Medical equipment monitoring analysis system and method based on full life cycle

Abstract

The invention discloses a full-life-cycle-based medical equipment monitoring analysis system and method, and relates to the technical field of medical equipment monitoring, and the method comprises the steps: collecting medical equipment data in real time, and dynamically constructing a full-life-cycle digital twin model of medical equipment; constructing a medical equipment knowledge graph based on the equipment type, the function association and the spatial distribution; when the medical equipment node detects abnormal data, an early warning signal is sent to a full-life-cycle digital twin model associated with the medical equipment in the medical equipment knowledge graph in combination with the medical equipment knowledge graph; preliminarily judging fault causes and fault location, and generating an analysis report; performing multi-dimensional verification on the diagnosis result in the digital twin environment, comprehensively evaluating the risk coefficient of the scheme, and outputting an optimal maintenance strategy; in the maintenance process, the maintenance process is recorded in real time, and maintenance data and equipment state updating are synchronously fed back to the equipment full-life-cycle digital twin model.

Description

Technical Field

[0001] The present invention relates to the field of medical equipment monitoring technology, and in particular to a medical equipment monitoring and analysis system and method based on the entire life cycle. Background Art

[0002] With the rapid development of medical technology, the types and complexity of medical equipment are constantly increasing, and the drawbacks of traditional equipment management models are becoming increasingly prominent. Existing monitoring methods mostly rely on static thresholds to judge equipment anomalies, which makes it difficult to adapt to dynamic changes during equipment operation. In addition, there is a lack of systematic integration and analysis of data throughout the equipment life cycle, making it impossible to effectively predict potential failures. The correlation between devices has not been fully explored. When a device fails, it is difficult to quickly assess the scope of its impact and the propagation path on other devices, resulting in inefficient fault response. In terms of maintenance decisions, they are mostly based on experience and judgment, lack scientific verification and cost-benefit analysis, and can easily lead to waste of resources or excessive equipment maintenance. In addition, the problem of equipment aging is often ignored, and a dynamic management system covering the entire life cycle of the equipment has not been formed, making it difficult to meet the needs of modern medicine for efficient, safe, and precise management of equipment. Summary of the Invention

[0003] The purpose of the present invention is to provide a medical equipment monitoring and analysis system and method based on the entire life cycle to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a medical device monitoring and analysis method based on the entire life cycle, the method comprising:

[0005] S100: Collect medical equipment operating data in real time and collect data on the entire life cycle of medical equipment. Based on the collected data, dynamically build a digital twin model of the entire life cycle of medical equipment, and combine the operating data of each medical device with its corresponding digital twin model of the entire life cycle;

[0006] S200, connect each medical device to the swarm intelligence network, making it a node in the network; construct a medical device knowledge graph based on device type, functional association, and spatial distribution, and establish an association network between medical devices;

[0007] S300: When a medical device node detects abnormal data, it sends a warning signal to the full life cycle digital twin model of the medical device associated in the medical device knowledge graph in combination with the medical device knowledge graph. The full life cycle digital twin model that receives the warning signal activates a self-check program.

[0008] S400: By comparing historical cases, evaluating related impacts, and simulating fault propagation paths, we can preliminarily determine the cause and location of the fault and generate an analysis report.

[0009] S500: Perform multi-dimensional verification of the diagnosis results in the digital twin environment. The multi-dimensional verification includes reproducing the fault scenario, simulating the maintenance plan, and deducing the aging trend. The risk factor of the plan is comprehensively evaluated to output the optimal maintenance strategy.

[0010] S600: During the maintenance process, the maintenance process is recorded in real time, and the maintenance data and equipment status updates are synchronously fed back to the digital twin model of the equipment's entire life cycle.

[0011] According to the above solution, the medical equipment operation data includes various medical operation parameters, environmental monitoring data and clinical usage data; the full life cycle data includes the procurement stage, installation stage, maintenance stage and business stage. The procurement stage includes the equipment model, manufacturer, warranty period and energy consumption level; the installation stage includes the installation location, initialization commissioning report and acceptance test record; the maintenance stage includes the fault type, fault description, maintenance time and maintenance strategy; the business data includes the frequency of use of medical equipment, usage time and usage feedback;

[0012] The collected data is denoised, normalized, and time-series aligned, and invalid data points are filtered out.

[0013] According to the above solution, step S100 includes:

[0014] S110. Dynamically construct a full-lifecycle digital twin model of the medical device, the full-lifecycle digital twin model comprising a geometric model and behavioral rules; the geometric model constructs a three-dimensional visual digital twin model based on the device's bill of materials (BOM) and CAD drawings; the behavioral rules define the medical device's operating mode transition logic through a state machine and embed the communication rules of the medical device control protocol. By parsing the device's original communication protocol, the device state switching instructions and parameter read and write rules are extracted and converted into transition conditions for the state machine;

[0015] S120. Use the collected full life cycle data of the medical device as the initial attributes of the full life cycle digital twin model of the medical device; synchronously update the collected medical device operation data and the corresponding parameters of the full life cycle digital twin model of the medical device; convert the data of the maintenance phase in the full life cycle data into a health status assessment factor to drive the adaptive adjustment of the full life cycle digital twin model.

[0016] According to the above solution, step S200 includes:

[0017] S210: Connect medical devices to the swarm intelligence network, assign a node to each device, and register basic information about the medical devices, including device type, unique identifier, communication protocol, and physical location. Computing resources and storage quotas are allocated to each node for local data processing. When a device enters a high-load state, cluster resources are dynamically scheduled through the edge controller to temporarily increase node computing power quotas to ensure real-time data processing.

[0018] S220, extracting key entities from the collected data, wherein the key entities include equipment entities, function entities, and space entities;

[0019] S230: Establish association relationships, including functional dependency relationships, physical connection relationships, and process linkage relationships, to construct a network structure of the medical device knowledge graph;

[0020] S240. Update the association weights in real time based on the frequency of data interaction between medical devices, and identify sets of medical devices with tight functional coupling or high-frequency interaction based on a graph clustering algorithm; automatically update the graph structure through an incremental learning algorithm when devices are added, retired, or upgraded.

[0021] According to the above solution, step S300 includes:

[0022] S310: The medical device node compares the collected operating data with the dynamic threshold in real time, and triggers anomaly detection when the data deviates from the normal range; calculates the deviation degree, duration, and impact range of the abnormal data, and classifies the anomaly into primary anomaly, intermediate anomaly, and advanced anomaly;

[0023] The degree of deviation of the abnormal data is expressed as follows:

[0024]

[0025] Where D represents the degree of deviation; x t It represents the specific data value collected at time t; μ represents the dynamic threshold; σ represents the dynamic standard deviation, which is used to measure the degree of dispersion of the data. The dynamic standard deviation changes as the data is updated.

