Medical equipment monitoring method and system based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a medical equipment monitoring method and system based on artificial intelligence. The method comprises the following steps: acquiring operation monitoring data of medical equipment; performing medical equipment operation state feature analysis based on the medical equipment operation monitoring data to generate medical equipment operation state feature data; acquiring historical medical equipment abnormal operation data; performing medical equipment abnormal operation global decision factor analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state feature data to generate an abnormal operation global decision factor; constructing an abnormal operation evaluation optimization decision tree model based on the abnormal operation global decision factor; and transmitting the medical equipment operation state characteristic data to the abnormal operation evaluation optimization decision tree model to execute medical equipment abnormal operation trend evaluation, and generating medical equipment abnormal operation trend evaluation data. According to the invention, intelligent and efficient monitoring and detection of the medical equipment are realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based medical equipment monitoring method and system. Background Art

[0002] In the modern medical industry, the widespread use of medical devices has significantly improved the efficiency and accuracy of diagnosis and treatment. With the increasing types and number of medical devices, real-time monitoring and abnormal management of their operating status have become important challenges faced by medical institutions. The rise of Internet of Things technology has provided new ideas for the networked monitoring of medical devices. By connecting different medical devices and realizing real-time data collection and transmission, it provides more comprehensive information support for equipment management. In addition, through artificial intelligence technology, such as intelligent algorithms with machine learning and deep learning as the core, it has shown great potential in the fields of data processing and anomaly detection in recent years, feature extraction, trend analysis and anomaly prediction of medical equipment operating status. However, the existing medical equipment monitoring methods based on artificial intelligence are mostly focused on a single device or a single algorithm model, lacking a comprehensive analysis of the complex correlation between devices and an in-depth exploration of the causal relationship of abnormal triggering. Due to the diversity, real-time and complexity of medical equipment operating data, traditional data processing and analysis methods are difficult to effectively extract the key features of the equipment operating status, and are even more unable to efficiently identify the potential risks of equipment operating abnormalities. Summary of the invention

[0003] Based on this, the present invention provides an artificial intelligence-based medical equipment monitoring method and system to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above purpose, a medical device monitoring method based on artificial intelligence includes the following steps: Step S1: performing a medical device interconnection operation on the medical device using a preset Internet of Things communication interface, and performing medical device operation monitoring processing on the medical device after the medical device interconnection operation to obtain medical device operation monitoring data; performing medical device operation perception topology data analysis on the medical device operation monitoring data to generate medical device operation perception topology data; Step S2: performing a state characteristic analysis of the operation of the medical device based on the medical device operation perception topology data to generate medical device operation state characteristic data; Step S3: Acquire historical medical equipment abnormal operation data; perform historical medical equipment abnormal operation state characteristic analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state characteristic data to generate historical medical equipment abnormal operation state characteristic data; perform global decision factor analysis of medical equipment abnormal operation based on the historical medical equipment abnormal operation state characteristic data to generate abnormal operation global decision factor; Step S4: Calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor to generate abnormal operation triggering condition distribution probability data; perform abnormal operation triggering causal contribution analysis based on the abnormal operation triggering condition distribution probability data to generate abnormal operation triggering causal contribution data; construct an abnormal operation evaluation decision tree model based on the abnormal operation global decision factor; use the abnormal operation triggering causal contribution data to perform training optimization processing on the abnormal operation evaluation decision tree model for the abnormal operation triggering causal contribution of the model to generate an abnormal operation evaluation optimization decision tree model; Step S5: Transmitting the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform abnormal operation trend evaluation processing of the medical equipment, and generate abnormal operation trend evaluation data of the medical equipment; establishing an automated feedback engine for medical equipment fault response based on the abnormal operation trend evaluation data of the medical equipment, and generating a medical equipment fault response intelligent feedback engine; executing the medical equipment monitoring fault response feedback task through the medical equipment fault response intelligent feedback engine.

[0005] This specification also provides an artificial intelligence-based medical device monitoring system for executing the artificial intelligence-based medical device monitoring method as described above. The artificial intelligence-based medical device monitoring system includes: The medical device operation perception topology analysis module is used to perform a medical device interconnection operation on the medical device using a preset Internet of Things communication interface, and perform medical device operation monitoring processing on the medical device after the medical device interconnection operation to obtain medical device operation monitoring data; perform medical device operation perception topology data analysis on the medical device operation monitoring data to generate medical device operation perception topology data; A medical equipment operation status characteristic analysis module is used to analyze the operation status characteristics of the medical equipment based on the medical equipment operation perception topology data, and generate the medical equipment operation status characteristic data; The abnormal operation global decision factor analysis module is used to obtain historical medical equipment abnormal operation data; perform historical medical equipment abnormal operation state characteristic analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state characteristic data, and generate historical medical equipment abnormal operation state characteristic data; perform global decision factor analysis on abnormal operation of medical equipment based on the historical medical equipment abnormal operation state characteristic data, and generate abnormal operation global decision factors; The abnormal operation evaluation optimization decision tree construction module is used to calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor, and generate abnormal operation triggering condition distribution probability data; perform abnormal operation triggering causal contribution analysis based on the abnormal operation triggering condition distribution probability data, and generate abnormal operation triggering causal contribution data; construct an abnormal operation evaluation decision tree model based on the abnormal operation global decision factor; use the abnormal operation triggering causal contribution data to perform training optimization processing on the abnormal operation evaluation decision tree model for the abnormal operation triggering causal contribution, and generate an abnormal operation evaluation optimization decision tree model; The medical equipment monitoring fault response feedback module is used to transmit the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform the medical equipment abnormal operation trend evaluation processing and generate the medical equipment abnormal operation trend evaluation data; establish the medical equipment fault response automation feedback engine based on the medical equipment abnormal operation trend evaluation data and generate the medical equipment fault response intelligent feedback engine; perform the medical equipment monitoring fault response feedback operation through the medical equipment fault response intelligent feedback engine.

[0006] The beneficial effect of the present application is that the preset Internet of Things communication interface is used to perform Unicom operations on medical devices, overcoming the difficulties caused by incompatible device protocols in traditional equipment monitoring. At the same time, through the medical equipment operation monitoring and processing, the operation data of medical equipment can be collected in real time, providing comprehensive basic data support for subsequent state analysis and abnormality detection. According to the Internet of Things communication interface corresponding to the medical equipment operation monitoring data, the monitoring perception node identification is carried out to accurately identify the perception node and its specific state of each medical device. Based on the identification monitoring perception node, the medical equipment operation monitoring data is analyzed for the medical equipment operation perception topology data, and the operation relationship and external condition interaction state between medical devices are intuitively displayed. According to the medical equipment operation perception topology data, the state variable analysis is carried out, and the core variable information reflecting the operation status of the medical equipment is extracted from the complex topology data, which effectively reduces the interference of redundant data on subsequent processing, improves the efficiency and accuracy of data processing, and performs key variable analysis on the medical equipment state variable data, improves the pertinence of state feature analysis, and significantly reduces data complexity. According to the key variable data of the medical equipment state, the state feature analysis of the medical equipment operation perception topology data can more deeply explore the characteristic mode and change law of the equipment operation. Achieve accurate characterization of the current state of the equipment, provide an important basis for early identification of abnormal states, and make the perception of the operating state of medical equipment more comprehensive and efficient. Obtain historical abnormal operation data of medical equipment to provide real and representative abnormal operation sample data for the current medical equipment monitoring system, covering various abnormal scenarios that occur in actual use of medical equipment. Perform state feature analysis and processing on historical abnormal operation data according to the characteristic data of medical equipment operation state, extract the characteristic pattern of historical abnormal operation, and generate abnormal operation state characteristic data to help identify abnormal characteristics, provide empirical support for the abnormal detection capability of the current monitoring system, and significantly improve the accuracy and efficiency of abnormal analysis. Perform correlation analysis on historical abnormal operation state characteristic data of medical equipment to determine the degree of correlation between abnormal operation state and other operation variables, reveal potential factors affecting the occurrence of abnormalities, and locate the root cause of abnormalities. Perform feature correlation analysis based on abnormal operation state feature correlation data to effectively explore the correlation and regularity of abnormal operation of medical equipment, and intuitively display the overall structure and distribution of abnormal state of equipment by designing abnormal operation state feature correlation map. Time series data analysis is performed on historical abnormal operating status characteristic data of medical equipment to capture the dynamic characteristics of abnormal status changing over time, provide basic information for predicting abnormal trends, reveal the temporal correlation and migration path between abnormal status, and provide a theoretical basis for establishing an effective abnormal intervention mechanism.The abnormal operation state feature association data is combined with the abnormal state migration relationship data, and transmitted to the abnormal operation state feature association map for global decision factor analysis, which comprehensively evaluates the triggering mechanism and influencing factors of abnormal operation. The generated abnormal operation global decision factor covers the multi-dimensional factors of abnormal occurrence, providing a highly reliable decision basis for abnormal prediction and early warning. The conditional distribution probability calculation of the abnormal operation global decision factor can clarify the probability condition distribution that affects the abnormal triggering, quantify the influence of different factors on the abnormal triggering, provide support for building a more accurate abnormal prediction model, and improve the scientificity and interpretability of anomaly detection. Causal contribution analysis based on the abnormal operation trigger condition distribution probability data can reveal the contribution of different conditional factors to abnormal operation and improve the depth of understanding of the abnormal triggering mechanism. The decision tree algorithm is used to design the tree nodes of abnormal operation evaluation, and the preliminary decision tree is established in combination with the abnormal operation state feature association data and migration relationship data, which intuitively displays the decision path of abnormal triggering, comprehensively integrates multi-dimensional data features, and improves the accuracy and efficiency of abnormal operation evaluation. By fitting and analyzing the global decision factors of abnormal operation and designing decision tree pruning rules, the structure of the decision tree model can be further optimized and redundant nodes can be reduced. The introduction of pruning rules significantly improves the generalization ability of the model and reduces the risk of overfitting, so that the decision tree model can more accurately reflect the key decision path of abnormal triggering. The evaluation decision tree model is trained and optimized using the abnormal operation triggering causal contribution data to generate an optimized decision tree model, which further enhances the model's ability to identify the causal relationship of abnormal triggering. The optimized model has higher prediction accuracy and adaptability, and realizes comprehensive abnormal monitoring and intelligent operation management. The characteristic data of the operation status of medical equipment is input into the abnormal operation evaluation optimization decision tree model, and the abnormal operation trend of medical equipment is evaluated and processed. The abnormal operation trend evaluation data containing perception topology nodes, trend time nodes and fault risk factors are generated, which fully reflects the potential risks and trend changes in the operation of the equipment, and provides a basis for predicting the time window and key nodes of abnormal occurrence. Combined with the perception topology nodes, trend time nodes and fault response feedback data, the established automated feedback engine realizes the intelligent management of the entire process from abnormal detection to fault response, and improves the timeliness and automation level of fault handling. Efficiently monitor and handle abnormal situations to ensure that the equipment can quickly resume normal operation when a failure occurs. The application of the intelligent feedback engine can reduce the risk of human intervention and misoperation and provide a more reliable response mechanism.

