Artificial intelligence-based intelligent management system and method for surgical equipment, and storage medium

Abstract

The application relates to the technical field of equipment management informatization, and discloses a surgical equipment intelligent management system and method based on artificial intelligence and a storage medium, wherein the system comprises an equipment management module, a surgery management module, an equipment use prediction module, an intelligent scheduling module, an equipment fault early warning module, a data interface module and a model optimization module. The equipment management module defines the equipment hierarchical relationship in a tree structure, and the surgery management module is associated with surgery and equipment information. After external data is acquired through the data interface module, the equipment use prediction module outputs demand prediction by using a time series model, the intelligent scheduling module generates a scheduling scheme in combination with a tree structure and a graph neural network model, and the equipment fault early warning module outputs fault early warning through a classification model. The model optimization module continuously optimizes the model. The system and the storage medium correspond to the method. The application realizes the intelligentization of surgical equipment management, and improves resource utilization efficiency, scheduling safety and operation reliability.

Description

Technical Field

[0001] This application relates to the field of information technology for equipment management, specifically an intelligent management system, method, and storage medium for surgical equipment based on artificial intelligence. Background Technology

[0002] In the daily operation of hospital operating rooms, the management and scheduling of surgical equipment is a complex and critical task. Currently, common management methods rely on manual records or simple spreadsheets, which have significant limitations. First, equipment status information (such as idle, in use, under maintenance) cannot be obtained and updated accurately and in real time, leading to frequent conflicts or delays in surgical scheduling due to unclear equipment availability. Second, equipment scheduling heavily relies on human experience, lacking intelligent analysis of historical usage data, making it impossible to effectively predict future surgical equipment needs and hindering efficient, conflict-free resource allocation across multiple surgeries and equipment. Third, equipment maintenance is typically based on fixed cycles or post-operative repairs, lacking early fault warning capabilities based on real-time equipment operating data, potentially impacting surgical safety and progress due to sudden equipment failures. Furthermore, existing solutions often isolate functional modules, hindering data flow and lacking a collaborative system that integrates equipment management, surgical scheduling, intelligent decision-making, and continuous optimization, making it difficult to improve overall operating room operational efficiency and resource utilization. Therefore, there is an urgent need for an integrated intelligent management solution that enables real-time equipment status monitoring, intelligent demand measurement, automatic scheduling optimization, and early fault warnings. Summary of the Invention

[0003] The purpose of this application is to provide an intelligent management system, method, and storage medium for surgical equipment based on artificial intelligence, so as to solve the technical problems mentioned in the background art.

[0004] To achieve the above objectives, this application discloses the following technical solutions: In a first aspect, this application discloses an intelligent management method for surgical equipment based on artificial intelligence, the system comprising: The equipment management module is used to manage surgical equipment in a tree structure, which defines the hierarchical relationship between the main equipment and its sub-equipment. The surgery management module is used to manage surgical information and associate the surgical information with the surgical equipment in the equipment management module according to the surgical schedule; The equipment usage prediction module is used to output equipment demand prediction data based on historical surgical data and historical equipment usage records through a time series prediction model; The intelligent scheduling module is used to output a conflict-free surgical equipment scheduling scheme based on the current surgical information and the real-time status of the equipment, combined with the tree structure, through a graph neural network model. The equipment fault early warning module is used to output fault early warning information of surgical equipment based on sensor data and maintenance records of surgical equipment and through an equipment fault classification model. The data interface module is used to obtain the historical surgical data, current surgical information, and sensor data from external medical systems; The model optimization module is used to collect the operational performance data of the equipment usage prediction module, the intelligent scheduling module, and the equipment fault early warning module, and to optimize the model based on the operational performance data. The demand forecast data output by the device prediction module, the surgical equipment scheduling plan output by the intelligent scheduling module, and the fault warning information output by the device fault warning module are synchronized to the device management module and the surgical management module to guide device scheduling and surgical arrangements.