[0026] The impact range is as follows:

[0027]

[0028] Where I represents the impact range index; w i Expressed as the weight of the i-th associated device; L i It represents the current load of the i-th device, which indicates the workload of the device at the current moment; i represents the index of the associated device; It is expressed as the average load; n is the total number of associated devices;

[0029] Map the impact range index to the impact range interval: low impact: I∈[0,0.3), impact range [0%,30%]; medium impact: I∈[0.3,0.7), impact range [30%,70%]; high impact: I∈[0.7,1], impact range [70%,100%];

[0030] The abnormality is graded as follows:

[0031] G=f(d,t,i)=α×D+β×T+γ×I;

[0032] Where G represents the anomaly score; D represents the degree of deviation; T represents the duration level; I represents the impact range index; α represents the weight coefficient of the degree of deviation; β represents the weight coefficient of the duration level; γ represents the weight coefficient of the impact range; the sum of the weight coefficients is 1;

[0033] The weight coefficient is based on the preset initial weight of the equipment type, and is dynamically adjusted based on the quarterly statistical analysis of the prediction accuracy of each dimension for actual failures. The formula is as follows:

[0034]

[0035] Among them, α′ represents the adjusted weight coefficient; A D Expressed as the historical accuracy of the deviation dimension; A arg Expressed as average accuracy;

[0036] S320: Starting from the abnormal device node, analyzing the associated medical devices based on the functional dependency, physical connection, emergency linkage, and association weight in the medical device knowledge graph;

[0037] S330, sending the warning signal to the associated medical device, wherein the warning signal includes a unique identifier of the abnormal device, abnormal parameters, a timestamp, and an abnormal level;

[0038] S340. The full life cycle digital twin model that receives the early warning signal activates the self-check program, collects the current operating parameters of the medical equipment, and checks the parameters related to the abnormal data.

[0039] According to the above solution, the dynamic threshold includes:

[0040] S311. Set basic threshold ranges and safety mandatory thresholds for each parameter based on medical device design parameters, historical operating benchmark data, and industry standards for similar devices;

[0041] S312. Extracting adjustment factors based on the full lifecycle data of the medical device, the adjustment factors including the age, frequency of use, maintenance history, and environmental data of the medical device; assigning dynamic weights to the adjustment factors, the dynamic weights being dynamically adjusted based on the device type and parameter characteristics;

[0042] 313. Utilize reinforcement learning algorithms, with the reduction of false alarm and missed alarm rates in anomaly detection as the reward function, the deviation between real-time operating parameters and current thresholds as the state space, and the threshold adjustment range as the action space. The action space definition includes the threshold increase or decrease range and the warning level adjustment. Based on real-time data and historical adjustment records, the basic threshold range is automatically optimized to form a dynamic threshold. Combined with historical fault data and real-time data feedback, the dynamic threshold model is updated every hour to ensure that the threshold adapts to equipment aging and environmental changes.

[0043] According to the above solution, step S400 includes:

[0044] S410: Compare the abnormal parameters and time series data of the abnormal equipment with the similar fault data in the historical data, analyze the matching degree between the current abnormality and the historical data in terms of fault manifestation, occurrence environment and related equipment status, and extract the corresponding fault cause, maintenance plan and maintenance result;

[0045] S420: Based on the functional dependencies, physical connections, and process linkages in the medical device knowledge graph, evaluate the potential impact of abnormal device failures on associated devices, calculate association weights, identify associated medical devices with association weights above a threshold, and analyze the chain reaction caused by the failure in combination with the current operating status of the associated medical devices;

[0046] S430. Utilizing the behavioral rules of the full lifecycle digital twin model and combining it with the medical device knowledge graph, dynamically simulate the fault propagation path. The dynamic simulation starts from the abnormal medical device node and, based on the connection relationship and operation logic between medical devices, deduces the propagation trajectory of the fault in the device network to assist in locating the root cause of the fault.

[0047] S440, based on historical case matching results, correlation impact assessment, and fault propagation path simulation, uses an evidence fusion algorithm to assign weights to candidate fault causes, screens out the most likely fault causes and fault locations, and generates an analysis report. The analysis report includes an overview of the abnormal event, preliminary diagnostic conclusions, fault impact range, abnormality level, historical case references, and recommended maintenance strategies. The report is then simultaneously pushed to the device management platform, the operation and maintenance personnel terminal, and the digital twin model of the abnormal device.

[0048] According to the above solution, step S500 includes:

[0049] S510. Based on the abnormal event overview, time series data, and fault propagation path in the analysis report, reproduce the equipment operating status before the fault occurs in the digital twin environment, synchronously load the environmental data and associated equipment operating parameters, and build a complete fault trigger condition.

[0050] S520. Using the control variable method, adjust the parameters in the recurrence scenario one by one to observe whether the fault phenomenon reappears or changes, and verify the sensitivity of the fault cause in the preliminary diagnosis conclusion; if the fault phenomenon disappears or is significantly alleviated after the parameter adjustment, mark the parameter as a possible fault cause; otherwise, exclude the parameter;

[0051] S530. Execute the recommended maintenance strategy in the analysis report in the digital twin model, simulate the impact of the maintenance strategy on the overall performance of the equipment, and generate quantitative evaluation results by calculating the performance recovery rate, maintenance time, and resource consumption to evaluate the effectiveness of the maintenance strategy.

[0052] The performance recovery rate is obtained by combining the performance index value at the time of failure and the performance index value after maintenance, and the formula is as follows:

[0053]

[0054] Where η is the performance recovery rate; P post Expressed as the performance index value after maintenance; P fall Expressed as the performance index value at the time of failure; P std Expressed as standard performance index values;

[0055] The maintenance time is recorded from the start to the end of the maintenance operation;

[0056] The resource consumption is analyzed according to the recommended maintenance strategy to obtain the maintenance efficiency. The formula is as follows:

[0057]

[0058] Where E represents maintenance efficiency; T a It is expressed as maintenance time; C is the resource consumption coefficient, which is used to measure the resources consumed during the maintenance process. The more resources are consumed, the greater the resource consumption coefficient; η is the performance recovery rate;

[0059] The resource consumption coefficient is as follows:

[0060] C = 0.4 × C time +0.3×C cost +0.2×C human +0.1×C energy ;

[0061] Where C represents the resource consumption coefficient; Ctime Expressed as time consumption, time consumption is the ratio of actual time consumption to standard time consumption; C cost Expressed as cost consumption, cost consumption is the ratio of the cost of accessories to the original value of the equipment; C human Expressed as manpower consumption, the required manpower qualifications are divided according to the abnormality level; C energy Expressed as energy consumption, energy consumption is the ratio of actual energy consumption to standard energy consumption;

[0062] S540: Build a device health model based on maintenance records, failure frequencies, and performance degradation trends in the full lifecycle data to quantify the current degree of device aging.