[0007] Therefore, the medical equipment monitoring method based on artificial intelligence of the present invention establishes the operation perception topology structure between devices through the analysis method based on the Internet of Things communication interface and perception topology data, identifies the complex correlation between multiple devices, and analyzes the state feature correlation and state migration relationship of the abnormal operation of medical equipment to comprehensively evaluate the global decision factors of abnormal operation, realizes a comprehensive analysis of equipment abnormalities and efficient identification and prediction of potential risks, and accurately captures the trend of abnormal operation of equipment. Through the calculation of the probability of abnormal trigger condition distribution and causal contribution analysis, the trigger mechanism and causal relationship of abnormal operation of equipment are deeply explored, and the accurate identification of the cause of abnormal operation is provided. In addition, the multi-scale specificity and dependency analysis of medical equipment state variables and the analysis of operation state characteristics realize the ability to extract key state variables and features from complex operation data, and improve the accuracy of medical equipment monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A schematic diagram of the steps of a medical device monitoring method based on artificial intelligence of the present invention; Figure 2 for Figure 1 Detailed implementation steps of step S4 in FIG. The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0009] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0010] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities are implemented in software form, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0011] It should be understood that, although the terms "first", "second", etc. are used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit is referred to as the second unit, and similarly the second unit is referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0012] To achieve this, please refer to Figure 1 to Figure 2 The present invention provides a medical device monitoring method based on artificial intelligence. In the embodiments of the present invention, please refer to Figure 1 FIG. 1 is a schematic diagram of a process flow of a medical device monitoring method based on artificial intelligence according to the present invention, wherein the medical device monitoring method based on artificial intelligence comprises the following steps: Step S1: performing a medical device interconnection operation on the medical device using a preset Internet of Things communication interface, and performing medical device operation monitoring processing on the medical device after the medical device interconnection operation to obtain medical device operation monitoring data; performing medical device operation perception topology data analysis on the medical device operation monitoring data to generate medical device operation perception topology data; In an embodiment of the present invention, the medical device is connected to the Internet of Things communication interface through an embedded hardware communication module to realize the interconnection operation of the device. The Internet of Things communication interface module supports standardized communication protocols, such as the MQTT protocol or the CoAP protocol, to ensure efficient communication between the device and the Internet of Things system. The interconnection operation first performs two-way authentication with the Internet of Things system through the unique identifier of the medical device (such as the MAC address or serial number). The authentication process includes device identity authentication and network communication authority verification to ensure that the device is legally connected to the system. The interconnection status of the device and the Internet of Things is recorded in the device management database, including the physical address of the device, the communication protocol type, and the real-time connection status information. The operation monitoring of medical equipment depends on a variety of sensors and data acquisition modules built into the device, such as temperature sensors, current sensors, pressure sensors, etc. During the operation monitoring process, the device sensors collect the operation status parameters in real time and transmit them to the cloud monitoring platform through the Internet of Things communication interface. After the medical device operation monitoring data is uploaded to the cloud, the monitoring platform identifies the sensing node according to the source address, device code, and other information in the data packet. After the identification is completed, the sensing node is mapped to a logical node in the monitoring system, which is convenient for subsequent data management and topology construction. By analyzing the operation monitoring data and communication interaction relationship of the sensing nodes, the medical equipment operation perception topology is constructed. The implementation logic includes aggregating and analyzing the operation data of the sensing nodes (such as transmission rate, interaction frequency, data content, etc.) to generate the topological association relationship between the devices. For example, in the imaging equipment network of a medical center, the analysis found that the data interaction frequency between the CT machine and the MRI machine was the highest (an average of 100 times per second), and the interactive data packet content between the cooling system and the CT machine contained multiple temperature alarm parameters, indicating that the cooling system has a direct impact on the normal operation of the CT machine. According to the analysis results, the CT machine and the MRI machine are identified as closely related nodes, and the cooling system and the CT machine are identified as high-dependency nodes. The attributes of each sensing node include the real-time operation status of the device (such as temperature, pressure, current), data communication status (such as bandwidth, delay), historical performance indicators, etc. Through these attributes, the topology structure reflects the operation status of the medical equipment and its association in real time. The perception topology data is updated in real time on the topology map, showing the logical structure and status changes between the devices, which is used to assist the monitoring decision of the system.

[0013] Step S2: performing a state characteristic analysis of the operation of the medical device based on the medical device operation perception topology data to generate medical device operation state characteristic data; In an embodiment of the present invention, the medical equipment operation perception topology data includes equipment nodes and their associated information. The operation state variables of each equipment are extracted by analyzing the data, including real-time operation parameters, key performance indicators (such as temperature, power consumption, load), equipment interaction traffic (such as transmission rate, delay) and historical performance records. The extracted state variables are classified and analyzed according to the equipment category and operation mode, for example, imaging equipment and surgical equipment are grouped and processed, and subdivided into startup, stable operation, idle and other modes. Combined with the equipment operation logic model, the correlation between variables is analyzed, such as the positive correlation between CT machine temperature and radiation dose, or the negative correlation between network delay and image transmission rate, to form a structured state variable data table. The state variable data is screened and analyzed to extract the key variables that have the most significant impact on the equipment operation state. The identification process of key variables includes causal analysis and statistical modeling, and the weight of the impact of each variable on equipment performance or failure risk is evaluated by regression analysis or correlation calculation. The key variable data is combined with the operation perception topology data to analyze the global operation state characteristics of the equipment. The key variables are mapped to the perception topology structure to form the attributes of the topological nodes. For example, the key variables of the CT machine, radiation dose and temperature, are set as the attributes of the topological node CT-001, and its associated path is analyzed to confirm the causal relationship between whether the increase in radiation dose leads to the increase in temperature. The characteristic pattern of the operating state is extracted through feature analysis algorithms, such as cluster analysis and pattern matching. The characteristic pattern analysis shows that the CT machine is prone to trigger an operating mode with increased network delay under high radiation dose (over 5mGy / s) and high temperature (over 50°C), which is classified as a high-risk operating characteristic mode. The generated characteristic data is recorded in a structured form, for example: the identification is CT-001, the characteristic mode is "high temperature and high radiation operation", the associated variables are radiation dose and temperature, and the risk level is high. These characteristic data are used for subsequent anomaly detection and system optimization decisions.

[0014] Step S3: Acquire historical medical equipment abnormal operation data; perform historical medical equipment abnormal operation state characteristic analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state characteristic data to generate historical medical equipment abnormal operation state characteristic data; perform global decision factor analysis of medical equipment abnormal operation based on the historical medical equipment abnormal operation state characteristic data to generate abnormal operation global decision factor; In an embodiment of the present invention, the log storage function in the medical equipment operation management system is used to extract abnormal operation data recorded in the past period of time, and the abnormal event data is filtered from the equipment operation log according to the identification of the abnormal event, and the filtered abnormal event data is integrated into a historical data set in chronological order, including information such as abnormal type, occurrence time, abnormal value range and impact degree. The generated historical medical equipment abnormal operation data is time series data, including information such as the type, variable value and time of the abnormal event. By comparing the historical abnormal operation data with the medical equipment operation state feature data, the state characteristics of the abnormal operation are extracted, and the abnormal operation data are divided into multiple state feature types according to the normal range defined by the operation state feature data. The key variable values ​​in the historical abnormal operation data, such as the power supply voltage, equipment temperature and radiation dose, are extracted, and the abnormal distribution range of each variable is counted. Based on the time series analysis technology, the change law of the abnormal operation state is extracted. The generated historical medical equipment abnormal operation state feature data includes abnormal state classification, feature value range and time series feature information. The historical medical equipment abnormal operation state feature data is used to analyze the correlation between different abnormal state features, and the Pearson correlation coefficient or mutual information algorithm is used to calculate the correlation between different features. Cluster abnormal state features with strong correlation into the same category. For example, cluster abnormal power supply voltage, abnormal temperature and abnormal radiation dose into the electrical abnormality category. Based on the results of correlation analysis, use causal inference method to determine the causal relationship between features. For example, the causal effect strength of temperature abnormality caused by power supply voltage fluctuation is 0.92. The generated abnormal operation state feature correlation data includes correlation coefficient matrix, feature clustering results and causal relationship description. Establish an association map of abnormal state features based on state feature correlation data, take abnormal state features as map nodes, and assign node attribute values. For example, take abnormal power supply voltage, abnormal temperature and abnormal radiation dose of CT machine as nodes respectively, and attach abnormal range values. Establish association relationships between nodes based on correlation and causal analysis results, and assign weight values ​​to edges. For example, the edge weight between abnormal power supply voltage and abnormal temperature is 0.85, and the edge weight between abnormal power supply voltage and abnormal radiation dose is 0.78. Design the visualization structure of abnormal operation state feature association map, including node size, color and edge thickness. Node size indicates the degree of abnormal impact, and edge thickness indicates the strength of correlation. The generated abnormal operation status feature association data includes graph nodes, edge relationships and weight information. The association graph presents the complex associations between abnormal features of medical equipment in a graphical form. Time series data analysis is performed based on historical abnormal operation status feature data of medical equipment, and time series processing is performed on historical abnormal operation status feature data of medical equipment. Assuming that the abnormal state feature data of a certain device includes abnormal values ​​and timestamps of variables such as temperature, humidity, and voltage, these data are sorted in chronological order to form time series data.Apply time series analysis methods, such as the autoregressive moving average (ARMA) model or LSTM neural network, to extract potential time series patterns from historical time series data and identify regular fluctuations of variables such as temperature and power supply voltage. For example, the temperature data of the CT machine continues to rise within a specific time period, indicating that the equipment overheats and causes a fault. Perform fluctuation analysis on the time series data of each variable to calculate the amplitude and duration of abnormal fluctuations. The generated time series abnormal operation state feature data includes the time series changes of the abnormal state characteristics of each device, marking the abnormal change points at key moments and their duration. Use the time series abnormal operation state feature data to perform migration relationship analysis, and establish a state transition matrix by analyzing the abnormal states of the equipment at different time points. For example, the temperature of the CT machine is abnormal at time t1 and returns to normal at time t2. If it changes from an abnormal state to a normal state, a state transition matrix is ​​constructed to record the transition probability from abnormal to normal. According to historical time series data, identify the migration pattern between different abnormal states. For example, the abnormal temperature state of the CT machine undergoes a change process of high temperature, continuous high temperature, and recovery to normal, and there is a certain transition probability between each state. Through the analysis of time series data, the transition law and probability between abnormal states are obtained. The migration relationship between abnormal states is analyzed by using Markov chain or migration matrix model. The abnormal operation state migration relationship data obtained includes the transition rules and transition probabilities between different abnormal states in equipment operation, which provides a key basis for subsequent fault prediction and risk assessment. The migration relationship data is combined with the state feature association map to expand the map node of the global decision factor analysis. Based on the abnormal operation state feature association data and the abnormal state migration relationship data, the abnormal state migration relationship data is used as the extension node of the map node. For example, the association relationship between high temperature abnormality and voltage abnormality and the state migration information (such as temperature abnormality, high temperature abnormality and normal) are added to the map node extension part. The global decision factor analysis is performed using the association relationship between each node in the map and its extension node. For example, the potential high-risk fault mode is analyzed through the map by combining the state changes and migration relationships of variables such as temperature, humidity, and voltage. In the map, the migration probability, risk level and fault influencing factors of each state are marked to generate a comprehensive abnormal operation global decision factor.