[0005] Optionally, the device uses a prediction module including: The feature extraction unit is used to perform time-series alignment and feature extraction on the surgical type, duration and department information in the historical surgical data and the equipment usage sequence in the historical equipment usage record according to the timestamp, and to construct the equipment usage training dataset. The prediction model training unit is used to train a long short-term memory network model using the training dataset on the device to obtain a trained time series prediction model. The prediction execution unit is used to input the current surgical scheduling data into the time series prediction model and output the predicted demand data for each surgical device in the future time period.

[0006] The intelligent scheduling module includes: The relationship graph construction unit is used to construct a scheduling relationship graph based on the tree structure, current surgical information and real-time equipment status, wherein the nodes of the scheduling relationship graph represent surgery or equipment, and the edges represent equipment occupancy, time conflict or equipment hierarchical dependency relationship. The conflict analysis unit is used to input the scheduling relationship graph into a pre-trained graph neural network model, aggregate node information through graph convolution operations, and strengthen the weight of key edges using an attention mechanism to identify device occupancy conflicts and time conflicts. The scheme generation unit is used to generate the surgical equipment scheduling scheme by adjusting the allocation and timing of surgical equipment through a constraint optimization algorithm based on the identified equipment occupancy conflicts and time conflicts, with the optimization goal of minimizing equipment idle time and balancing departmental load.

[0007] Optionally, the equipment fault early warning module includes: The feature fusion unit is used to perform feature fusion on the physical parameters in the sensor data and the maintenance and fault history in the maintenance record to construct a feature sample set of equipment health status; The early warning model training unit is used to train a random forest classification model or a gradient boosting decision tree classification model using the equipment health status feature sample set to obtain a trained equipment fault classification model. The early warning output unit is used to input the real-time collected sensor data into the equipment fault classification model. When the output fault probability exceeds a preset threshold, it generates fault early warning information with equipment identification and early warning level.

[0008] Optionally, the system further includes a natural language interaction module; the natural language interaction module is used for: It receives voice commands from users and converts them into text commands using automatic speech recognition technology; Natural language processing is performed on the text instructions to parse the user's intent and the associated device or surgical entity; Based on the parsed user intent, the device management module is invoked to query the device status, or the surgical management module is invoked to operate the surgical record. When the surgical management module is invoked to perform an operation involving surgical time or equipment changes, or when the device management module is invoked to perform an operation to modify the device status, the natural language interaction module triggers the intelligent scheduling module to perform conflict detection and update the surgical equipment scheduling scheme.

[0009] Optionally, the surgical management module includes: The device list creation unit is used to load the doctor list and the available device list based on the tree structure in parallel in response to a request to create or modify a surgical record; A time availability verification unit is used to verify the availability of the surgeon at the selected surgery time in response to the user's selection of the surgeon. The occupancy conflict verification unit, in response to the user's selection of a sub-device through the tree structure, checks the current status of the selected sub-device and its occupancy conflicts within a selected time period through the intelligent scheduling module; The surgical record management unit is used to save surgical records after verifying surgical information, physician availability, and equipment conflicts.

[0010] Optionally, the model optimization module includes: The data collection unit is used to collect the prediction deviation data between the demand prediction data output by the equipment prediction module and the actual demand data, the conflict frequency data of the surgical equipment scheduling scheme output by the intelligent scheduling module in actual execution, and the accuracy data of the fault warning information output by the equipment fault warning module. The optimization analysis unit is used to perform parameter tuning or incremental learning analysis on the time series prediction model, the graph neural network model, and the equipment fault classification model based on the prediction deviation data, conflict occurrence frequency data, and early warning accuracy data collected by the data collection unit. The model update unit is used to update the model after parameter tuning or incremental learning to the corresponding module based on the analysis results of the optimization analysis unit.

[0011] Optionally, the constrained optimization algorithm includes: Step 1: Define decision variables , indicating surgery Is it in the device? time Start execution; Step 2: Define the objective function, the expression of which is: in, For the total scheduling duration, For surgery Duration, The total number of devices. Total number of departments; Indicates equipment The average conflict score during the scheduling period. The device output by the conflict analysis unit In time Conflict score; Indicates department Total surgical workload Indicates the average departmental workload. Belonging to the department A collection of surgical procedures; Indicates department The average conflict score of the scheduled surgeries; and The preset weighting coefficient is positive; Step 3: When the constrained optimization algorithm solves the objective function, the following constraints must be satisfied: Constraint 1: Each surgery must be scheduled and only once. ; Constraint 2, Equipment Time Exclusivity Constraint: The same equipment can only be used by one surgery at a time; Constraint 3, Conflict Avoidance Constraint: If a time constraint exists... make Surgery is prohibited if... In time Start using the equipment ,in The preset conflict threshold; Constraint 4, Equipment Tree Structure Dependency Constraint: If surgery Sub-device used Then its main device must be used simultaneously. And the usage time is the same: .