[0063] The current aging degree of the quantitative equipment is as follows:

[0064]

[0065] Where H(t) represents the health status of the device at time t; H0 represents the initial health status; λ(ι) represents the aging rate, the aging rate of the device at time ι; t represents the time index;

[0066] The aging rate is expressed as follows:

[0067]

[0068] Where λ(t) represents the aging rate of the device at time t; w k Expressed as the weight of the kth aging factor, f k (t) represents the cumulative value of the kth aging factor at time t; k represents the index of the aging factor; m represents the total number of aging factors;

[0069] The cumulative value f of the kth aging factor at time t k (t) = b k ×t;b k Expressed as the basic equipment aging rate, the basic equipment aging rate is obtained based on the historical aging situation of similar equipment;

[0070] S550. Based on fault scenario reproduction, maintenance plan simulation, and aging trend deduction, risk assessment indicators are extracted, a decision tree model is constructed, and the optimal maintenance strategy is output. The maintenance strategy includes continued maintenance and recommended scrapping. The continued maintenance includes preventive maintenance, corrective maintenance, and upgrade and modification. The recommended scrapping includes emergency scrapping and planned scrapping.

[0071] According to the above scheme, the risk assessment indicators include technical indicators, time indicators, economic indicators, compliance indicators and aging indicators; the technical indicators include the complexity and success rate of maintenance operations and the impact on equipment accuracy; the time indicators include maintenance time, downtime and service recovery time; the economic indicators include maintenance costs and potential losses; the compliance indicators include medical device regulatory requirements; the aging indicators include the impact of maintenance strategies on the remaining life of the equipment;

[0072] The extraction of the risk assessment indicators relies on multi-module data fusion, including real-time equipment operation data, historical maintenance cases, industry standards and digital twin simulation results.

[0073] A medical equipment monitoring and analysis system based on the entire life cycle, which includes: data acquisition module, digital twin module, knowledge graph module, intelligent diagnosis module, maintenance strategy module and database module;

[0074] The data acquisition module includes an operating data module, a full life cycle data module, and a data preprocessing module; the operating data module acquires medical equipment operating parameters, environmental data, and clinical usage data in real time; the full life cycle data module integrates procurement, installation, maintenance, and business data of medical equipment; and the data preprocessing module is used to perform noise reduction, normalization, and time series alignment on the collected data.

[0075] The digital twin module includes a modeling module, a behavior rule module, and a dynamic update module; the modeling module builds a digital twin model of the medical device based on the BOM list or CAD drawing of the medical device; the behavior rule module implements state machine control logic and device protocol parsing; the dynamic update module synchronizes the real-time operating data of the medical device with the parameters of the digital twin model;

[0076] The knowledge graph module includes a node management module, a graph construction module, and a graph optimization module. The node management module assigns a unique identifier, computing resources, and communication links to each medical device; the graph construction module extracts key entities and builds association relationships; and the graph optimization module identifies device groups through a graph clustering algorithm and incrementally updates the graph.

[0077] The intelligent diagnosis module includes a dynamic threshold module, a multi-level warning module, and a fault location module; the dynamic threshold module dynamically adjusts the threshold based on reinforcement learning; the multi-level warning module sends warning signals according to the abnormality classification standard; and the fault location module performs root cause analysis by combining fault propagation simulation and historical case matching.

[0078] The maintenance strategy module includes a virtual verification module, an aging assessment module, and a decision tree module; the virtual verification module reproduces faults in a digital twin environment and simulates maintenance strategies; the aging assessment module predicts the remaining life of medical equipment; and the decision tree module outputs the optimal maintenance strategy.

[0079] The database module includes a maintenance tracking module, a knowledge base module and a performance feedback module; the maintenance tracking module is used to record maintenance operation steps, spare parts batches and quality inspection results; the knowledge base module adds new failure modes to the case library and optimizes maintenance strategies; the performance feedback module is used to record various evaluation indicators of medical equipment.

[0080] Compared with the prior art, the present invention has the following beneficial effects:

[0081] 1. By collecting medical equipment operation data and full life cycle data, building a digital twin model and knowledge graph, we can achieve digital management of the entire process from procurement to scrapping of equipment, and complete visual monitoring throughout the process;

[0082] 2. This application combines dynamic threshold setting, knowledge graph association analysis, and digital twin simulation to quickly detect equipment anomalies and accurately locate the root cause of the fault;

[0083] 3. This application verifies the maintenance plan through digital twin environment simulation, and further optimizes the maintenance strategy to reduce ineffective maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 This is a flowchart of the steps of the medical equipment monitoring and analysis method based on the entire life cycle of the present invention;

[0085] Figure 2 This is a schematic diagram of the structure of the medical equipment monitoring and analysis system based on the entire life cycle of the present invention. DETAILED DESCRIPTION

[0086] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0087] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, a medical equipment monitoring and analysis method based on the entire life cycle, the method comprising the steps of:

[0088] S100: Collect medical equipment operating data in real time and collect data on the entire life cycle of medical equipment. Based on the collected data, dynamically build a digital twin model of the entire life cycle of medical equipment, and combine the operating data of each medical device with its corresponding digital twin model of the entire life cycle;

[0089] Specifically, for example, the operating data of CT scanner A is collected. Medical equipment operating data includes various medical operating parameters, environmental monitoring data, and clinical usage data. The full life cycle data includes the procurement phase, installation phase, maintenance phase, and business phase. The procurement phase includes the equipment model, manufacturer, warranty period, and energy consumption level. The installation phase includes the installation location, initialization and commissioning report, and acceptance test record. The maintenance phase includes the fault type, fault description, maintenance time, and maintenance strategy. Business data includes the frequency of medical equipment use, usage time, and usage feedback. The operating data of the CT scanner includes: tube temperature: current value is 75°C, tube voltage: 120kV, tube current: 300mA, ambient temperature and humidity: 25°C / 50%RH. Clinical usage data: 48 cases scanned that day, with an average dose of 8mSv.