[0015] Step S4: Calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor to generate abnormal operation triggering condition distribution probability data; perform abnormal operation triggering causal contribution analysis based on the abnormal operation triggering condition distribution probability data to generate abnormal operation triggering causal contribution data; construct an abnormal operation evaluation decision tree model based on the abnormal operation global decision factor; use the abnormal operation triggering causal contribution data to perform training optimization processing on the abnormal operation evaluation decision tree model for the abnormal operation triggering causal contribution of the model to generate an abnormal operation evaluation optimization decision tree model; In an embodiment of the present invention, based on the historical operation data and global decision factors of the medical device, various trigger condition data that cause abnormal operation are collected. It is assumed that the abnormal operation of the medical device is jointly affected by multiple factors, involving device state variables (such as temperature, voltage, humidity, etc.), and the abnormal value of each variable is used as a trigger condition. According to the historical data and the probability model, the probability distribution under each trigger condition is generated. The calculated probability data of the trigger condition distribution is used to perform causal contribution analysis, and a causal reasoning model is constructed based on the probability data of the trigger condition. A causal reasoning method (such as Granger causality test or a causal model based on a Bayesian network) is used to identify the causal relationship between each trigger condition and the abnormal operation of the device. Assuming that the conditions such as device temperature, humidity, and voltage have different contributions to the abnormal operation state, analyze how these factors cause device failure. Using statistical analysis technology, based on the trigger condition probability data and the device failure data, the causal contribution of each trigger condition to the device failure is calculated. According to the analysis results, the contribution data of each trigger condition to the abnormal operation state is generated. The causal relationship between different trigger conditions and the abnormal operation of the device is described, which provides a key basis for the establishment of the subsequent decision tree. According to the collected global decision factors, the decision tree algorithm is used to design nodes for these factors. The root node of the decision tree usually selects the most critical factor, and the tree nodes are designed through multiple decision factors to form a multi-level decision tree. The abnormal operation state feature association data is integrated into the decision tree as the combination factor of each tree node. The abnormal operation state feature association data includes the characteristic values ​​of the equipment under different operation states (such as temperature, humidity, voltage, etc.) and their correlation with the occurrence of faults. The abnormal operation state transition relationship data is used as the expansion factor of the tree node. Assume that the fault state of the equipment does not occur immediately, but undergoes a migration process of multiple states. For example, the equipment will experience the "overheating" state, then the "overtemperature" state, and finally enter the fault state under high temperature. Therefore, these migration relationships also need to be represented by decision tree nodes. Under each tree node, add conditions related to other state transitions. For example, under the high temperature node, further determine whether there are state transition factors such as voltage drop. Each state transition will affect the probability of the final fault. According to the combination factors and expansion factors of the tree nodes, a preliminary decision tree for abnormal operation evaluation is established. Assume that factors such as temperature, voltage, and humidity constitute the root node and child node structure of the tree, and the branches of each node represent different equipment operating states (such as whether the temperature exceeds the standard, whether the voltage fluctuates, etc.). According to the branch conditions and associated data of the tree nodes, the decision tree algorithm (such as ID3, CART, etc.) is used to generate the tree. Through the step-by-step judgment of these conditions and factors, a decision tree that can evaluate the abnormal operation of the equipment is finally formed, and a preliminary decision tree model is obtained. The model determines whether the equipment is abnormal and provides accurate evaluation results.On the basis of the existing preliminary decision tree model for abnormal operation evaluation, multiple global decision factors (such as temperature, voltage, humidity, etc.) related to the operation status of medical equipment are collected and analyzed. For example, based on the historical data of equipment operation and abnormal failure data, these decision factors are fitted using statistical methods such as regression analysis to obtain fitted decision factors. For each decision factor, linear or nonlinear fitting analysis is performed to generate the fitted decision factor. Combined with the data of equipment operation, the contribution value of each decision factor to the equipment failure is calculated, and the abnormal operation fitting decision factor is generated. According to the results of the analysis of the fitting decision factor, the pruning rules of the decision tree are designed. The design basis of the pruning rule is: when the branches of some decision factors contribute very little to the final decision in the model, or fail to effectively improve the prediction accuracy in multiple decisions, these branches need to be pruned. For example, in the branch judgment of the temperature factor, if the fitting results show that the temperature has little effect on the failure, and the nodes in some temperature ranges do not provide effective information, the pruning rules are set to delete these nodes from the decision tree, simplify the tree structure, and avoid overfitting. When designing the pruning rules, consider the interactive effects of multiple factors and set the pruning threshold. For example, if the intersection of the two factors of temperature and humidity cannot effectively distinguish the probability of failure in multiple states, then the design rules are used for pruning to improve the operation efficiency and interpretability of the decision tree. Using the designed pruning rules, combined with the preliminary decision tree model for abnormal operation evaluation, pruning is performed on each tree node. Assuming that some branches in the preliminary decision tree model (such as temperature overload and voltage fluctuation) fail to significantly improve the prediction accuracy of the model, the pruning rules are applied to delete these branches from the decision tree. The pruning operation starts from the bottom node and is performed step by step upward. For each decision factor node, the influence of its child nodes is analyzed. If the contribution of some child nodes is too low (for example, the equipment failure rate does not change much in different temperature ranges), these nodes are pruned. The generated abnormal operation evaluation decision tree model is simpler, more accurate, and more computationally efficient than the preliminary model. The model can effectively predict the abnormal state of medical equipment and provide maintenance personnel with more reliable fault warnings. The abnormal operation trigger causal contribution data is used to train and optimize the decision tree model. The causal contribution data is obtained through historical data analysis and indicates how certain specific events (such as equipment overtemperature, fault alarm, etc.) trigger abnormal operation states. During the training process, the causal contribution data is matched with the historical operation data of the equipment, and the training model learns the contribution value of each decision node to the abnormal state trigger. Through optimized training, the decision tree model can not only predict abnormal operation based on the current state, but also optimize the judgment based on the triggering causal relationship of the equipment history. The final abnormal operation evaluation optimization decision tree model has higher prediction ability than the preliminary decision tree model, and can more finely reflect the fault triggering mechanism of the equipment under different states. This model can monitor the equipment operation status in real time and provide accurate fault prediction and early warning.

[0016] Step S5: Transmitting the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform abnormal operation trend evaluation processing of the medical equipment, and generate abnormal operation trend evaluation data of the medical equipment; establishing an automated feedback engine for medical equipment fault response based on the abnormal operation trend evaluation data of the medical equipment, and generating a medical equipment fault response intelligent feedback engine; executing the medical equipment monitoring fault response feedback task through the medical equipment fault response intelligent feedback engine.

[0017] In an embodiment of the present invention, the characteristic data of the operation status of the medical device (such as real-time sensor data such as temperature, voltage, pressure, etc.) is transmitted to the equipment monitoring platform through the Internet of Things interface and input into the optimized decision tree model for processing. The model evaluates the trend of the equipment operation status based on historical data and abnormal operation mode, and generates abnormal operation trend evaluation data, including abnormal operation perception topology nodes, trend time nodes and fault risk factors. For example, the perception topology node identifies key abnormal areas, such as components with excessively high temperatures; the trend time node predicts the time of occurrence of faults, such as equipment failures that may occur within the next 30 minutes; the fault risk factor quantifies the impact of each variable on the fault, such as the contribution of temperature increase to the risk of faults is 40%. According to the evaluation results, the fault modes caused by different fault risk factors are further analyzed, and the fault mode data is generated by combining historical data and real-time status, such as high temperature causing circuit damage, and the risk factors are quantified to determine the weight of each factor. These data are integrated into fault response feedback data for establishing an intelligent feedback engine. Combined with the perception topology node, trend time node and feedback data, an intelligent feedback engine is designed to achieve automated fault response. When an anomaly is detected (such as voltage overload or temperature rise), the feedback engine triggers emergency operations, such as starting the cooling system, reducing the load, or sending an alarm message to maintenance personnel. The system can determine the appropriate response measures based on the real-time status, start the automatic diagnosis and repair procedures of the equipment, or remotely control the equipment to perform protective operations. For example, when the voltage is abnormal, the feedback engine can trigger the power protection mechanism, automatically cut off the power supply and notify the maintenance personnel. After the fault response operation is completed, the equipment resumes normal operation and the system continues to monitor to ensure safe operation and reduce the risk of failure.

[0018] Further, step S1 includes the following steps: Step S11: performing a medical device connection operation on the medical device using a preset Internet of Things communication interface, and performing a medical device operation monitoring process on the medical device after the medical device connection operation to obtain medical device operation monitoring data; In an embodiment of the present invention, the Internet of Things communication interface module monitors the operation of the medical device in real time through a sensor connected to the Internet of Things. The medical device is connected to the Internet of Things communication interface through an embedded hardware communication module. The interface is implemented by a standardized protocol stack, such as the MQTT protocol or the CoAP protocol, to ensure efficient communication between the device and the Internet of Things system. The medical device communication operation performs two-way authentication with the Internet of Things system through a pre-configured device unique identifier (such as a MAC address or a serial number), and a stable communication connection is established between the device and the monitoring platform after the authentication is completed. The implementation logic of the operation monitoring process includes real-time acquisition of operation status data from the built-in sensors and data acquisition modules of the medical device, such as parameters such as device temperature, pressure, speed, and current. The acquired operation monitoring data is transmitted to the cloud monitoring platform through the communication module and stored in a distributed database for subsequent analysis. For example, when a CT machine is running, the power current data collected by its sensor is transmitted to the cloud in real time and used to dynamically monitor the device status.

[0019] Step S12: performing monitoring sensing node identification according to the Internet of Things communication interface corresponding to the medical equipment operation monitoring data, and generating an identified monitoring sensing node; In an embodiment of the present invention, after the root medical device operation monitoring data is uploaded through the Internet of Things communication interface, the monitoring platform identifies the perception node based on the device identifier and the communication interface address in the communication data packet. The implementation logic includes parsing the source address, destination address, device code and other meta-information in the data packet, and comparing it with the registration information in the device database to confirm the uniqueness of the perception node. For example, after parsing the uploaded CT machine data packet, it is found that its source address is 192.168.1.10 and the device code is CT12345, then it is identified as node CT12345, and the network properties of its perception node are recorded, including the communication protocol type, bandwidth occupancy, etc. After the identification is completed, the perception node is mapped to a logical node in the monitoring system, which is convenient for subsequent data management and analysis.

[0020] Step S13: Perform medical equipment operation perception topology data analysis on the medical equipment operation monitoring data based on the identified monitoring perception node to generate medical equipment operation perception topology data.