[0012] Secondly, this application discloses an intelligent management method for surgical equipment based on artificial intelligence, which includes the following steps: Obtain historical surgical data, current surgical information, and sensor data from surgical equipment from external medical systems; Surgical equipment is managed using a tree structure, which is used to define the hierarchical relationship between the main equipment and its sub-equipment. Manage surgical information and associate the surgical information with the surgical equipment in the tree structure according to the surgical schedule; Based on the historical surgical data and historical equipment usage records, equipment demand forecast data is output through a time series prediction model. Based on the current surgical information and real-time equipment status, and combined with the tree structure, a conflict-free surgical equipment scheduling scheme is output through a graph neural network model. Based on the sensor data and maintenance records of the surgical equipment, fault warning information of the surgical equipment is output through the equipment fault classification model; Collect the output device demand forecast data, output surgical equipment scheduling scheme, and output fault early warning information corresponding to the operation effect data, and optimize the time series prediction model, graph neural network model, and equipment fault classification model based on the operation effect data; The equipment demand forecast data, the surgical equipment scheduling plan, and the fault warning information are used to guide equipment scheduling and surgical arrangements.

[0013] Thirdly, this application discloses a computer-readable storage medium storing a computer program that can be executed by a processor. When the computer program is executed by the processor, it implements the artificial intelligence-based intelligent management method for surgical equipment as described above.

[0014] Beneficial Effects: The artificial intelligence-based intelligent management system, method, and storage medium of this application, by constructing a collaborative system that integrates equipment tree structure management, demand prediction based on time series models, intelligent scheduling that combines graph neural networks and constraint optimization, fault early warning based on classification models, and continuous self-optimization of models, can achieve refined and intelligent management of the entire life cycle of surgical equipment, improve the utilization rate and scheduling efficiency of operating room equipment, ensure the smoothness of surgical arrangements through accurate demand prediction and automated conflict resolution, and improve the reliability of equipment operation and surgical safety through early fault warning, ultimately achieving the effect of optimizing the allocation of medical resources and ensuring the efficient and safe operation of the operating room. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A block diagram of an AI-based intelligent management system for surgical equipment provided in this application embodiment; Figure 2 A flowchart illustrating an artificial intelligence-based intelligent management method for surgical equipment provided in this application embodiment. Detailed Implementation

[0017] To facilitate understanding of the technical solutions provided in the embodiments of this application, the background technology involved in the embodiments of this application will be described below.

[0018] In the daily operation of hospital operating rooms, surgical equipment is a core resource, and its management and scheduling is a crucial and complex task that directly affects surgical efficiency, safety, and cost. Current management practices have significant shortcomings in terms of digitalization and intelligentization, specifically in the following aspects: 1. Delayed equipment status awareness and severe information silos. Current management methods largely rely on medical staff manually recording equipment usage, cleaning, and maintenance status in paper logs or entering this information into isolated spreadsheets. This approach results in equipment status updates lagging significantly behind actual changes, and information is fragmented and difficult to share in real time. Operating room managers and clinicians cannot immediately and accurately know whether a particular piece of equipment is currently "idle," "in use," or "under maintenance," leading to situations where equipment is already in use but is still scheduled for surgery, causing temporary scheduling conflicts and surgical delays. Furthermore, complex equipment (such as large imaging systems) often consists of a main unit and multiple specialized accessories; current methods lack effective modeling and management of the internal hierarchical relationships of such equipment, further increasing scheduling complexity.