[0090] Furthermore, step S100 specifically includes:

[0091] S110. Dynamically construct a full-lifecycle digital twin model of medical equipment. The full-lifecycle digital twin model includes a geometric model and behavioral rules. The geometric model builds a 3D visual digital twin model based on the equipment BOM list and CAD drawings. The behavioral rules define the medical equipment operating mode conversion logic through a state machine and embed the communication rules of the medical equipment control protocol. By parsing the original communication protocol of the equipment, the equipment state switching instructions and parameter reading and writing rules are extracted and converted into the transition conditions of the state machine. For example: Based on the CAD drawing of CT scanner A, a digital twin model is constructed, the positions of key components such as the tube, detector, and cooling system are marked, and the mode conversion logic of standby-preheating-scanning-cooling is defined through the state machine.

[0092] S120. Use the collected life cycle data of the medical device as the initial attributes of the life cycle digital twin model of the medical device; synchronously update the collected medical device operation data and the corresponding parameters of the life cycle digital twin model of the medical device; convert the maintenance phase data in the life cycle data into health status assessment factors to drive the adaptive adjustment of the life cycle digital twin model; for example, synchronously update the collected real-time data of CT scanner A with the corresponding digital twin model;

[0093] S200, connect each medical device to the swarm intelligence network, making it a node in the network; construct a medical device knowledge graph based on device type, functional association, and spatial distribution, and establish an association network between medical devices;

[0094] Specifically, step S200 includes:

[0095] S210. Connect medical devices to the swarm intelligence network, assign nodes to each device, and register basic information about the medical devices, including device type, unique identifier, communication protocol, and physical location. For example, register the technical information of medical device A for CT scanner A: assign the node to CT-1, the device type to imaging, the communication protocol to Modbus TCP, the physical location to Radiology Room 1, and the edge computing resources to a 4-core CPU and 2TB of storage. Each node is allocated computing resources and storage quotas for local data processing. When the device enters a high-load state, the edge controller dynamically schedules cluster resources and temporarily increases the node computing power quota to ensure real-time data processing.

[0096] S220: Extract key entities from the collected data. Key entities include equipment entities, function entities, and space entities. For example, extract equipment entities: CT scanner A: CT-1, high-pressure injector: HS-1, PACS server: PACS-1.

[0097] S230. Establish association relationships, including functional dependency relationships, physical connection relationships, and process linkage relationships, to construct a network structure of the medical equipment knowledge graph. For example, the physical connection relationship is: CT-1 - public power supply - radiology department power distribution circuit L1; the functional dependency relationship is: CT-1 - image data generation - PACS-1; the process linkage relationship is: CT-1 - examination process dependency - HS-1;

[0098] S240. Update the association weights in real time based on the frequency of data interaction between medical devices, and identify sets of medical devices with tight functional coupling or high-frequency interaction based on a graph clustering algorithm; automatically update the graph structure through an incremental learning algorithm when devices are added, retired, or upgraded.

[0099] S300: When a medical device node detects abnormal data, it sends a warning signal to the full life cycle digital twin model of the medical device associated in the medical device knowledge graph in combination with the medical device knowledge graph. The full life cycle digital twin model that receives the warning signal activates a self-check program.

[0100] Specifically, step S300 includes:

[0101] S310: The medical device node compares the collected operating data with the dynamic threshold in real time, triggering anomaly detection when the data deviates from the normal range; calculating the degree of deviation, duration, and impact range of the abnormal data, and classifying the anomaly into primary, intermediate, and advanced levels;

[0102] Furthermore, S311 sets a basic threshold range and a safety mandatory threshold for each parameter based on the medical device design parameters, historical operating benchmark data, and industry standards for similar devices. For example, the bulb temperature is designed to be 40-70°C, and the safety mandatory threshold is 75°C. If the mandatory threshold is reached, the device is forced to shut down.

[0103] S312. Extract adjustment factors based on the full lifecycle data of the medical device, including the age, frequency of use, maintenance history, and environmental data of the medical device; assign dynamic weights to the adjustment factors, which are dynamically adjusted based on the device type and parameter characteristics;

[0104] S313. Utilize a reinforcement learning algorithm, with the reward function being to reduce the false alarm rate and missed alarm rate of anomaly detection. The state space is defined by the deviation between the real-time operating parameters and the current threshold, and the action space is defined by the threshold adjustment range. The action space definition includes the threshold adjustment range, as well as the warning level adjustment. Based on real-time data and historical adjustment records, the basic threshold range is automatically optimized to form a dynamic threshold. Combined with historical fault data and real-time data feedback, the dynamic threshold model is updated hourly to ensure that the threshold adapts to equipment aging and environmental changes. For example, if the service life is 3 years, the weight is 40%, the threshold is tightened by 5%, and the upper limit is adjusted to 70°C × 0.95 = 66.5°C. If the frequency of use is 45 cases / day, which is 30 cases above the average, the weight is 30%, the threshold is relaxed by 3%, and the upper limit is adjusted to 66.5°C × 1.03 ≈ 68.5°C. If the maintenance history is: no major faults in the past year, the weight is 20%, the ambient temperature is 25°C, and the standard temperature range is 18-28°C, the weight is 10%, and the comprehensive dynamic threshold is 40-68.5°C. This is for illustration only and is not intended to be limiting.

[0105] S320. Starting from the abnormal device node, analyze the associated medical devices based on the functional dependency, physical connection, emergency linkage, and association weight in the medical device knowledge graph. For example, identify strongly associated devices such as PACS servers and high-pressure injectors through the knowledge graph and send an alert to their digital twin models.

[0106] S330. Send an early warning signal to the associated medical device. The early warning signal includes the unique identifier of the abnormal device, abnormal parameters, timestamp, and abnormal level. For example, if the current tube temperature of 70°C is greater than the dynamic threshold of 68.5°C, triggering a high-level abnormality, the early warning signal includes: device ID: CT-1, abnormal parameters: tube temperature 70°C, timestamp: 2025-05-01-14:30, and level: high.

[0107] S340. The full life cycle digital twin model that receives the early warning signal activates the self-check program, collects the current operating parameters of the medical equipment, and checks the parameters related to the abnormal data.

[0108] S400: By comparing historical cases, evaluating related impacts, and simulating fault propagation paths, we can preliminarily determine the cause and location of the fault and generate an analysis report.