[0021] In an embodiment of the present invention, the operation monitoring data of the identification sensing node is aggregated and analyzed to construct the operation sensing topology of the medical device. By analyzing the data interaction frequency, data content and physical connection relationship of the sensing node, a topological relationship diagram between the devices is generated. For example, among the imaging devices (CT machine, MRI machine, ultrasound equipment) in a medical center, the image data interaction of the CT machine and the MRI machine is most frequent through the Internet of Things network, so these two devices are set as close nodes. The attributes of each sensing node include operating parameters, data communication status and historical performance indicators. For example, the attributes of the CT machine node include parameters such as temperature and radiation dose during its operation. According to the changes in the operation data of the medical equipment, the sensing topology is updated in real time. For example, when a device exits due to a fault, its corresponding node is marked as "unavailable" and its topological association relationship is adjusted. The operation sensing topology data is obtained through comprehensive analysis to describe the logical structure and real-time operation status between medical devices. For example, the operation sensing topology diagram of the CT machine and the MRI machine in a medical center shows that the data upload rate of the CT machine has dropped to 50KB / s, which is 30% lower than its normal state. The topological structure indicates that there is a problem with the communication module, triggering further monitoring or maintenance decisions.

[0022] Further, step S13 includes the following steps: Step S131: performing sensor node parsing processing on the identification monitoring sensor node to generate identification monitoring sensor node parsing data; In an embodiment of the present invention, an identified monitoring perception node data packet is obtained from a monitoring system, and the meta-information contained therein, such as a unique device identifier, operating status attributes, IoT interface address, and communication protocol type, is decoded. A data parsing module is used to analyze the data packet layer by layer, extract important fields, and normalize the field values. For example, the perception node data of a CT machine is parsed, and the extracted fields include the device identifier CT-001, the status attribute running, the interface address 192.168.0.2, and the communication protocol MQTT. The parsed fields are organized into structured records, stored as a record in a database, and parsed data for identifying the monitoring perception node is generated. For example, the parsing result is recorded as: the node ID is CT-001, the current status is running, and the device category is an imaging device.

[0023] Step S132: using the identification monitoring perception node parsing data to perform heterogeneous monitoring data synchronization processing on the medical equipment operation monitoring data to generate synchronized medical equipment operation monitoring data; In an embodiment of the present invention, the synchronization processing of heterogeneous data is completed by coordinating the timestamps, data formats and parameter definitions of the operation data of medical devices from different sources through parsing data. According to the equipment category and the attributes contained in the parsed data, the field name, data unit and encoding method of the monitoring data are converted. For example, the radiation dose of CT machine data from different manufacturers is expressed in mGy or μGy, which needs to be uniformly converted to mGy. The time reference in the sensor node parsed data is used to adjust the synchronization timestamp of the data uploaded by each sensor node. For example, if the sensor node data of the CT machine and the MRI machine use different time references, the timestamps of both need to be re-marked based on the unified server time to ensure that all data are synchronized. Abnormal values ​​or redundant data are eliminated and missing information is filled. For example, if a data packet of the CT machine is lost due to network interruption, the interpolation algorithm is used to fill the missing part according to the data at the upper and lower times. After the synchronization processing is completed, a medical equipment operation monitoring data record in a unified format is generated, for example, the radiation dose of the CT machine at a certain moment and the magnetic field strength of the MRI machine are recorded in the same time series.

[0024] Step S133: Establish a medical equipment perception topology structure by identifying monitoring perception nodes, transmit synchronized medical equipment operation monitoring data to the medical equipment perception topology structure to perform medical equipment operation perception topology data analysis, and generate medical equipment operation perception topology data.

[0025] In an embodiment of the present invention, a node model is created based on the unique identifier and network connection relationship of the perception node in the parsed data, and the logical association relationship between the nodes is defined. For example, a real-time data connection is established between the CT machine node CT-001 and the image server IS-002 through the Internet of Things interface 192.168.0.2, and the initial topological connection description generated by the node relationship diagram is CT-001 -> IS-002. The synchronized operation monitoring data is allocated to the perception topology structure according to the node model definition. For example, the real-time radiation dose and temperature monitoring data uploaded by the CT machine are updated to the topology node CT-001 in real time through the network interface, and data is associated with the image server node. Based on the data flow of all nodes in the topology structure, the operation status is evaluated using a graph analysis algorithm. The specific analysis includes communication delay analysis between nodes, data flow consistency check, abnormal state warning, etc. For example, if the data flow delay from CT-001 to IS-002 in the topology diagram exceeds a preset threshold, a performance alarm is issued, and the problem node and its associated impact range are recorded. After completing the above processing, the generated medical equipment operation perception topology data not only describes the real-time status of the equipment operation, but also provides the correlation characteristics between the equipment and the network performance. For example, a topology analysis result shows that the data exchange volume between the CT machine CT-001 and the MRI machine MRI-003 has dropped sharply, indicating a potential network failure.

[0026] Further, step S2 includes the following steps: Step S21: performing medical device state variable analysis according to the medical device operation perception topology data to generate medical device state variable data; In an embodiment of the present invention, the node and associated information in the medical equipment operation perception topology data is used to perform multi-dimensional variable extraction and analysis operations on the operation status of the medical equipment. The operation parameters of each medical device are extracted from the perception topology data, including the real-time operation status of the device, key performance indicators (such as temperature, power consumption, load), device interaction traffic (such as data packet transmission rate, delay) and historical performance records. For example, for a CT machine, the extracted state variables include the current radiation dose of 3mGy / s, the device temperature of 45°C, the network transmission rate of 50Mbps, etc. The extracted state variables are classified according to the device category and operation mode. For example, the state variables of the imaging device and the state variables of the surgical auxiliary device are processed separately; the operation mode is divided into startup, stable operation, idle, shutdown, etc. The correlation between variables is analyzed according to the device operation logic model, such as whether the increase in CT machine temperature is positively correlated with the increase in radiation dose, or whether the increase in network delay leads to a decrease in image transmission speed. After completing the state variable analysis, a medical device state variable data table including device identification, variable category, variable value and associated characteristics is generated. For example, the state variable data of a CT machine is recorded as follows: identification CT-001, radiation dose 3mGy / s, temperature 45°C, and stable operation mode.

[0027] Step S22: performing key variable analysis on the medical device state variable data to generate medical device state key variable data; In an embodiment of the present invention, by screening and analyzing the state variable data, the key variables that have the greatest impact on the operating state of the medical device are determined. The causal analysis and statistical modeling methods are used to evaluate the degree of influence of each state variable on the equipment performance or failure risk, and the influence weight of the variable is calculated. For example, through regression analysis, it is found that the radiation dose and temperature variables of the CT machine have a significant impact on the stability of the equipment, and are assigned weight values ​​of 0.7 and 0.6 respectively, while the impact of network delay is small, and the weight value is 0.3. Set the screening rules, such as selecting the top N variables according to the weight value sorting, or screening out significant variables according to the actual degree of influence. For example, for the CT machine, the key variables screened out are radiation dose, temperature and network transmission rate. The screened key variables are bound and stored in a marked form with the equipment identification to form structured key variable data. For example, the record is: identification CT-001, the key variables include radiation dose 3mGy / s, temperature 45°C. After the analysis is completed, the key variable data of the medical device status is generated for subsequent status feature analysis.

[0028] Step S23: Performing a state characteristic analysis of the medical device operation on the medical device operation perception topology data according to the medical device state key variable data to generate medical device operation state characteristic data.

[0029] In an embodiment of the present invention, the global state characteristics of the equipment operation are analyzed by combining key variables with the perception topology data. The selected key variables are mapped to the perception topology structure to form a topological map of the key variables. For example, the key variables radiation dose and temperature of the CT machine are respectively used as the attributes of the topological nodes, and the association relationship between the nodes is established to analyze whether the conduction path of excessive radiation dose leads to temperature increase. The characteristic pattern of the operation state is mined through a feature analysis algorithm (such as cluster analysis and pattern matching). For example, under high radiation dose and high temperature conditions, the CT machine is prone to trigger an operation mode with increased network delay, and the analysis result is a high-risk operation characteristic mode. The feature analysis results are converted into structured feature data, including equipment identification, state feature pattern, associated variables and risk level. For example, the generated feature data record is: identification CT-001, feature pattern is high temperature and high radiation operation, associated variables are radiation dose and temperature, and the risk level is high. After completing the state feature analysis, the medical equipment operation state feature data is used to support subsequent abnormality detection and decision analysis.

[0030] Further, step S22 includes the following steps: Step S221: performing medical equipment operation condition analysis of each topology data according to the medical equipment operation perception topology data, and generating medical equipment operation condition data; In an embodiment of the present invention, by analyzing the device nodes and their connection relationships in the medical device operation perception topology data, the condition data required for the device operation is extracted, and based on the node attribute information in the perception topology data, the external environment conditions for the operation of the medical device, such as power supply stability, network bandwidth adequacy, etc., are analyzed. Taking the MRI device as an example, it is analyzed whether the power supply voltage is maintained at 220±10V and whether the network bandwidth reaches 100Mbps to ensure real-time image transmission. The operation dependency between devices is analyzed from the edge relationship information in the topological structure, such as whether the data interaction frequency between the surgical robot and the imaging device meets the real-time requirements. Specifically analyze whether the delay time of the surgical robot control signal is less than 10ms. Combined with the real-time state variables recorded in the perception topology, it is judged whether the device operation parameters are within the safe range, such as whether the radiation dose of the CT machine is less than 5mGy / s, and whether the temperature is between 35-50°C. After the analysis is completed, the generated medical device operation condition data includes operating environment conditions, equipment interaction conditions and operating parameter range information, which are recorded as: device identification MRI-001, power supply voltage 220V, network bandwidth 120Mbps, and operating temperature range 38°C.

[0031] Step S222: performing medical device state variable-specific analysis processing of the medical device operating condition on the medical device state variable data based on the medical device operating condition data to generate medical device state variable-specific data; In an embodiment of the present invention, by comparing the medical equipment operating condition data with the state variable data, specific data with special significance in the state variable is identified, and based on the operating condition data, data points in the state variable that deviate from the normal range are detected. For example, the operating temperature condition of the CT machine is 35-50°C. If it is recorded as 55°C in the state variable, it is marked as specific variable data. Analyze whether the specific variable violates the key constraints of the equipment operation. For example, the power supply voltage of the MRI equipment is lower than 210V, resulting in unstable magnetic field strength, and the voltage variable is recorded as specific data. Use pattern matching technology to analyze whether the change law of the variable meets the equipment operation conditions, such as whether the frequency fluctuation of the surgical robot control signal is within the range of ±5%. Data points that exceed the range are marked as specific variables. After completing the specific analysis, the generated medical equipment state variable specific data includes abnormal point marks, variables that violate conditional constraints, and their characteristic values.