[0019] 2. Scheduling decisions rely heavily on human experience, lacking foresight and global optimization capabilities. The allocation of surgical equipment depends heavily on the scheduler's individual experience and on-the-spot judgment. When faced with surgical plans spanning multiple days, multiple operating rooms, and various types of surgeries, manual methods struggle to deeply analyze massive amounts of historical surgical data and equipment usage records, making it impossible to scientifically predict peak and trough demand for different equipment during specific future periods. This easily leads to resource misallocation, with some equipment overworked while others remain idle. More challenging is the scheduling of multiple surgeries and equipment; manual methods are almost incapable of accurately calculating and avoiding all potential time conflicts and resource competition in real time. Scheduling solutions are often not optimal, and may even contain hidden conflicts, resulting in bottlenecks in improving equipment utilization efficiency and overall operating room throughput.

[0020] 3. Passive equipment maintenance and insufficient preventative measures lead to operational risks. Current equipment maintenance largely follows fixed time cycles or is performed reactively after a significant malfunction. This approach lacks continuous monitoring and intelligent analysis of real-time equipment operating conditions (such as sensor data on vibration, temperature, and power consumption), failing to identify early warning signals before malfunctions occur. If critical surgical equipment malfunctions during surgery, it not only interrupts the procedure but may also endanger patient safety, causing serious medical risks and economic losses.

[0021] 4. Fragmented system functions and lack of collaborative intelligence. Some existing IT tools on the market may solve one or two of the above problems, such as independent equipment file systems or simple appointment calendars. However, the functional modules of these systems are often isolated from each other, and data cannot be integrated, forming information silos. There is a lack of effective linkage and integration between equipment status data, surgical scheduling data, maintenance record data, and real-time sensor data, which cannot support a closed-loop intelligent management process from "perception" to "prediction" to "decision-making" and "optimization". Therefore, from an overall operational perspective, there is still a lack of an integrated intelligent management solution that can integrate the entire process and has continuous learning and self-optimization capabilities.

[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application. Secondly, in this document, the term "comprising" is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements, but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0023] Firstly, this embodiment provides an intelligent management system for surgical equipment based on artificial intelligence, such as... Figure 1 As shown, the system includes an equipment management module, a surgical management module, an equipment usage prediction module, an intelligent scheduling module, an equipment fault early warning module, a data interface module, and a model optimization module, which are described in detail below.

[0024] The device management module is used to manage surgical devices in a tree structure, which defines the hierarchical relationship between the main device and its sub-devices.

[0025] In practical implementation, the equipment management module uses a tree-structured management system on the server by creating an equipment information table and an equipment relationship table. The equipment information table records each device's unique ID, name, type, current status (e.g., 0 - idle, 1 - in use, 2 - under maintenance), and the department to which it belongs. The equipment relationship table records the hierarchical relationships between devices, including master device ID and sub-device ID fields, thus defining the hierarchical relationship between the master device and its sub-devices. For example, a "high-definition laparoscopic system" (master device) may be associated with multiple sub-devices such as a "camera host," a "cold light source," and a "pneumoperitoneum machine."

[0026] The surgery management module is used to manage surgical information and associate surgical information with surgical equipment in the equipment management module according to the surgical schedule.

[0027] In practice, the surgery management module manages surgical information through a surgery information table. The surgery information table includes fields such as surgery ID, patient name, surgery name, planned start time, planned end time, and surgeon ID. A surgery-equipment association table is used to associate surgical information with surgical equipment in the equipment management module. This association table records the surgery ID and the equipment ID used.

[0028] The equipment usage prediction module is used to output equipment demand prediction data based on historical surgical data and historical equipment usage records through a time series prediction model.

[0029] In practice, the equipment usage prediction module is a software module deployed on the AI ​​server. It receives historical surgical data and historical equipment usage records from the data interface module, internally calls and executes a pre-trained time series prediction model (such as an LSTM model), and outputs the demand prediction data for each surgical device in the next week (for example, in the format: {device ID: date: predicted demand}).

[0030] The intelligent scheduling module is used to output a conflict-free surgical equipment scheduling scheme based on the current surgical information and the real-time status of the equipment, combined with a tree structure and a graph neural network model.

