient's blood pressure recovery curve after executing the command, the stabilization time of the ventilator parameters, the number of times the nurse touches the ventilator interface, and the consumption of normal saline.

[0118] The evaluation index calculation module defines the product of the blood pressure recovery slope and the device stabilization time as the physiological recovery efficiency index, and the ratio of the number of touches to the device stabilization time as the operation convenience index.

[0119] The comprehensive deviation calculation module compares the current operation convenience index with the preset value. When the deviation exceeds the threshold, the feedback correction module is triggered to adjust the weight distribution of the environmental interference dimension in the three-dimensional matrix through the back propagation algorithm.

[0120] The correction execution module recalculates the collaborative control priority of the ventilator and infusion pump after the weights are updated, forming a closed-loop optimization mechanism.

[0121] Through the above technical solution, this application realizes the fusion processing of multi-source heterogeneous data, effectively identifies the essential differences between clinical interference factors and real critical events, and improves the accuracy of alarm verification through a dynamic weight adjustment mechanism. The closed-loop feedback system continuously optimizes the equipment collaborative control strategy, reduces invalid operations caused by misjudgment of single-dimensional data, and improves the efficiency of medical resource utilization. The synergistic effect of multi-dimensional evaluation indicators ensures that the system maintains stable decision-making performance in complex clinical scenarios and avoids false alarm interference in traditional single-machine monitoring mode.

[0122] Furthermore, in order to optimize the use management of medical instruments and equipment, the following solutions are proposed.

[0123] Among them, in order to facilitate management, the positioning function of medical instruments and equipment is added; Zigbee communication module tags are set on the medical instruments and equipment; signal anchor points are deployed in the ICU ward based on multi-node Zigbee positioning technology, and then a centimeter-level positioning network is built; the real-time location coordinates of the equipment are uploaded, and finally the spatial distribution location of the medical instruments and equipment is dynamically displayed on the electronic map.

[0124] Among them, in order to facilitate the use and maintenance of medical instruments and equipment, a cloud-based equipment archive management system is constructed, which includes a basic information library, an operation knowledge base, and a use record; a two-dimensional code is pasted on the equipment, and after scanning the code with a mobile phone, the cloud-based equipment archive management system can be entered; then the basic information library can be viewed, and the basic information library stores the model, number, body picture, service life, and maintenance record data; the operation knowledge base can also be viewed, and the operation knowledge base includes a standardized operation process manual, parameter setting, and alarm processing method; the maintenance record is used to record the use time, operator, and cumulative use time, forming an unalterable electronic log.

[0125] The intelligent early warning monitoring system also includes a maintenance manual corresponding to the medical instrument and equipment, which records the service life and cumulative time of the medical instrument and equipment corresponding to the maintenance strategy method; the service life and cumulative time of the medical instrument and equipment are obtained, and when the service life and cumulative time reach the service life and cumulative time required for maintenance specified in the maintenance manual, the instrument maintenance person in charge is reminded through a short message.

[0126] The technical scope of the present application is not limited to the content in the above description, and those skilled in the art can make various modifications and changes to the above embodiments without departing from the technical idea of the present application, and these modifications and changes should all belong to the protection scope of the present application.

Claims

1. A digital intelligent management method for ICU medical instruments and equipment, characterized by: The steps include: Real-time collection of device layer data, physiological layer data and environmental layer data; Constructing a three-dimensional correlation matrix, mapping the device layer data, physiological layer data, and environmental layer data to the device status dimension, physiological relevance dimension, and environmental interference dimension, respectively, calculating the dynamic coupling weights between the dimensions, and generating a comprehensive credibility index; Based on the comprehensive credibility index, the authenticity of multiple device alarm events is verified, a dynamic priority sequence is generated according to the verification results, and device collaborative control instructions are output for events that meet the preset critical threshold; Obtaining multi-dimensional feedback data after the execution of the device collaborative control instruction, the multi-dimensional feedback data including physiological data response curve, device status recovery time, medical staff operation time and resource consumption data; Preprocessing the multi-dimensional feedback data, and calculating a physiological recovery efficiency index, an equipment coordination efficiency index, an operation convenience index, and a resource utilization efficiency index based on the preprocessed multi-dimensional feedback data; Compare each evaluation indicator with the preset expected indicator value to obtain the deviation of each evaluation indicator, assign weights to each evaluation indicator according to the preset importance rules, and calculate the comprehensive deviation; Determine whether the comprehensive deviation exceeds a preset adjustment threshold, and if so, reversely correct the weight coefficient of the three-dimensional correlation matrix according to a predefined optimization strategy; Based on the weight coefficient of the revised three-dimensional correlation matrix, the comprehensive credibility index is recalculated, and the authenticity of multiple device alarm events is verified.