[0032] Step S223: performing multi-scale analysis processing on the medical device state variable specific data to generate medical device state variable multi-scale specific data; performing medical device state variable dependency analysis on the medical device state variable specific data to generate medical device state variable dependency data; In an embodiment of the present invention, statistics and time series analysis of different scales are performed on specific data, and specific data are counted according to time granularity. For example, the frequency of abnormal 205V power supply voltage of MRI equipment is 3 times per hour, and there are 2 abnormal sections lasting more than 10 minutes. In the medical device perception topology, the influence range of specific variables on adjacent nodes is analyzed. For example, the temperature increase of MRI equipment affects the magnetic field stability of adjacent equipment, with a coverage range of 3 meters. Statistics are performed on the distribution of specific variables in different operating modes. For example, the abnormal temperature of CT machine 55°C is concentrated in high load mode, and there is no abnormal record in low load mode. After the analysis is completed, the generated multi-scale specific data of medical device state variables includes time distribution, spatial influence range and mode distribution information. The correlation relationship between specific data and other state variables is used to analyze the dependency between variables, and the conditional correlation between specific variables and other variables is analyzed through the association rule algorithm. For example, the abnormal 205V power supply voltage of MRI equipment is strongly correlated with the increase of 45°C in equipment temperature, with a support of 80% and a confidence of 95%. Through causal inference technology, the direct impact of specific variables on other variables is determined, such as low voltage directly causes MRI magnetic field instability and delay increases by 0.2s. The generated medical device state variable dependency data includes variable association rules and causal relationship information, for example, the record is: low voltage 205V causes MRI magnetic field stability to decrease and delay increases by 0.2s.

[0033] Step S224: Perform key variable analysis on the medical device state variable data according to the medical device state variable multi-scale specificity data and the medical device state variable dependency data to generate medical device state key variable data.

[0034] In an embodiment of the present invention, multi-scale specific data and dependency data are integrated to screen out the key variables that have the greatest impact on the operation of the device, and weights are assigned to each variable in the multi-scale specific data and dependency data. For example, the weight of the MRI power supply voltage is 0.7, and the weight of the temperature variable is 0.6. According to the weight value and dependency relationship of the variable, its importance to the operation of the device is evaluated. For example, low voltage directly causes fluctuations in the magnetic field of the device, and it is determined to be a key variable. The key variables are marked and bound to the device identification, for example, the key variables of the MRI device are marked as power supply voltage and temperature. The generated key variable data of the medical device status includes the identification, weight and impact description of the key variables.

[0035] Further, step S3 includes the following steps: Step S31: Acquire historical abnormal operation data of medical equipment; In an embodiment of the present invention, the log storage function in the medical equipment operation management system is used to extract abnormal operation data recorded in the past period of time, and filter them from the equipment operation log according to the identification of abnormal events. For example, the abnormal operation data in the past six months are extracted according to the abnormal identification code E101 of the CT machine. The screened abnormal event data are integrated into a historical data set in chronological order, including information such as abnormal type, occurrence time, abnormal value range and impact degree. For example, the abnormal operation data of the CT machine is recorded, including temperature abnormality of 60°C, power supply voltage abnormality of 200V, and the occurrence time and duration of the abnormal event. The extracted data is formatted in a unified manner to ensure that various types of abnormal data are stored in a unified structure. The generated historical medical equipment abnormal operation data is time series data, including information such as the type, variable value and time of the abnormal event.

[0036] Step S32: performing historical abnormal operation state characteristic analysis on the historical abnormal operation state data of medical equipment according to the medical equipment operation state characteristic data, and generating historical abnormal operation state characteristic data of medical equipment; In an embodiment of the present invention, by comparing historical abnormal operation data with the operation status characteristic data of the medical equipment, the state characteristics of the abnormal operation are extracted, and the abnormal operation data are divided into multiple state characteristic types according to the normal range defined by the operation status characteristic data. For example, the abnormal temperature data of the CT machine is divided into two state characteristic types: too high and too low. More than 50°C is defined as too high, and less than 30°C is defined as too low. The key variable values ​​in the historical abnormal operation data, such as the power supply voltage, equipment temperature and radiation dose, are extracted, and the abnormal distribution range of each variable is counted. Based on the time series analysis technology, the change law of the abnormal operation state is extracted. For example, the abnormal fluctuation frequency and amplitude of the power supply voltage of the CT machine are analyzed to obtain the time series characteristics of the abnormal state: the voltage fluctuates by more than 5V per hour, and the duration is 15 minutes. The generated historical abnormal operation state characteristic data of medical equipment includes abnormal state classification, characteristic value range and time series characteristic information.

[0037] Step S33: performing abnormal operation state characteristic correlation analysis based on historical abnormal operation state characteristic data of medical equipment to generate abnormal operation state characteristic correlation data; In an embodiment of the present invention, the historical abnormal operation state feature data of medical equipment is used to analyze the correlation between different abnormal state features, and the Pearson correlation coefficient or mutual information algorithm is used to calculate the correlation between different features. For example, the correlation between the abnormal temperature and the abnormal power supply voltage of the CT machine is analyzed, and the correlation coefficient is 0.85, indicating a strong correlation. Abnormal state features with strong correlation are clustered into the same category. For example, the power supply voltage abnormality, temperature abnormality and radiation dose abnormality are clustered into the electrical abnormality category. Based on the results of the correlation analysis, the causal relationship between the features is determined using a causal inference method. For example, the causal effect strength of the temperature abnormality caused by the power supply voltage fluctuation is 0.92. The generated abnormal operation state feature correlation data includes a correlation coefficient matrix, feature clustering results and a causal relationship description.

[0038] Step S34: performing abnormal operation state feature correlation analysis according to the abnormal operation state feature correlation data, generating abnormal operation state feature correlation data, and designing an abnormal operation state feature correlation map using the abnormal operation state feature correlation data as a main node; In an embodiment of the present invention, an association map of abnormal state features is established based on state feature correlation data, abnormal state features are used as map nodes, and node attribute values ​​are assigned. For example, the abnormal power supply voltage, temperature abnormality, and radiation dose abnormality of a CT machine are respectively used as nodes, and an abnormal range value is added, such as the abnormal power supply voltage range of 200-210V. According to the results of correlation and causal analysis, an association relationship between nodes is established, and weight values ​​are assigned to edges. For example, the edge weight between abnormal power supply voltage and abnormal temperature is 0.85, and the edge weight between abnormal power supply voltage and abnormal radiation dose is 0.78. A visualization structure of an association map of abnormal operation state features is designed, including node size, color, and edge thickness. The node size indicates the degree of abnormal impact, and the edge thickness indicates the strength of correlation. For example, the abnormal power supply voltage node is larger and the edge line is thicker, indicating that it is the main abnormal feature. The generated abnormal operation state feature association data includes map nodes, edge relationships, and weight information, and the association map presents the complex associations between abnormal features of medical equipment in a graphical form.

[0039] Step S35: performing time series data analysis of abnormal operation state characteristics on historical abnormal operation state characteristic data of medical equipment to generate time series abnormal operation state characteristic data; performing abnormal operation state migration relationship analysis based on the time series abnormal operation state characteristic data to generate abnormal operation state migration relationship data; In an embodiment of the present invention, time series data analysis is performed based on the historical abnormal operation state characteristic data of medical equipment, and the historical abnormal operation state characteristic data of medical equipment is processed in a time series manner. Assuming that the abnormal state characteristic data of a certain device includes abnormal values ​​of variables such as temperature, humidity, and voltage and their timestamps, these data are sorted in chronological order to form time series data. For example, the abnormal fluctuation data of the temperature of the CT machine in each hour in the past three months is recorded: (2024-09-01 12:00, 55°C), (2024-09-01 13:00, 58°C), etc. Time series analysis methods, such as the autoregressive moving average (ARMA) model or the LSTM neural network, are applied to extract potential time series patterns from historical time series data to identify regular fluctuations of variables such as temperature and power supply voltage. For example, the temperature data of the CT machine continues to rise within a specific time period, indicating that the equipment is overheated and causes a fault. Fluctuation analysis is performed on the time series data of each variable to calculate the abnormal fluctuation amplitude and duration. For example, when the fluctuation amplitude of the temperature of the CT machine exceeds 5°C and lasts for more than 10 hours, it is determined to be a major abnormality. By analyzing the abnormal change pattern in the time series data, a basis is provided for subsequent fault prediction. The generated time series abnormal operation state characteristic data includes the time series change of the abnormal state characteristics of each device, marking the abnormal change points and their duration at the key moments. The time series abnormal operation state characteristic data is used to perform migration relationship analysis, and a state transition matrix is ​​established by analyzing the abnormal state of the device at different time points. For example, the temperature of the CT machine is abnormal at time t1 and returns to normal at time t2. If it changes from an abnormal state to a normal state, a state transition matrix is ​​constructed to record the probability of transition from abnormal to normal. According to the historical time series data, the migration pattern between different abnormal states is identified. For example, the abnormal temperature state of the CT machine experiences a change process of high temperature, continuous high temperature, and recovery to normal, and each state has a certain transition probability. Through the analysis of time series data, the transition rules and probabilities between abnormal states are obtained. The migration relationship between abnormal states is analyzed using the Markov chain or migration matrix model. For example, if the abnormal temperature state of the device lasts for more than a certain period of time, it migrates to the complete failure state of the device. This relationship is obtained by analyzing the transition probability in the time series data. The abnormal operation state migration relationship data includes the transition rules and transition probabilities between different abnormal states in equipment operation, which provides a key basis for subsequent fault prediction and risk assessment.

[0040] Step S36: The abnormal operation state migration relationship data corresponding to the abnormal operation state feature association data is transmitted as an extension node of the main node to the abnormal operation state feature association map to perform global decision factor analysis on abnormal operation of the medical device and generate a global decision factor for abnormal operation.

[0041] In an embodiment of the present invention, the migration relationship data is combined with the state feature association map to perform global decision factor analysis map node expansion, and based on the abnormal operation state feature association data, combined with the abnormal state migration relationship data, the abnormal state migration relationship data is used as an extension node of the map node. For example, the association relationship between high temperature anomaly and voltage anomaly and the state migration information (such as temperature abnormality, high temperature abnormality and normal) are added to the extended part of the map node. The global decision factor analysis is performed using the association relationship between each node in the map and its extended nodes. For example, combining the state changes of variables such as temperature, humidity, and voltage and their migration relationships, potential high-risk failure modes are analyzed through the map. In the map, the migration probability, risk level and failure influencing factors of each state are marked, so as to generate a comprehensive abnormal operation global decision factor, which provides a global perspective for the risk assessment of equipment operation.

[0042] Further, as an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S4 in FIG. 1 , in this embodiment, step S4 includes the following steps: Step S41: Calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor to generate abnormal operation triggering condition distribution probability data; In an embodiment of the present invention, various trigger condition data that lead to abnormal operation are collected based on the historical operation data of the medical device and the global decision factor. It is assumed that the abnormal operation of the medical device is jointly affected by multiple factors, involving device state variables (such as temperature, voltage, humidity, etc.), and the abnormal value of each variable is used as a trigger condition. For example, the temperature of the CT machine rises to 70°C, the power supply voltage drops to 180V, etc. Based on the historical data of the device, the probability of abnormal operation under various trigger conditions is calculated. Through statistical analysis, such as using maximum likelihood estimation or Bayesian inference method, the distribution probability of each trigger condition is calculated. For example, if the historical data shows that the probability of device failure is 80% when the temperature exceeds 70°C, the trigger probability of this condition is 0.8. Based on the historical data and the probability model, the probability distribution under each trigger condition is generated. For example, when the device is in a "high temperature" state, the conditions that lead to abnormal operation of the device include temperature, humidity, power supply voltage, etc., and the distribution probability of each condition is obtained by the above calculation method.