[0031] In practical implementation, the intelligent scheduling module is another AI software module. It receives current surgical information (from the surgical management module) and real-time equipment status (from the equipment management module), and combines the tree-structured relationships between the equipment. It then analyzes the data through a pre-trained graph neural network model and finally outputs a conflict-free surgical equipment scheduling scheme (e.g., a list containing the equipment allocated to each surgery and the specific time period for its use).

[0032] The equipment fault early warning module is used to output fault early warning information of surgical equipment based on sensor data and maintenance records of surgical equipment and through equipment fault classification model.

[0033] In practical implementation, the equipment fault early warning module runs on the edge computing gateway or server. It obtains sensor data (such as vibration, temperature, current) and maintenance records of the surgical equipment through the data interface module, calls an equipment fault classification model (such as a random forest model) for processing, and outputs equipment fault early warning information (e.g., {equipment ID: warning level: high, possible fault: bearing wear}).

[0034] The data interface module is used to obtain historical surgical data, current surgical information, and sensor data from external medical systems.

[0035] In practical implementation, the data interface module consists of a series of API interfaces and adapter programs, which are used to obtain historical surgical data, current surgical information and sensor data from the Hospital Information System (HIS), Laboratory Information System (LIS) and IoT sensors deployed on surgical equipment through HL7, FHIR or custom TCP protocols.

[0036] The model optimization module is used to collect operational performance data from the equipment usage prediction module, intelligent scheduling module, and equipment fault early warning module, and to optimize the model based on the operational performance data.

[0037] In practical implementation, the model optimization module is a background service program used to periodically collect operational performance data from system logs and business databases, such as the prediction accuracy of the equipment usage prediction module, the actual conflict rate of the scheme generated by the intelligent scheduling module, and the early warning accuracy of the equipment fault early warning module. Based on these data (such as prediction deviation exceeding 10%), it triggers parameter tuning or incremental learning of the corresponding time series prediction model, graph neural network model, and equipment fault classification model.

[0038] The demand forecast data output by the equipment usage prediction module, the surgical equipment scheduling plan output by the intelligent scheduling module, and the fault warning information output by the equipment fault warning module are synchronized to the equipment management module and the surgical management module via message queues (such as RabbitMQ) or database update events to guide equipment scheduling and surgical arrangements. For example, when the intelligent scheduling module generates a new plan, it updates the time field in the "Surgery-Equipment Association Table" and triggers the equipment management module to update the status of the relevant equipment to "pre-occupancy," thereby guiding subsequent equipment scheduling and surgical arrangements.

[0039] Based on the above, this system connects to external data sources through a data interface module, and the core business data model is constructed by the equipment management and surgery management modules. Three AI modules provide intelligent decision support from the dimensions of prediction, scheduling, and early warning, respectively. Their outputs are fed back to the business modules in real time, forming a decision-making closed loop, and the model optimization module ensures the continuous evolution of AI capabilities. This system achieves an automated process from data perception and intelligent analysis to decision execution, improving the accuracy of equipment management, optimizing scheduling efficiency, and enhancing the reliability of equipment operation.

[0040] Traditional equipment demand forecasting relies on human experience and cannot automatically learn and quantify future demand from massive and diverse historical data, resulting in coarse and inaccurate predictions. To address this issue and improve management reliability, as an optional implementation method in this embodiment, the equipment usage prediction module includes a feature extraction unit, a prediction model training unit, and a prediction execution unit.

[0041] The feature extraction unit is specifically used to perform time-series alignment and feature extraction on the surgical type, duration, and department information in historical surgical data and the equipment usage sequence in historical equipment usage records, based on timestamps, to construct a training dataset for equipment usage.

[0042] In practice, this unit first reads data from the database and associates a surgery with all the devices used, using the surgery ID and timestamp. Then, for each device, it aggregates the total usage time and number of times by day or hour, and combines this with the total number of surgeries that day, the proportion of each type of surgery, etc., to generate a time series sample containing multiple feature vectors. For example, the feature vector generated for an "ultrasound scalpel" device might be [date, number of general surgery surgeries that day, number of gynecological surgeries that day, total device usage time (minutes), number of times the device was used].

[0043] The prediction model training unit is specifically used to train a Long Short-Term Memory (LSTM) network model using the training dataset on the device, resulting in a trained time series prediction model.