2. The digital intelligent management method for ICU medical instruments and equipment according to claim 1, characterized in that: The correction formula of the weight coefficient of the three-dimensional correlation matrix is: ,in, is the weight coefficient before correction, is the modified weight coefficient, is the preset learning rate.

3. The digital intelligent management method for ICU medical instruments and equipment according to claim 2, characterized in that: Also includes: Obtain historical operation success rate data, and build a feedback weight adjustment factor based on the historical operation success rate data. When the success rate is lower than the preset threshold in N consecutive operations, the preset learning rate Increase to (1 + weight adjustment factor) times the original value, where the weight adjustment factor is calculated based on the logarithmic function of the number of historical failures.

4. The digital intelligent management method for ICU medical instruments and equipment according to claim 2, characterized in that: Also includes: Set independent time decay factors for each weight coefficient When the preset optimization period is exceeded and no correction is triggered, the weight coefficient is automatically restored, specifically: ； in, is the weight coefficient after automatic recovery, The time interval since the last correction.

5. The digital intelligent management method for ICU medical instruments and equipment according to claim 1, characterized in that: The calculation process of the comprehensive credibility index is: Normalize the sensor offset in the device layer data to generate the device health index; Perform trend correlation analysis on the vital signs change rate in physiological layer data and the equipment alarm parameters to generate a physiological correlation index; Perform weighted calculation on the equipment movement trajectory data and pipeline mechanics data in the environmental layer data to generate an environmental interference assessment value; Comprehensive credibility index = device health index * weight coefficient of device layer data + physiological correlation index * weight coefficient of physiological layer data + environmental interference assessment value * weight coefficient of environmental layer data.

6. The digital intelligent management method for ICU medical instruments and equipment according to claim 1, characterized in that: The authenticity verification of multiple device alarm events includes: When the single device alarm parameters in the device layer data conflict with the physiological layer data trend, the reverse verification mechanism is triggered to obtain the device hardware status parameters for secondary judgment.

7. The digital intelligent management method for ICU medical instruments and equipment according to claim 1, characterized in that: The weights of the evaluation indicators are allocated according to the ICU scenario, and a scenario-based comprehensive evaluation indicator system is constructed. When the scenario switches, the weights of the evaluation indicators of the corresponding scenario are automatically called to calculate the comprehensive deviation and correct the weight coefficient.

8. The digital intelligent management method for ICU medical instruments and equipment according to claim 1, characterized in that: Generating a dynamic priority sequence according to the verification result includes: For verified real alarm events, the basic priority weight is set according to the device type. When multiple device alarms are causally related, the upstream device alarm priority is increased to 1.5 times that of the downstream device alarm.

9. The digital intelligent management method for ICU medical instruments and equipment according to claim 1, characterized in that: Also includes: Real-time monitoring of pre-processed multi-dimensional feedback data. When a data point deviates from the historical data distribution by more than the preset abnormal threshold, it is marked as abnormal data. Analyze the causes of abnormal data. If it is determined that the abnormality is caused by equipment failure or human error, suspend the weight coefficient correction and issue a corresponding warning message.

10. ICU medical equipment digital intelligent management and control system, characterized by: The method for digital intelligent management of ICU medical instruments and equipment according to any one of claims 1 to 9 is applied, comprising: Multi-source data acquisition module, used to collect device layer data, physiological layer data and environmental layer data in real time; A three-dimensional data processing and analysis module is used to construct a three-dimensional correlation matrix, map the device layer data, physiological layer data, and environmental layer data to the device status dimension, physiological relevance dimension, and environmental interference dimension, calculate the dynamic coupling weights between the dimensions, and generate a comprehensive credibility index; A dynamic priority control and execution module, configured to verify the authenticity of multiple device alarm events based on the comprehensive credibility index, generate a dynamic priority sequence based on the verification results, and output device collaborative control instructions for events that meet a preset critical threshold; A feedback data acquisition module is used to obtain multi-dimensional feedback data after the device collaborative control instruction is executed. The multi-dimensional feedback data includes physiological data response curves, device status recovery time, medical staff operation time and resource consumption data; An evaluation index calculation module is used to preprocess the multi-dimensional feedback data and calculate the physiological recovery efficiency index, the equipment coordination efficiency index, the operation convenience index and the resource utilization efficiency index based on the preprocessed multi-dimensional feedback data; The comprehensive deviation calculation module is used to compare each evaluation indicator with the preset expected indicator value to obtain the deviation of each evaluation indicator, assign weights to each evaluation indicator according to the preset importance rules, and calculate the comprehensive deviation; A feedback correction module is used to determine whether the comprehensive deviation exceeds a preset adjustment threshold, and if so, reversely correct the weight coefficient of the three-dimensional correlation matrix according to a predefined optimization strategy; The correction execution module is used to recalculate the comprehensive credibility index based on the weight coefficient of the corrected three-dimensional correlation matrix and verify the authenticity of multiple device alarm events.

Images