[0043] Step S42: performing abnormal operation trigger causal contribution analysis based on the abnormal operation trigger condition distribution probability data to generate abnormal operation trigger causal contribution data; In the embodiment of the present invention, the calculated probability data of the trigger condition distribution is used to perform causal contribution analysis, and a causal reasoning model is constructed based on the probability data of the trigger condition. A causal reasoning method (such as Granger causality test or a causal model based on a Bayesian network) is used to identify the causal relationship between each trigger condition and the abnormal operation of the device. Assuming that the conditions such as the temperature, humidity, and voltage of the device have different contributions to the abnormal operation state, analyze how these factors cause the device failure. Using statistical analysis techniques, based on the probability data of the trigger condition and the device failure data, the causal contribution of each trigger condition to the device failure is calculated. For example, if the frequency of the device failing under high temperature is 70%, and the frequency of the failure caused by voltage drop is 30%, the causal contribution of the high temperature condition to the failure is 0.7, and the voltage condition is 0.3. According to the analysis results, the contribution data of each trigger condition to the abnormal operation state is generated. For example, the causal contribution of the conditions such as excessive temperature, voltage fluctuation, and excessive humidity of a certain device to the occurrence of the failure is 0.6, 0.3, and 0.1 respectively. These contribution data provide an in-depth understanding of the abnormal operation of the device. The generated abnormal operation trigger causal contribution data describes in detail the causal relationship between different trigger conditions and the abnormal operation of the equipment, providing a key basis for the establishment of the subsequent decision tree.

[0044] Step S43: Designing a tree node for abnormal operation evaluation for the abnormal operation global decision factor using a decision tree algorithm to generate an abnormal operation evaluation tree node; using the abnormal operation state feature association data as a tree node combination factor, and using the abnormal operation state migration relationship data as a tree node expansion factor; establishing a preliminary decision tree for abnormal operation evaluation for the abnormal operation evaluation tree node according to the tree node combination factor and the tree node expansion factor, so as to obtain a preliminary decision tree model for abnormal operation evaluation; In an embodiment of the present invention, based on the collected global decision factors, a decision tree algorithm is used to design nodes for these factors. The root node of the decision tree usually selects the most critical factor, such as the device temperature. If the temperature exceeds 75°C, the probability of a fault occurring is greatly increased; then, it is determined whether the voltage is lower than 200V, and the node is designed based on whether the humidity exceeds 85%. Each decision factor will be used as a tree node, and the branches of the node represent different states when the factor reaches a certain threshold. For example, the branches of the temperature node are "temperature ≤ 75°C" and "temperature > 75°C", and the branches of the voltage node are "voltage > 200V" and "voltage ≤ 200V". In this way, tree nodes are designed through multiple decision factors to form a multi-level decision tree. The abnormal operation state feature association data is integrated into the decision tree as a combination factor of each tree node. The abnormal operation state feature association data includes the characteristic values ​​of the device under different operating states (such as temperature, humidity, voltage, etc.) and their correlation with the occurrence of faults. For example, if the failure rate of the device is 60% when the temperature is 75°C and 80% when the humidity is 90%, these data will be used as the combination factors of the tree node. The combination factors of each tree node are weighted according to historical data. For example, if the probability of failure is higher when the temperature is higher than 75°C, the weight of temperature will be set to a higher value in the combination factors of this node. The abnormal operation state transition relationship data is used as the expansion factor of the tree node. Assume that the failure state of the device does not occur immediately, but undergoes a migration process of multiple states. For example, the device will experience the "overheated" state, then the "too high temperature" state, and finally enter the failure state under high temperature. Therefore, these migration relationships also need to be represented by decision tree nodes. Under each tree node, add conditions related to other state transitions. For example, under the high temperature node, further determine whether there are state transition factors such as voltage drop. Each state transition will affect the probability of the final failure. According to the combination factors and expansion factors of the tree nodes, a preliminary decision tree for abnormal operation evaluation is established. Assume that factors such as temperature, voltage, and humidity constitute the root node and child node structure of the tree, and the branches of each node represent different equipment operating states (such as whether the temperature exceeds the standard, whether the voltage fluctuates, etc.). For example, the root node determines whether the temperature exceeds 75°C, and then further determines factors such as voltage and humidity based on different temperature states. According to the branch conditions and associated data of the tree nodes, the decision tree algorithm (such as ID3, CART, etc.) is used to generate the tree. Through the step-by-step judgment of these conditions and factors, a decision tree that can evaluate the abnormal operation of the equipment is finally formed. This decision tree will determine whether the equipment has an abnormality and give processing suggestions based on the type of abnormality. After the tree is built, a preliminary decision tree model is obtained by analyzing the path and leaf nodes of the decision tree. This model determines whether the equipment has an abnormality and provides accurate evaluation results.For example, if the equipment is in a high temperature and low voltage state, and the humidity is also too high, the leaf nodes of the decision tree will output the probability of equipment failure and provide early warnings to the operation and maintenance personnel. The preliminary decision tree model can be used as a basic model to provide equipment failure early warnings and conduct subsequent optimization and training. By further collecting equipment data, the node and branch conditions of the decision tree are gradually optimized to improve the accuracy and prediction ability of the model.

[0045] Step S44: performing fitting decision factor analysis on the abnormal operation global decision factor to generate the abnormal operation fitting decision factor; designing decision tree pruning rules according to the abnormal operation fitting decision factor; In an embodiment of the present invention, on the basis of the existing preliminary decision tree model for abnormal operation evaluation, multiple global decision factors (such as temperature, voltage, humidity, etc.) related to the operation status of medical equipment are collected and analyzed. For example, based on the equipment operation history data and abnormal failure data, these decision factors are fitted using statistical methods such as regression analysis to obtain fitted decision factors. For each decision factor, a linear or nonlinear fitting analysis is performed. Assuming that through the fitting model, the relationship between the temperature factor and the equipment failure is expressed by a linear equation, when the temperature exceeds a certain critical value, the probability of equipment failure increases rapidly. Similarly, factors such as voltage and humidity are also fitted and analyzed to generate fitted decision factors. Combined with the data of equipment operation, the contribution value of each decision factor to the equipment failure is calculated, and the abnormal operation fitting decision factor is generated. For example, the fitting model of the temperature factor shows that when the temperature exceeds 75°C, the probability of equipment failure increases by 20%, and this contribution value will be recorded as the output of the fitting decision factor. According to the results of the analysis of the fitting decision factor, the pruning rules of the decision tree are designed. The design basis of pruning rules is: when the branches of certain decision factors contribute very little to the final decision in the model, or fail to effectively improve the prediction accuracy in multiple decisions, these branches need to be pruned. For example, in the branch judgment of the temperature factor, if the fitting results show that the temperature has little effect on the fault, and the nodes in certain temperature ranges do not provide effective information, then set pruning rules to delete these nodes from the decision tree, simplify the tree structure, and avoid overfitting. When designing pruning rules, consider the interaction of multiple factors and set the pruning threshold. For example, if the intersection of the two factors of temperature and humidity cannot effectively distinguish the probability of fault occurrence in multiple states, then design rules for pruning to improve the operating efficiency and interpretability of the decision tree.

[0046] Step S45: performing decision tree pruning processing on the preliminary decision tree model for abnormal operation evaluation according to the decision tree pruning rules to generate an abnormal operation evaluation decision tree model; In the embodiment of the present invention, the designed pruning rules are used, combined with the abnormal operation evaluation preliminary decision tree model, and pruning is performed for each tree node. Assuming that some branches in the preliminary decision tree model (such as temperature exceeding the standard and voltage fluctuation) fail to significantly improve the prediction accuracy of the model, the pruning rules are applied to delete these branches from the decision tree. The pruning operation starts from the bottom node and is performed step by step upward. For each decision factor node, the influence of its child nodes is analyzed. If the contribution of some child nodes is too low (for example, the equipment failure rate does not change much in different temperature ranges), these nodes are pruned. Through pruning, the complexity and computational burden of the decision tree are reduced. The pruned decision tree simplifies the original model and retains key decision nodes, such as the branches of the two factors "temperature> 75°C" and "voltage fluctuation". The combination of these factors has a high predictive ability for equipment failure. Through pruning, the depth of the tree is reduced, which improves the efficiency and interpretability of the model. After pruning, the accuracy and predictive ability of the decision tree model are recalculated. It is assumed that the pruned decision tree can judge equipment failure more efficiently and reduce the risk of overfitting. The generated abnormal operation evaluation decision tree model is more concise, accurate and computationally efficient than the preliminary model. This model can effectively predict the abnormal state of medical equipment and provide maintenance personnel with more reliable fault warnings.

[0047] Step S46: using the abnormal operation triggering causal contribution data, the abnormal operation evaluation decision tree model is trained and optimized for the abnormal operation triggering causal contribution, to generate an abnormal operation evaluation optimized decision tree model.

[0048] In an embodiment of the present invention, abnormal operation triggering causal contribution data is used to train an optimized decision tree model. The causal contribution data is obtained through historical data analysis, indicating how certain specific events (such as equipment overheating, fault alarm, etc.) trigger abnormal operation states. Assuming that the probability of triggering a fault when the temperature of the device exceeds 75°C is 70%, and the probability of triggering a fault when the voltage fluctuation exceeds 10% is 50%, these causal contribution data will be used as input for training the optimized decision tree. During the training process, the causal contribution data is matched with the historical operation data of the device, and the training model learns the contribution value of each decision node to the abnormal state triggering. For example, when the contribution value of the temperature factor is greater than 75°C, it will be given a higher weight, while the contribution value of the voltage factor is smaller, and its influence is appropriately adjusted. Through optimized training, the decision tree model can not only predict abnormal operation based on the current state, but also optimize judgments based on the triggering causal relationship of the device history. For example, if historical data shows that the device is prone to failure under high temperature conditions, the model will adjust the weight of the temperature node to reduce dependence on other less relevant factors. After the training is completed, the accuracy of the model is verified by cross-validation and test sets. The optimized decision tree can accurately assess whether the equipment is in an abnormal operating state based on the trigger causal contribution data, and has a high prediction accuracy for the timing and risk factors of fault occurrence. The final abnormal operation evaluation optimization decision tree model has a higher prediction ability than the preliminary decision tree model, and can more finely reflect the fault triggering mechanism of the equipment in different states. This model can monitor the equipment operating status in real time and provide accurate fault prediction and warning.