[0109] Specifically, step S400 includes:

[0110] S410. Compare the abnormal parameters and time series data of the abnormal device with similar fault data in historical data, analyze the degree of match between the current abnormality and historical data in terms of fault manifestation, occurrence environment, and associated device status, and extract the corresponding fault cause, maintenance plan, and maintenance result. For example, a similar fault in 2024 is matched: high tube temperature, the historical cause is: cooling system fan failure, the maintenance plan is: fan replacement, and the result is: successful repair.

[0111] S420. Based on the functional dependencies, physical connections, and process linkages in the medical device knowledge graph, assess the potential impact of abnormal device failures on associated devices, calculate association weights, identify associated medical devices with association weights above a threshold, and analyze the chain reactions caused by the failure in combination with the current operating status of the associated medical devices. For example, an idle high-pressure injector may increase the load of the adjacent CT room by 20%.

[0112] S430. Utilize the behavioral rules of the full lifecycle digital twin model, combined with the medical device knowledge graph, to dynamically simulate the fault propagation path. This dynamic simulation starts from the abnormal medical device node and, based on the connection relationships and operating logic between medical devices, deduce the fault's diffusion trajectory in the device network to assist in locating the root cause of the fault. For example, the fault propagation path may be: tube overheating - cooling system failure - triggering hardware protection - CT shutdown - radiology examination queue backlog.

[0113] S440, based on historical case matching results, correlation impact assessment, and fault propagation path simulation, uses an evidence fusion algorithm to assign weights to candidate fault causes, screens out the fault causes and fault locations with the highest probability, and generates an analysis report. The analysis report includes an overview of the abnormal event, preliminary diagnostic conclusions, fault impact range, abnormality level, historical case references, and recommended maintenance strategies. The report is then simultaneously pushed to the equipment management platform, the operation and maintenance personnel terminal, and the digital twin model of the abnormal equipment.

[0114] S500: Perform multi-dimensional verification of diagnostic results in a digital twin environment. This verification includes fault scenario reproduction, maintenance plan simulation, and aging trend deduction. It comprehensively evaluates the risk factor of the plan and outputs the optimal maintenance strategy.

[0115] Specifically, step S500 includes:

[0116] S510. Based on the abnormal event overview, time series data, and fault propagation path in the analysis report, reproduce the equipment operating status before the fault occurs in the digital twin environment, synchronously load the environmental data and associated equipment operating parameters, and build a complete fault trigger condition.

[0117] S520. Adjust the parameters in the recurrence scenario one by one by using the control variable method to observe whether the fault phenomenon reappears or changes, and verify the sensitivity of the fault cause in the preliminary diagnosis conclusion; if the fault phenomenon disappears or is significantly alleviated after the parameter adjustment, mark the parameter as a possible fault cause; otherwise, exclude the parameter;

[0118] S530. Execute the recommended maintenance strategy in the analysis report in the digital twin model, simulate the impact of the maintenance strategy on the overall performance of the equipment, and generate quantitative evaluation results by calculating the performance recovery rate, maintenance time, and resource consumption to evaluate the effectiveness of the maintenance strategy.

[0119] S540: Build a device health model based on maintenance records, failure frequencies, and performance degradation trends in the full lifecycle data to quantify the current degree of device aging.

[0120] S550, based on fault scenario reproduction, maintenance plan simulation and aging trend deduction, extract risk assessment indicators, build a decision tree model, and output the optimal maintenance strategy. The maintenance strategy includes continued maintenance and recommended scrapping. Continued maintenance includes preventive maintenance, corrective maintenance and upgrade and modification. Recommended scrapping includes emergency scrapping and planned scrapping.

[0121] Furthermore, risk assessment indicators include technical indicators, time indicators, economic indicators, compliance indicators and aging indicators; technical indicators include maintenance operation complexity, success rate and impact on equipment accuracy; time indicators include maintenance time, downtime and business recovery time; economic indicators include maintenance costs and potential losses; compliance indicators include medical device regulatory requirements, and aging indicators include the impact of maintenance strategies on the remaining life of the equipment.

[0122] S600: During the maintenance process, the maintenance process is recorded in real time, and the maintenance data and equipment status updates are synchronously fed back to the digital twin model of the equipment's entire life cycle.

[0123] The present invention provides another technical solution for CT scanner fault verification and maintenance strategy optimization based on a full life cycle medical equipment monitoring and analysis method;

[0124] Abnormal event data obtained from the analysis report: At 2:30 PM on May 1, 2025, the tube temperature reached 70°C (exceeding the dynamic threshold of 68.5°C), triggering a shutdown. One hour before the failure, the tube temperature continued to rise from 55°C at an average rate of 3°C / 10 minutes, and the cooling fan speed dropped sharply from 2000 rpm to 0 rpm. The ambient temperature was 28°C and the humidity was 45% RH.

[0125] Related equipment parameters: Cooling system water pump flow rate 15L / min, standard value ≥20L / min, heat sink temperature 65℃, normal ≤50℃;

[0126] Data was loaded synchronously in the digital twin environment to simulate dual scenarios: cooling fan stall and insufficient water pump flow. The tube temperature was observed to rise at a rate of 2.5°C / minute, reaching 70°C after 18 minutes. This was consistent with the actual fault, confirming that the triggering conditions were fan stall and reduced cooling efficiency.

[0127] By adjusting parameters and observing whether the fault phenomenon reappears or changes, the sensitivity of the fault cause in the preliminary diagnosis conclusion was verified. Analysis showed that fan speed has the greatest impact on temperature, with a sensitivity coefficient of 0.8, water pump flow sensitivity of 0.4, and room temperature sensitivity of 0.3. It was confirmed that fan failure was the main cause, and other factors were auxiliary factors.

[0128] The maintenance strategy was simulated. Replacing the fan took 2 hours, and cleaning the heat sink took 0.5 hours. After the simulation, the tube temperature stabilized at 55°C, and the performance recovery rate was 100%, confirming that the maintenance strategy was feasible.

[0129] Based on maintenance records, failure frequency, and performance degradation trends in the full lifecycle data, an equipment health status model was constructed to quantify the current degree of equipment aging. Maintenance records included: major maintenance every three years, with a weight of 30%; failure frequency: no failures in the past year, with a weight of 20%; performance degradation: tube noise increased from 15HU to 18HU, with a weight of 50%. The resulting health score was 74. Based on data from similar equipment, a health score of 74 corresponds to a remaining lifespan of approximately 2.8 years (design life is 6 years).