[0044] Specifically, an LSTM model is constructed using the TensorFlow or PyTorch framework. This model consists of an input layer (receiving, for example, feature sequences from the past 7 days), two LSTM hidden layers (64 and 32 neurons respectively, with tanh activation function), and a fully connected output layer (outputting predicted device usage time for days 1 to 7). During training, mean squared error (MSE) is used as the loss function, the Adam optimizer is employed, and the learning rate is set to 0.001. Iterative training is performed on the training set until the loss function converges. In practical applications, other recurrent neural network structures such as GRU can also be used for the LSTM model; this embodiment does not limit this choice.

[0045] The prediction execution unit is specifically used to input current surgical scheduling data (i.e., information on future scheduled but not yet performed surgeries) into the time series prediction model, and output predicted demand data for each surgical device within the future time period. This unit converts the latest surgical scheduling data into feature sequences in the same format as the training data in real time, inputs them into the loaded LSTM model, and the model output is the predicted value of the demand for each device within a future period (e.g., one week).

[0046] Based on the above, the feature extraction unit transforms unstructured business data into structured features suitable for time series models. The LSTM model, with its powerful ability to capture temporal dependencies, can effectively learn the complex correlations between surgical schedules, departmental activities, and equipment usage. The prediction execution unit applies the model to real-time data, enabling quantitative and forward-looking predictions of future equipment needs. This solution transforms equipment preparation and procurement planning from experience-driven to data-driven, improving the accuracy of resource prediction.

[0047] Scheduling in scenarios involving multiple surgeries and multiple devices is a complex combinatorial optimization problem. Manual scheduling struggles to fully consider device status, time conflicts, and hierarchical dependencies between devices, easily leading to conflicts and resource waste. To address this issue and improve management reliability, as an optional implementation method in this embodiment, the intelligent scheduling module includes a relationship graph construction unit, a conflict analysis unit, and a solution generation unit.

[0048] The relational graph construction unit is specifically used to construct a scheduling relational graph based on a tree structure, current surgical information, and real-time equipment status. The nodes of the scheduling relational graph represent surgeries or equipment, and the edges represent equipment occupancy, time conflicts, or equipment hierarchical dependencies.

[0049] Specifically, the implementation involves creating a "surgery node" for each scheduled surgery and a "device node" for each available device (including primary and secondary devices). If a surgery is planned to use a particular device, an "occupancy" edge is established between the corresponding surgery node and device node. If two surgeries have overlapping time slots and both plan to use the same device, a "time conflict" edge is established between the two surgery nodes. Based on the tree structure in the device management module, a "hierarchical dependency" edge is established between the primary device node and secondary device nodes, indicating that the use of the secondary device depends on the primary device.

[0050] The conflict analysis unit is specifically used to input the scheduling relationship graph into a pre-trained graph neural network model, aggregate node information through graph convolution operations, and strengthen the weights of key edges using an attention mechanism to identify device occupancy conflicts and time conflicts.

[0051] Specifically, a Graph Attention Network (GAT) model is used. This model encodes features for each node in the graph (e.g., surgical node features include surgical duration and urgency; equipment node features include equipment type and current status). Information is passed and aggregated through multiple graph attention layers, with each layer calculating the attention coefficient between nodes. The formula is as follows: in, and It is a node and eigenvectors, It is a weight matrix. It is an attention vector. This indicates a splicing operation. It is a node The model uses the neighbor set of the neural network, with LeakyReLU as the activation function. Ultimately, the model outputs a conflict probability score for each "temporal conflict" edge, with values ​​closer to 1 indicating a higher risk of conflict.

[0052] The scheme generation unit is specifically used to generate a surgical equipment scheduling scheme by adjusting the allocation and timing of surgical equipment through a constraint optimization algorithm, based on the identified equipment occupancy conflicts and time conflicts, with the optimization goal of minimizing equipment idle time and balancing departmental load.