[0049] Further, step S42 includes the following steps: Step S421: performing abnormal operation trigger condition probability density analysis according to the abnormal operation trigger condition distribution probability data to generate abnormal operation trigger condition probability density data, and performing cluster analysis on the abnormal operation trigger condition probability density data to generate abnormal operation trigger condition probability density cluster data; In an embodiment of the present invention, the distribution probability of abnormal operation trigger conditions is collected and calculated based on the abnormal state records in the historical medical equipment operation data. These trigger conditions include various equipment operation environment parameters such as temperature, voltage, pressure, humidity, etc. Assuming that the temperature range is 60°C to 100°C in a certain period of time, the trigger probability of each temperature interval (i.e., the probability of equipment failure) is recorded, and a probability data of the trigger condition distribution is generated. Probability density analysis is performed on these data. Assuming that temperature is the main influencing factor, the trigger probability density distribution of equipment failure presents a normal distribution, and the trigger probability of equipment failure is highest when the temperature is close to 80°C. By fitting the historical data, a probability density function (PDF) is obtained, and the trigger probability density data of each condition is calculated. Cluster analysis is performed on the probability density data of abnormal operation trigger conditions, with the aim of discovering the clustering pattern of equipment failure under different operating conditions. For example, the K-means clustering algorithm is used to cluster the failure probability density under different temperature intervals. Assume that through cluster analysis, three main failure probability clusters are obtained: low probability interval (60°C to 70°C), medium probability interval (70°C to 85°C), and high probability interval (85°C to 100°C). These cluster data will help identify high-risk operating intervals. Based on the results of cluster analysis, abnormal operation trigger condition probability density cluster data is generated to indicate the density distribution of the probability of equipment failure within a specific temperature range. Cluster data is used to identify high-incidence intervals of equipment failures, thereby guiding the improvement of fault prediction models.

[0050] Step S422: performing abnormal operation trigger condition probability weight analysis according to abnormal operation trigger condition probability density clustering data to generate abnormal operation trigger condition probability weight data; In an embodiment of the present invention, a weight analysis is performed based on the probability density clustering data of abnormal operation trigger conditions. The goal of the weight analysis is to determine the contribution of different trigger conditions to equipment failure. Assuming that the temperature interval "70°C to 85°C" obtained by cluster analysis has a higher probability of failure, the weight of the temperature factor in the overall equipment failure should be higher. The relative weight is obtained by calculating the ratio of the failure probability of each cluster interval to the total failure probability. For example, if the probability of equipment failure is 80% at a temperature of 75°C and 20% in other intervals, the weight of the 75°C interval will be higher, and the weight is calculated by weighted average method or normalization method. Generate weight data of abnormal operation trigger conditions for subsequent fault assessment and decision analysis. Assuming that the weight is 0.6 at a temperature of 75°C, and the weights are 0.2 and 0.2 at 65°C and 85°C respectively, these data will serve as the basis for subsequent decisions.

[0051] Step S423: Perform abnormal operation trigger causal contribution analysis based on the abnormal operation trigger condition distribution probability data and the abnormal operation trigger condition probability weight data to generate abnormal operation trigger causal contribution data.

[0052] In the embodiment of the present invention, causal contribution analysis is performed based on the abnormal operation trigger condition distribution probability data and the abnormal operation trigger condition probability weight data. The purpose of this analysis is to determine the influence of each trigger condition (such as temperature, voltage, etc.) on the occurrence of equipment failure. First, through the trigger condition distribution probability data, identify which trigger conditions have a greater impact on the probability of equipment failure. Combined with the weight data, the contribution of each condition is further analyzed. For example, the trigger probability density of 75°C is high, and the weight is 0.6, which indicates that temperature is the main causal factor of equipment failure. Although voltage fluctuation has a certain probability under different conditions, its causal contribution is small due to its low weight (such as 0.1). Generate abnormal operation trigger causal contribution data to record the specific causal contribution of each trigger condition in the occurrence of equipment failure. For example, the causal contribution of temperature (75°C) is 60%, while the causal contribution of voltage fluctuation is 10%. These causal contribution data will be used to identify the most critical failure factors in the subsequent decision tree model. The causal contribution data are summarized to generate the final trigger causal contribution data. The causal contribution value of each factor (such as temperature, voltage, pressure, etc.) is regarded as the weight of the decision factor, which further affects the subsequent fault assessment and optimization.

[0053] Further, step S5 includes the following steps: Step S51: transmitting the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform abnormal operation trend evaluation processing on the medical equipment, and generating abnormal operation trend evaluation data of the medical equipment, wherein the abnormal operation trend evaluation data of the medical equipment includes abnormal operation perception topological nodes of the medical equipment, abnormal operation trend time nodes of the medical equipment, and abnormal operation failure risk factors of the medical equipment; In an embodiment of the present invention, the real-time operation data (including temperature, voltage, pressure, etc.) of the medical device is transmitted to the device monitoring platform through the Internet of Things interface. The data content includes a variety of sensor data and monitoring information collected by the device during operation, involving device status such as operating time, load, fault records, etc. These real-time operation data are input into the optimized decision tree model. The decision tree model analyzes whether there is a risk of failure by judging the change trend of the device operation status based on historical data and previous abnormal operation modes. The data is processed by the optimized decision tree model to evaluate the abnormal operation trend of the device. The model outputs abnormal operation trend data according to the state feature data, including: medical device abnormal operation perception topology nodes, which are used to identify key nodes where the device fails, such as high temperature areas or device nodes with abnormal loads; medical device abnormal operation trend time nodes, which are used to determine when the failure occurs, such as when the temperature continues to rise, the failure of the device occurs within the next 30 minutes; medical device abnormal operation failure risk factors, which are used to calculate the risk contribution of various factors to the occurrence of failures during the operation of the device, for example, temperature increase, load increase, etc. are the main risk factors. The final medical device abnormal operation trend evaluation data is generated as the basis for subsequent fault prediction, early warning and decision-making. These data help predict the time and specific location of equipment failure and take maintenance measures in advance.

[0054] Step S52: performing abnormal operation failure mode analysis of the medical equipment according to the abnormal operation failure risk factors of the medical equipment, and generating abnormal operation failure mode data of the medical equipment; performing quantitative analysis of the abnormal operation failure risk factors of the medical equipment according to the abnormal operation failure risk factors of the medical equipment, and generating quantitative data of the abnormal operation failure risk factors of the medical equipment; marking the abnormal operation failure mode data of the medical equipment and the quantitative data of the abnormal operation failure risk factors of the medical equipment as abnormal operation failure response feedback data of the medical equipment; In an embodiment of the present invention, according to the data of the fault risk factors of abnormal operation of the medical device, the fault mode of the device under the action of different fault risk factors is analyzed. For example, when the temperature is too high, the circuit is damaged due to overheating, and when the pressure is too high, the valve fails or the pipe ruptures. The analysis of the historical fault mode is combined with the real-time operation data to determine the fault mode that occurs. For example, if the device is under certain specific working conditions, such as overload operation, temperature increase, etc., the fault mode will be manifested as battery damage or system overload, etc. The fault mode data of abnormal operation of the medical device is output, and the factors affecting the stability and safety of the device under different fault modes are identified. Each risk factor is quantitatively analyzed to determine the specific impact of each risk factor. For example, the risk factor of temperature increase affects the probability of failure of the device by 40%, while the risk impact of load increase is 25%, and the impact of excessive pressure is 15%. Through quantitative analysis, the weight data of each risk factor is generated. The fault mode data and the quantified risk factor data are identified as the fault response feedback data of the abnormal operation of the medical device, which is used for subsequent fault response and early warning.

[0055] Step S53: Establishing an automated feedback engine for medical device fault response based on the medical device abnormal operation perception topological nodes, the medical device abnormal operation trend time node data, and the medical device abnormal operation fault response feedback data, and generating a medical device fault response intelligent feedback engine; In an embodiment of the present invention, an automated feedback engine for medical equipment fault response is established by combining the abnormal operation perception topology nodes, trend time nodes and fault response feedback data of medical equipment. For example, when the temperature of the equipment is sensed to exceed the set threshold, the feedback engine will automatically trigger related operations, such as starting the backup cooling system or alarming maintenance personnel. The feedback engine can automatically determine the required maintenance measures, early warning processing or alarm measures based on the fault response data and the real-time status of the equipment. For example, if the equipment is highly loaded and close to the fault mode, the feedback engine can start measures to reduce the load of the equipment to avoid the occurrence of failures. Various feedback mechanisms (including automatic equipment response, remote control, personnel notification, etc.) are integrated to design an intelligent feedback system. When the system evaluates that the equipment operation has an abnormal trend, the intelligent feedback engine can automatically generate response instructions and notify maintenance personnel or start an automatic repair program.

[0056] Step S54: Execute the medical equipment monitoring fault response feedback operation through the medical equipment fault response intelligent feedback engine.

[0057] In an embodiment of the present invention, once the medical equipment monitoring system detects an abnormality in the equipment and analyzes the failure mode and risk through the intelligent feedback engine, the feedback engine will perform a fault response operation. This operation includes: starting the emergency cooling system, load reduction device, shutting down certain dangerous modules, etc.; starting the equipment diagnostic program, troubleshooting, and automatically performing self-repair; sending a fault alarm to maintenance personnel or operators, starting on-site repair or remote repair. Assuming that the equipment detects a voltage overload, the intelligent feedback engine will trigger the self-protection mechanism of the power supply system, automatically shut down the power supply and notify the operator to check. At this time, the equipment can not only reduce the probability of failure, but also take measures before the failure occurs through the automatic feedback mechanism to reduce losses. After completing the feedback operation, the equipment resumes normal operation, the fault response feedback operation is completed, and the system continues to monitor the equipment status to ensure that similar failures do not occur in the equipment again.

[0058] This specification provides an artificial intelligence-based medical device monitoring system for executing the artificial intelligence-based medical device monitoring method as described above. The artificial intelligence-based medical device monitoring system includes: The medical device operation perception topology analysis module is used to perform a medical device interconnection operation on the medical device using a preset Internet of Things communication interface, and perform medical device operation monitoring processing on the medical device after the medical device interconnection operation to obtain medical device operation monitoring data; perform medical device operation perception topology data analysis on the medical device operation monitoring data to generate medical device operation perception topology data; A medical equipment operation status characteristic analysis module is used to analyze the operation status characteristics of the medical equipment based on the medical equipment operation perception topology data, and generate the medical equipment operation status characteristic data; The abnormal operation global decision factor analysis module is used to obtain historical medical equipment abnormal operation data; perform historical medical equipment abnormal operation state characteristic analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state characteristic data, and generate historical medical equipment abnormal operation state characteristic data; perform global decision factor analysis on abnormal operation of medical equipment based on the historical medical equipment abnormal operation state characteristic data, and generate abnormal operation global decision factors; The abnormal operation evaluation optimization decision tree construction module is used to calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor, and generate abnormal operation triggering condition distribution probability data; perform abnormal operation triggering causal contribution analysis based on the abnormal operation triggering condition distribution probability data, and generate abnormal operation triggering causal contribution data; construct an abnormal operation evaluation decision tree model based on the abnormal operation global decision factor; use the abnormal operation triggering causal contribution data to perform training optimization processing on the abnormal operation evaluation decision tree model for the abnormal operation triggering causal contribution, and generate an abnormal operation evaluation optimization decision tree model; The medical equipment monitoring fault response feedback module is used to transmit the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform the medical equipment abnormal operation trend evaluation processing and generate the medical equipment abnormal operation trend evaluation data; establish the medical equipment fault response automation feedback engine based on the medical equipment abnormal operation trend evaluation data and generate the medical equipment fault response intelligent feedback engine; perform the medical equipment monitoring fault response feedback operation through the medical equipment fault response intelligent feedback engine.