[0130] Based on fault scenario reproduction, maintenance plan simulation, and aging trend deduction, risk assessment indicators were extracted and a decision tree model was constructed. The optimal maintenance strategy was output as follows: replace the fan, clean the heat sink quarterly, and monitor the fan speed monthly. The current health score of 74 points is greater than the scrap threshold of 60 points, so the scrapping process is not triggered for the time being.

[0131] The present invention provides another technical solution, wherein a CT scanner detects abnormal data;

[0132] The CT scanner compares the collected operating data with dynamic thresholds in real time, triggering anomaly detection when the data deviates from the normal range; calculates the degree of deviation, duration, and impact range of the abnormal data, and classifies the anomaly into primary, intermediate, and advanced levels;

[0133] The degree of deviation of the abnormal data is expressed as follows:

[0134]

[0135] Where D represents the degree of deviation; x t It represents the specific data value collected at time t; μ represents the dynamic threshold; σ represents the dynamic standard deviation, which is used to measure the degree of dispersion of the data. The dynamic standard deviation changes as the data is updated.

[0136] The impact range is as follows:

[0137]

[0138] Where I represents the impact range index; w i Expressed as the weight of the i-th associated device; L i It represents the current load of the i-th device, which indicates the workload of the device at the current moment; i represents the index of the associated device; It is expressed as the average load; n is the total number of associated devices;

[0139] Map the impact range index to the impact range interval: low impact: I∈[0,0.3), impact range [0%,30%]; medium impact: I∈[0.3,0.7), impact range [30%,70%]; high impact: I∈[0.7,1], impact range [70%,100%];

[0140] CT scanner associated equipment:

[0141] PACS server: w1 = 0.6, L1 = 85%;

[0142] High-pressure injector: w2 = 0.4, L2 = 30%;

[0143] Load Average

[0144] Impact range index I = 0.6 × |85-65| / 65 + 0.4 × |30-65| / 65 = 0.4; determined to be a medium impact;

[0145] The abnormality is graded as follows:

[0146] G=f(d,t,i)=α×D+β×T+γ×I;

[0147] Where G represents the anomaly score; D represents the degree of deviation; T represents the duration level; I represents the impact range index; α represents the weight coefficient of the degree of deviation; β represents the weight coefficient of the duration level; γ represents the weight coefficient of the impact range; the sum of the weight coefficients is 1;

[0148] The weight coefficient is based on the preset initial weight of the equipment type, and is dynamically adjusted based on the quarterly statistical analysis of the prediction accuracy of each dimension for actual failures. The formula is as follows:

[0149]

[0150] Among them, α′ represents the adjusted weight coefficient; A D Expressed as the historical accuracy of the deviation dimension; A arg Expressed as average accuracy;

[0151] Abnormality level: Primary abnormality (0, 3) records the error in the log and checks it during the next maintenance. Intermediate abnormality (0.3, 0.6) schedules diagnostic maintenance within 48 hours. Advanced abnormality (0.6, 1) immediately shuts down the system and initiates the emergency plan. This is for illustrative purposes only and is not intended to be limiting.

[0152] The present invention provides another technical solution, in which a CT scanner detects abnormal data, simulates the impact of a maintenance strategy on the overall performance of the equipment, and evaluates the effectiveness of the maintenance strategy;

[0153] The performance recovery rate is obtained by combining the performance index value at the time of failure and the performance index value after maintenance, and the formula is as follows:

[0154]

[0155] Where η is the performance recovery rate; P post Expressed as the performance index value after maintenance; P fall Expressed as the performance index value at the time of failure; P std Expressed as standard performance index values;

[0156] The maintenance time is recorded from the start to the end of the maintenance operation;

[0157] The resource consumption is analyzed according to the recommended maintenance strategy to obtain the maintenance efficiency. The formula is as follows:

[0158]

[0159] Where E represents maintenance efficiency; T aIt is expressed as maintenance time; C is the resource consumption coefficient, which is used to measure the resources consumed during the maintenance process. The more resources are consumed, the greater the resource consumption coefficient; η is the performance recovery rate;

[0160] The resource consumption coefficient is as follows:

[0161] C = 0.4 × C time +0.3×C cost +0.2×C human +0.1×C energy ;

[0162] Where C represents the resource consumption coefficient; C time Expressed as time consumption, time consumption is the ratio of actual time consumption to standard time consumption; C cost Expressed as cost consumption, cost consumption is the ratio of the cost of accessories to the original value of the equipment; C human Expressed as manpower consumption, the required manpower qualifications are divided according to the abnormality level; C energy Expressed as energy consumption, energy consumption is the ratio of actual energy consumption to standard energy consumption;

[0163] Maintenance of the CT scanner actually takes 2 hours (the standard is 4 hours), C time =0.5;

[0164] The cost of the accessories is E (the original value of the equipment is F), C cost =E / F;

[0165] The human qualification required for abnormal level is Level 2 Engineer, C human =0.4;

[0166] Actual consumption is 2kWh (standard energy consumption is 10kWh), C energy =0.2;

[0167] Resource consumption coefficient C = 0.333;

[0168] Build an equipment health model based on maintenance records, failure frequencies, and performance degradation trends in the full lifecycle data to quantify the current degree of equipment aging;

[0169] The current aging degree of the quantitative equipment is as follows:

[0170]

[0171] Where H(t) represents the health status of the device at time t; H0 represents the initial health status; λ(ι) represents the aging rate, the aging rate of the device at time ι; t represents the time index;

[0172] The aging rate is expressed as follows:

[0173]

[0174] Where λ(t) represents the aging rate of the device at time t; w k Expressed as the weight of the kth aging factor, f k (t) represents the cumulative value of the kth aging factor at time t; k represents the index of the aging factor; m represents the total number of aging factors;

[0175] The cumulative value f of the kth aging factor at time t k (t) = b k ×t;b k Expressed as the basic equipment aging rate, the basic equipment aging rate is obtained based on the historical aging situation of similar equipment;

[0176] CT scanner aging factors include X-ray tube degradation, wear of mechanical rotating parts, decreased detector sensitivity, and reduced cooling system efficiency;

[0177] The basic aging rate of X-ray tube attenuation is 0.08 / year. The CT scanner has been in use for 3 years. The cumulative value of X-ray tube attenuation in 3 years is f1(3)=0.08×3=0.24;

[0178] S550. Based on fault scenario reproduction, maintenance plan simulation, and aging trend deduction, risk assessment indicators are extracted, a decision tree model is constructed, and the optimal maintenance strategy is output. The maintenance strategy includes continued maintenance and recommended scrapping. The continued maintenance includes preventive maintenance, corrective maintenance, and upgrade and modification. The recommended scrapping includes emergency scrapping and planned scrapping.