[0053] Specifically, this unit will output the conflict probability from the GAT model. As key inputs, define decision variables, objective function, and constraints, and then call an optimization solver (such as OR-Tools, CPLEX) to solve the problem. Objective function The study comprehensively considered equipment utilization and departmental equity. For example, a simplified objective function can be expressed as minimizing the weighted sum of total conflict risk and equipment idle time: ,in These are the weighting coefficients. Constraints include: each surgery must be allocated the required equipment; the same equipment can only be used by one surgery at a time; and sub-equipment must be allocated to the same surgery simultaneously with its main equipment. The solver outputs the optimized, conflict-free surgical equipment scheduling scheme.

[0054] Based on the above, the graph construction unit abstracts the complex scheduling problem into a graph data structure, clearly expressing the complex relationships between entities. The graph neural network model can efficiently and accurately uncover hidden conflict patterns in the graph, especially demonstrating a strong ability to identify indirect and potential conflicts. The solution generation unit quantifies the identified conflicts as input to an optimization problem, and automatically generates globally optimal or near-optimal scheduling solutions by solving the constrained optimization problem. This solution automates and intelligentizes the scheduling process, reduces manual intervention and scheduling conflicts, and improves the overall operational efficiency of the operating room.

[0055] Furthermore, in surgical equipment scheduling, it is necessary to organically combine the two objectives of minimizing equipment idle time and balancing departmental load with the "conflict risk" identified by graph neural networks, and perform global, mathematically precise optimization. Simple weighted summation or priority sorting is insufficient to achieve such a refined trade-off. Therefore, based on the aforementioned intelligent scheduling module, as a further preferred implementation of this embodiment, unlike the simplified objective function form, the constrained optimization algorithm includes: Step 1: Define decision variables , indicating surgery Is it in the device? time Start execution; Step 2: Define the objective function. The expression for the objective function is: in, For the total scheduling duration, For surgery Duration, The total number of devices. Total number of departments; Indicates equipment The average conflict score during the scheduling period. Devices output by the conflict analysis unit In time The conflict score (between 0 and 1); Indicates department Total surgical workload (hours) This indicates the average workload of each department (in hours). Belonging to the department A collection of surgical procedures; Indicates department The average conflict score of the scheduled surgeries is calculated based on the scheduling scheme. ) and the specific time conflict score of the equipment ( ); and A positive preset weighting coefficient is used to adjust the relative importance of conflict avoidance, equipment utilization and load balancing. Step 3: When solving the objective function using a constrained optimization algorithm, the following constraints must be satisfied: Constraint 1: Each surgery must be scheduled and only once. ; Constraint 2, Equipment Time Exclusivity Constraint: The same equipment can only be used by one surgery at a time. Its mathematical expression is: For any equipment... and any time ,have ; Constraint 3, Conflict Avoidance Constraint: If a time constraint exists... make Then forced That is, surgery is prohibited. In time Start using the equipment ,in The preset conflict threshold (e.g., 0.8); Constraint 4, Equipment Tree Structure Dependency Constraint: If surgery Sub-device used Then its main device must be used simultaneously. And the usage time is the same: .

[0056] This optimization problem is a mixed-integer nonlinear programming problem. In practice, it can be solved directly using mathematical programming solvers (such as Gurobi and Baron), or by linearizing / relaxing it, or by using metaheuristic algorithms (such as genetic algorithms and simulated annealing) to find high-quality feasible solutions. The solution generation unit obtains the decision variables by calling the aforementioned solvers. The optimal or near-optimal solution is obtained, thereby generating a specific surgical equipment scheduling scheme.

[0057] Based on the above, the core innovation of this optimization algorithm lies in the design of its objective function. It divides the device idle time penalty term by... This reduces the idle cost of devices with high conflict risk, causing the algorithm to tend to let these devices rest, thus proactively avoiding high-risk assignments. Simultaneously, it multiplies the departmental load balancing penalty by the department's average conflict score. This means that assigning more workload to high-conflict departments will incur greater penalties, thus prompting the workload to be shifted to low-conflict, smoother-running departments. The function applies a mild non-linear penalty to load deviations, avoiding extreme allocations. Instead of simply linearly weighting multiple objectives, the algorithm uses a conflict score... and As a dynamic coupling factor, conflict avoidance, a core security and efficiency requirement, is deeply integrated into and intelligently adjusts the trade-off between the two operational goals of improving utilization and balancing load, achieving synergistic optimization of the three and thus obtaining a safer, more efficient, and fairer global scheduling solution.