[0059] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0060] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A medical equipment monitoring method based on artificial intelligence, characterized in that: The following steps are involved: Step S1: performing a medical device interconnection operation on the medical device using a preset Internet of Things communication interface, and performing medical device operation monitoring processing on the medical device after the medical device interconnection operation to obtain medical device operation monitoring data; performing medical device operation perception topology data analysis on the medical device operation monitoring data to generate medical device operation perception topology data; Step S2: performing a state characteristic analysis of the operation of the medical device based on the medical device operation perception topology data to generate medical device operation state characteristic data; Step S3: Acquire historical medical equipment abnormal operation data; perform historical medical equipment abnormal operation state characteristic analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state characteristic data to generate historical medical equipment abnormal operation state characteristic data; perform global decision factor analysis of medical equipment abnormal operation based on the historical medical equipment abnormal operation state characteristic data to generate abnormal operation global decision factor; Step S4: Calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor to generate abnormal operation triggering condition distribution probability data; Perform abnormal operation trigger causal contribution analysis based on abnormal operation trigger condition distribution probability data to generate abnormal operation trigger causal contribution data; Construct an abnormal operation evaluation decision tree model based on the abnormal operation global decision factor; use the abnormal operation trigger causal contribution data to train and optimize the abnormal operation trigger causal contribution of the abnormal operation evaluation decision tree model to generate an abnormal operation evaluation optimization decision tree model; Step S5: transmitting the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform abnormal operation trend evaluation processing on the medical equipment to generate abnormal operation trend evaluation data of the medical equipment; Establish an automated feedback engine for medical equipment failure response based on abnormal operation trend assessment data of medical equipment, and generate an intelligent feedback engine for medical equipment failure response; Medical equipment monitoring fault response feedback tasks are performed through the medical equipment fault response intelligent feedback engine.

2. The medical equipment monitoring method based on artificial intelligence according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: performing a medical device connection operation on the medical device using a preset Internet of Things communication interface, and performing a medical device operation monitoring process on the medical device after the medical device connection operation to obtain medical device operation monitoring data; Step S12: performing monitoring sensing node identification according to the Internet of Things communication interface corresponding to the medical equipment operation monitoring data, and generating an identified monitoring sensing node; Step S13: Perform medical equipment operation perception topology data analysis on the medical equipment operation monitoring data based on the identified monitoring perception node to generate medical equipment operation perception topology data.

3. The medical equipment monitoring method based on artificial intelligence according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: performing sensor node parsing processing on the identification monitoring sensor node to generate identification monitoring sensor node parsing data; Step S132: using the identification monitoring perception node parsing data to perform heterogeneous monitoring data synchronization processing on the medical equipment operation monitoring data to generate synchronized medical equipment operation monitoring data; Step S133: Establish a medical equipment perception topology structure by identifying monitoring perception nodes, transmit synchronized medical equipment operation monitoring data to the medical equipment perception topology structure to perform medical equipment operation perception topology data analysis, and generate medical equipment operation perception topology data.

4. The medical equipment monitoring method based on artificial intelligence according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing medical device state variable analysis according to the medical device operation perception topology data to generate medical device state variable data; Step S22: performing key variable analysis on the medical device state variable data to generate medical device state key variable data; Step S23: Performing a state characteristic analysis of the medical device operation on the medical device operation perception topology data according to the medical device state key variable data to generate medical device operation state characteristic data.

5. The medical equipment monitoring method based on artificial intelligence according to claim 4 is characterized in that: Step S22 includes the following steps: Step S221: performing medical equipment operation condition analysis of each topology data according to the medical equipment operation perception topology data, and generating medical equipment operation condition data; Step S222: performing medical device state variable-specific analysis processing of the medical device operating condition on the medical device state variable data based on the medical device operating condition data to generate medical device state variable-specific data; Step S223: performing multi-scale analysis processing on the medical device state variable specific data to generate medical device state variable multi-scale specific data; performing medical device state variable dependency analysis on the medical device state variable specific data to generate medical device state variable dependency data; Step S224: Perform key variable analysis on the medical device state variable data according to the medical device state variable multi-scale specificity data and the medical device state variable dependency data to generate medical device state key variable data.

6. The medical equipment monitoring method based on artificial intelligence according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: Acquire historical abnormal operation data of medical equipment; Step S32: performing historical abnormal operation state characteristic analysis on the historical abnormal operation state data of medical equipment according to the medical equipment operation state characteristic data, and generating historical abnormal operation state characteristic data of medical equipment; Step S33: performing abnormal operation state characteristic correlation analysis based on historical abnormal operation state characteristic data of medical equipment to generate abnormal operation state characteristic correlation data; Step S34: performing abnormal operation state feature correlation analysis according to the abnormal operation state feature correlation data, generating abnormal operation state feature correlation data, and designing an abnormal operation state feature correlation map using the abnormal operation state feature correlation data as a main node; Step S35: performing time series data analysis of abnormal operation state characteristics on historical abnormal operation state characteristic data of medical equipment to generate time series abnormal operation state characteristic data; performing abnormal operation state migration relationship analysis based on the time series abnormal operation state characteristic data to generate abnormal operation state migration relationship data; Step S36: The abnormal operation state migration relationship data corresponding to the abnormal operation state feature association data is transmitted as an extension node of the main node to the abnormal operation state feature association map to perform global decision factor analysis on abnormal operation of the medical device and generate a global decision factor for abnormal operation.

7. The medical equipment monitoring method based on artificial intelligence according to claim 6, characterized in that: Step S4 includes the following steps: Step S41: Calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor to generate abnormal operation triggering condition distribution probability data; Step S42: performing abnormal operation trigger causal contribution analysis based on the abnormal operation trigger condition distribution probability data to generate abnormal operation trigger causal contribution data; Step S43: Designing a tree node for abnormal operation evaluation for the abnormal operation global decision factor using a decision tree algorithm to generate an abnormal operation evaluation tree node; using the abnormal operation state feature association data as a tree node combination factor, and using the abnormal operation state migration relationship data as a tree node expansion factor; establishing a preliminary decision tree for abnormal operation evaluation for the abnormal operation evaluation tree node according to the tree node combination factor and the tree node expansion factor, so as to obtain a preliminary decision tree model for abnormal operation evaluation; Step S44: performing fitting decision factor analysis on the abnormal operation global decision factor to generate the abnormal operation fitting decision factor; designing decision tree pruning rules according to the abnormal operation fitting decision factor; Step S45: performing decision tree pruning processing on the preliminary decision tree model for abnormal operation evaluation according to the decision tree pruning rules to generate an abnormal operation evaluation decision tree model; Step S46: using the abnormal operation triggering causal contribution data, the abnormal operation evaluation decision tree model is trained and optimized for the abnormal operation triggering causal contribution, to generate an abnormal operation evaluation optimized decision tree model.

8. The medical equipment monitoring method based on artificial intelligence according to claim 7, characterized in that: Step S42 includes the following steps: Step S421: performing abnormal operation trigger condition probability density analysis according to the abnormal operation trigger condition distribution probability data to generate abnormal operation trigger condition probability density data, and performing cluster analysis on the abnormal operation trigger condition probability density data to generate abnormal operation trigger condition probability density cluster data; Step S422: performing abnormal operation trigger condition probability weight analysis according to abnormal operation trigger condition probability density clustering data to generate abnormal operation trigger condition probability weight data; Step S423: Perform abnormal operation trigger causal contribution analysis based on the abnormal operation trigger condition distribution probability data and the abnormal operation trigger condition probability weight data to generate abnormal operation trigger causal contribution data.

9. The medical equipment monitoring method based on artificial intelligence according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: transmitting the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform abnormal operation trend evaluation processing on the medical equipment, and generating abnormal operation trend evaluation data of the medical equipment, wherein the abnormal operation trend evaluation data of the medical equipment includes abnormal operation perception topological nodes of the medical equipment, abnormal operation trend time nodes of the medical equipment, and abnormal operation failure risk factors of the medical equipment; Step S52: performing abnormal operation failure mode analysis of the medical equipment according to the abnormal operation failure risk factors of the medical equipment, and generating abnormal operation failure mode data of the medical equipment; performing quantitative analysis of the abnormal operation failure risk factors of the medical equipment according to the abnormal operation failure risk factors of the medical equipment, and generating quantitative data of the abnormal operation failure risk factors of the medical equipment; marking the abnormal operation failure mode data of the medical equipment and the quantitative data of the abnormal operation failure risk factors of the medical equipment as abnormal operation failure response feedback data of the medical equipment; Step S53: Establishing an automated feedback engine for medical device fault response based on the medical device abnormal operation perception topological nodes, the medical device abnormal operation trend time node data, and the medical device abnormal operation fault response feedback data, and generating a medical device fault response intelligent feedback engine; Step S54: Execute the medical equipment monitoring fault response feedback operation through the medical equipment fault response intelligent feedback engine.

10. A medical equipment monitoring system based on artificial intelligence, characterized in that: For executing the artificial intelligence-based medical device monitoring method as claimed in claim 1, the artificial intelligence-based medical device monitoring system comprises: The medical device operation perception topology analysis module is used to perform a medical device interconnection operation on the medical device using a preset Internet of Things communication interface, and perform medical device operation monitoring processing on the medical device after the medical device interconnection operation to obtain medical device operation monitoring data; perform medical device operation perception topology data analysis on the medical device operation monitoring data to generate medical device operation perception topology data; A medical equipment operation status characteristic analysis module is used to analyze the operation status characteristics of the medical equipment based on the medical equipment operation perception topology data, and generate the medical equipment operation status characteristic data; The abnormal operation global decision factor analysis module is used to obtain historical medical equipment abnormal operation data; perform historical medical equipment abnormal operation state characteristic analysis on the historical medical equipment abnormal operation data according to the medical equipment operation state characteristic data, and generate historical medical equipment abnormal operation state characteristic data; perform global decision factor analysis on abnormal operation of medical equipment based on the historical medical equipment abnormal operation state characteristic data, and generate abnormal operation global decision factors; The abnormal operation evaluation optimization decision tree construction module is used to calculate the conditional distribution probability of abnormal operation triggering for the abnormal operation global decision factor, and generate abnormal operation triggering condition distribution probability data; perform abnormal operation triggering causal contribution analysis based on the abnormal operation triggering condition distribution probability data, and generate abnormal operation triggering causal contribution data; construct an abnormal operation evaluation decision tree model based on the abnormal operation global decision factor; use the abnormal operation triggering causal contribution data to perform training optimization processing on the abnormal operation evaluation decision tree model for the abnormal operation triggering causal contribution, and generate an abnormal operation evaluation optimization decision tree model; The medical equipment monitoring fault response feedback module is used to transmit the medical equipment operation status characteristic data to the abnormal operation evaluation optimization decision tree model to perform the medical equipment abnormal operation trend evaluation processing and generate the medical equipment abnormal operation trend evaluation data; establish the medical equipment fault response automation feedback engine based on the medical equipment abnormal operation trend evaluation data and generate the medical equipment fault response intelligent feedback engine; perform the medical equipment monitoring fault response feedback operation through the medical equipment fault response intelligent feedback engine.

Images