[0179] The present invention provides another technical solution, a medical equipment monitoring and analysis system based on the entire life cycle, which includes: a data acquisition module, a digital twin module, a knowledge graph module, an intelligent diagnosis module, a maintenance strategy module and a database module;

[0180] The data acquisition module includes an operating data module, a full life cycle data module, and a data preprocessing module; the operating data module acquires medical equipment operating parameters, environmental data, and clinical usage data in real time; the full life cycle data module integrates procurement, installation, maintenance, and business data of medical equipment; and the data preprocessing module is used to perform noise reduction, normalization, and time series alignment on the collected data.

[0181] The digital twin module includes a modeling module, a behavior rule module, and a dynamic update module. The modeling module builds a digital twin model of the medical device based on the BOM list or CAD drawing of the medical device. The behavior rule module implements state machine control logic and device protocol parsing. The dynamic update module synchronizes the real-time operating data of the medical device with the parameters of the digital twin model.

[0182] The knowledge graph module includes a node management module, a graph construction module, and a graph optimization module. The node management module assigns a unique identifier, computing resources, and communication links to each medical device; the graph construction module extracts key entities and builds associations; and the graph optimization module uses a graph clustering algorithm to identify device groups and incrementally update the graph.

[0183] The intelligent diagnosis module includes a dynamic threshold module, a multi-level warning module, and a fault location module. The dynamic threshold module dynamically adjusts the threshold based on reinforcement learning. The multi-level warning module sends warning signals according to abnormality classification standards. The fault location module combines fault propagation simulation and historical case matching to perform root cause analysis.

[0184] The maintenance strategy module includes a virtual verification module, an aging assessment module, and a decision tree module; the virtual verification module reproduces faults and simulates maintenance strategies in a digital twin environment; the aging assessment module predicts the remaining life of medical equipment; and the decision tree module outputs the optimal maintenance strategy.

[0185] The database module includes a maintenance tracking module, a knowledge base module and a performance feedback module; the maintenance tracking module is used to record maintenance operation steps, spare parts batches and quality inspection results; the knowledge base module adds new failure modes to the case library and optimizes maintenance strategies; the performance feedback module is used to record various evaluation indicators of medical equipment.

[0186] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A medical equipment monitoring and analysis method based on the entire life cycle, characterized by: The method includes: S100: Collect medical equipment operating data in real time and collect data on the entire life cycle of medical equipment. Based on the collected data, dynamically build a digital twin model of the entire life cycle of medical equipment, and combine the operating data of each medical device with its corresponding digital twin model of the entire life cycle; S200, connect each medical device to the swarm intelligence network, making it a node in the network; construct a medical device knowledge graph based on device type, functional association, and spatial distribution, and establish an association network between medical devices; S300: When a medical device node detects abnormal data, it sends a warning signal to the full life cycle digital twin model of the medical device associated in the medical device knowledge graph in combination with the medical device knowledge graph. The full life cycle digital twin model that receives the warning signal activates a self-check program. S400: By comparing historical cases, evaluating related impacts, and simulating fault propagation paths, we can preliminarily determine the cause and location of the fault and generate an analysis report. S500: Perform multi-dimensional verification of the diagnosis results in the digital twin environment. The multi-dimensional verification includes reproducing the fault scenario, simulating the maintenance plan, and deducing the aging trend. The risk factor of the plan is comprehensively evaluated to output the optimal maintenance strategy. S600: During the maintenance process, the maintenance process is recorded in real time, and the maintenance data and equipment status updates are synchronously fed back to the digital twin model of the equipment's entire life cycle.

2. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 1 is characterized in that: The medical equipment operation data includes various medical operation parameters, environmental monitoring data and clinical usage data; the full life cycle data includes the procurement stage, installation stage, maintenance stage and business stage. The procurement stage includes the equipment model, manufacturer, warranty period and energy consumption level; the installation stage includes the installation location, initialization commissioning report and acceptance test record; the maintenance stage includes the fault type, fault description, maintenance time and maintenance strategy; the business data includes the frequency of use of medical equipment, usage time and usage feedback; The collected data is denoised, normalized, and time-series aligned, and invalid data points are filtered out.

3. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 1 is characterized in that: Step S100 includes: S110. Dynamically construct a full-lifecycle digital twin model of the medical device, the full-lifecycle digital twin model including a geometric model and behavioral rules; the geometric model constructs a three-dimensional visual digital twin model based on the device's bill of materials (BOM) and CAD drawings; the behavioral rules define the medical device's operating mode transition logic through a state machine and embed the communication rules of the medical device control protocol; S120. Use the collected full life cycle data of the medical device as the initial attributes of the full life cycle digital twin model of the medical device; synchronously update the collected medical device operation data and the corresponding parameters of the full life cycle digital twin model of the medical device; convert the data of the maintenance phase in the full life cycle data into a health status assessment factor to drive the adaptive adjustment of the full life cycle digital twin model.

4. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 1 is characterized in that: Step S200 includes: S210: Connect medical devices to the swarm intelligence network, assign a node to each device, and register basic information about the medical devices, including device type, unique identifier, communication protocol, and physical location. Computing resources and storage quotas are allocated to each node for local data processing. S220, extracting key entities from the collected data, wherein the key entities include equipment entities, function entities, and space entities; S230: Establish association relationships, including functional dependency relationships, physical connection relationships, and process linkage relationships, to construct a network structure of the medical device knowledge graph; S240. Update the association weights in real time based on the frequency of data interaction between medical devices, and identify sets of medical devices with tight functional coupling or high-frequency interaction based on a graph clustering algorithm; automatically update the graph structure through an incremental learning algorithm when devices are added, retired, or upgraded.

5. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 1 is characterized in that: Step S300 includes: S310: The medical device node compares the collected operating data with the dynamic threshold in real time, and triggers anomaly detection when the data deviates from the normal range; calculates the deviation degree, duration, and impact range of the abnormal data, and classifies the anomaly into primary anomaly, intermediate anomaly, and advanced anomaly; S320: Starting from the abnormal device node, analyzing the associated medical devices based on the functional dependency, physical connection, emergency linkage, and association weight in the medical device knowledge graph; S330, sending the warning signal to the associated medical device, wherein the warning signal includes a unique identifier of the abnormal device, abnormal parameters, a timestamp, and an abnormal level; S340. The full life cycle digital twin model that receives the early warning signal activates the self-check program, collects the current operating parameters of the medical equipment, and checks the parameters related to the abnormal data.

6. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 5, characterized in that: The dynamic threshold includes: S311. Set basic threshold ranges and safety mandatory thresholds for each parameter based on medical device design parameters, historical operating benchmark data, and industry standards for similar devices; S312. Extracting adjustment factors based on the full lifecycle data of the medical device, the adjustment factors including the age, frequency of use, maintenance history, and environmental data of the medical device; assigning dynamic weights to the adjustment factors, the dynamic weights being dynamically adjusted based on the device type and parameter characteristics; S313. Utilize a reinforcement learning algorithm, with the reduction of the false alarm rate and missed alarm rate of anomaly detection as the reward function, the deviation between the real-time operating parameters and the current threshold as the state space, and the threshold adjustment range as the action space. Based on real-time data and historical adjustment records, automatically optimize the basic threshold range to form a dynamic threshold.

7. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 1 is characterized in that: Step S400 includes: S410: Compare the abnormal parameters and time series data of the abnormal equipment with the similar fault data in the historical data, analyze the matching degree between the current abnormality and the historical data in terms of fault manifestation, occurrence environment and related equipment status, and extract the corresponding fault cause, maintenance plan and maintenance result; S420: Based on the functional dependencies, physical connections, and process linkages in the medical device knowledge graph, evaluate the potential impact of abnormal device failures on associated devices, calculate association weights, identify associated medical devices with association weights above a threshold, and analyze the chain reaction caused by the failure in combination with the current operating status of the associated medical devices; S430. Utilizing the behavioral rules of the full lifecycle digital twin model and combining it with the medical device knowledge graph, dynamically simulate the fault propagation path. The dynamic simulation starts from the abnormal medical device node and, based on the connection relationship and operation logic between medical devices, deduces the propagation trajectory of the fault in the device network to assist in locating the root cause of the fault. S440, based on historical case matching results, correlation impact assessment, and fault propagation path simulation, uses an evidence fusion algorithm to assign weights to candidate fault causes, screens out the most likely fault causes and fault locations, and generates an analysis report. The analysis report includes an overview of the abnormal event, preliminary diagnostic conclusions, fault impact range, abnormality level, historical case references, and recommended maintenance strategies. The report is then simultaneously pushed to the device management platform, the operation and maintenance personnel terminal, and the digital twin model of the abnormal device.

8. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 1 is characterized in that: Step S500 includes: S510. Based on the abnormal event overview, time series data, and fault propagation path in the analysis report, reproduce the equipment operating status before the fault occurs in the digital twin environment, synchronously load the environmental data and associated equipment operating parameters, and build a complete fault trigger condition. S520. Using the control variable method, adjust the parameters in the recurrence scenario one by one to observe whether the fault phenomenon reappears or changes, and verify the sensitivity of the fault cause in the preliminary diagnosis conclusion; if the fault phenomenon disappears or is significantly alleviated after the parameter adjustment, mark the parameter as a possible fault cause; otherwise, exclude the parameter; S530. Execute the recommended maintenance strategy in the analysis report in the digital twin model, simulate the impact of the maintenance strategy on the overall performance of the equipment, and generate quantitative evaluation results by calculating the performance recovery rate, maintenance time, and resource consumption to evaluate the effectiveness of the maintenance strategy. S540: Build a device health model based on maintenance records, failure frequencies, and performance degradation trends in the full lifecycle data to quantify the current degree of device aging. S550. Based on fault scenario reproduction, maintenance plan simulation, and aging trend deduction, risk assessment indicators are extracted, a decision tree model is constructed, and the optimal maintenance strategy is output. The maintenance strategy includes continued maintenance and recommended scrapping. The continued maintenance includes preventive maintenance, corrective maintenance, and upgrade and modification. The recommended scrapping includes emergency scrapping and planned scrapping.

9. The medical equipment monitoring and analysis method based on the entire life cycle according to claim 8, characterized in that: The risk assessment indicators include technical indicators, time indicators, economic indicators, compliance indicators and aging indicators; the technical indicators include the complexity and success rate of maintenance operations and the impact on equipment accuracy; the time indicators include maintenance time, downtime and business recovery time; the economic indicators include maintenance costs and potential losses; the compliance indicators include medical device regulatory requirements, and the aging indicators include the impact of maintenance strategies on the remaining life of the equipment.

10. A medical equipment monitoring and analysis system based on the entire life cycle, characterized by: The system includes: data acquisition module, digital twin module, knowledge graph module, intelligent diagnosis module, maintenance strategy module and database module; The data acquisition module includes an operating data module, a full life cycle data module, and a data preprocessing module; the operating data module acquires medical equipment operating parameters, environmental data, and clinical usage data in real time; the full life cycle data module integrates procurement, installation, maintenance, and business data of medical equipment; and the data preprocessing module is used to perform noise reduction, normalization, and time series alignment on the collected data. The digital twin module includes a modeling module, a behavior rule module, and a dynamic update module; the modeling module builds a digital twin model of the medical device based on the BOM list or CAD drawing of the medical device; the behavior rule module implements state machine control logic and device protocol parsing; the dynamic update module synchronizes the real-time operating data of the medical device with the parameters of the digital twin model; The knowledge graph module includes a node management module, a graph construction module, and a graph optimization module. The node management module assigns a unique identifier, computing resources, and communication links to each medical device; the graph construction module extracts key entities and builds association relationships; and the graph optimization module identifies device groups through a graph clustering algorithm and incrementally updates the graph. The intelligent diagnosis module includes a dynamic threshold module, a multi-level warning module, and a fault location module; the dynamic threshold module dynamically adjusts the threshold based on reinforcement learning; the multi-level warning module sends warning signals according to the abnormality classification standard; and the fault location module performs root cause analysis by combining fault propagation simulation and historical case matching. The maintenance strategy module includes a virtual verification module, an aging assessment module, and a decision tree module; the virtual verification module reproduces faults in a digital twin environment and simulates maintenance strategies; the aging assessment module predicts the remaining life of medical equipment; and the decision tree module outputs the optimal maintenance strategy. The database module includes a maintenance tracking module, a knowledge base module and a performance feedback module; the maintenance tracking module is used to record maintenance operation steps, spare parts batches and quality inspection results; the knowledge base module adds new failure modes to the case library and optimizes maintenance strategies; the performance feedback module is used to record various evaluation indicators of medical equipment.

Images