[0058] Equipment maintenance relies on fixed cycles or reactive repairs, making it impossible to provide early fault warnings based on the real-time operating status of the equipment. This can lead to unplanned downtime and affect surgical safety. To address this issue and improve management reliability, as an optional implementation method in this embodiment, the equipment fault warning module includes a feature fusion unit, a warning model training unit, and a warning output unit.

[0059] The feature fusion unit is specifically used to fuse the physical parameters in the sensor data with the maintenance and fault history in the maintenance records to construct a feature sample set of equipment health status.

[0060] Specifically, sensor data (such as vibration acceleration, temperature, and operating current) is acquired from the IoT gateway in time-series format. The feature fusion unit first preprocesses the raw sensor data (denoising and normalization), and then calculates time-domain features (such as mean, variance, and peak value) and frequency-domain features (such as the spectral centroid and mean square frequency obtained through Fast Fourier Transform). These features are concatenated with features extracted from maintenance records (such as the number of days since the last maintenance and the number of historical failures) to form a comprehensive feature vector representing a sample of the device's health status at a specific point in time.

[0061] The early warning model training unit is specifically used to train a random forest classification model or a gradient boosting decision tree classification model using a sample set of equipment health status features, so as to obtain a trained equipment fault classification model.

[0062] Taking random forest as an example, this unit uses the Scikit-learn library, setting parameters such as the number of decision trees in the forest (n_estimators=100) and the maximum depth (max_depth=10). Each sample in the training data is labeled as "normal" or "faulty" (fault categories can be further subdivided, such as "motor overheating" or "tool wear"). During training, the algorithm constructs multiple decision trees, each trained based on a different subset of sample features. In practical applications, this unit can also choose other classification models such as support vector machines or neural networks; this embodiment does not limit this choice.

[0063] The early warning output unit is specifically used to input the real-time collected sensor data into the equipment fault classification model. When the output fault probability exceeds the preset threshold, it generates fault warning information with equipment identification and warning level.

[0064] This unit periodically (e.g., every minute) acquires the latest sensor data. After undergoing the same feature extraction and fusion process as in the training phase, it is input into the loaded random forest model. The model outputs the probability of the device being in various fault states. If the probability of the "bearing wear" category exceeds a threshold (e.g., 0.7), an early warning message is generated: {device_id:“Device_001”,alert_level:“medium”,predicted_fault:“bearing_wear”,timestamp:“2023-10-2714:30:00”}, and pushed to the device administrator via the system message center or interface.

[0065] Based on the above, the feature fusion unit integrates multi-source heterogeneous data, providing the model with a comprehensive information foundation to describe the health status of the equipment. Ensemble learning models such as random forests can effectively handle high-dimensional features and have good robustness against noise and overfitting, making them suitable for fault classification based on complex sensor data. The early warning output unit realizes the automated conversion from real-time data to actionable early warning information. This solution transforms passive maintenance into proactive early warning, extending equipment lifespan and ensuring the smooth progress of surgery and patient safety.

[0066] Traditional software interfaces are not convenient or efficient enough, failing to allow for quick information retrieval or command issuance through natural means. To address this issue and improve management reliability and efficiency, as an optional implementation in this embodiment, the system further includes a natural language interaction module. The natural language interaction module is used for: It receives voice commands from users and converts them into text commands using automatic speech recognition technology; Natural language processing is performed on text commands to parse user intent and related device or surgical entities; Based on the parsed user intent, the system calls the device management module to query the device status or calls the surgery management module to operate on the surgical records. When the system calls the surgery management module to perform operations involving surgical time or equipment changes, or when the system calls the device management module to perform operations to modify the device status, the natural language interaction module triggers the intelligent scheduling module to perform conflict detection and update the surgical equipment scheduling plan.

[0067] In practical implementation, this module can be integrated into a mobile app or workstation. For example, when a user says "Query the status of the laparoscopy in operating room number three," the module first calls a cloud-based ASR (Automatic Speech Recognition) service provided by iFlytek or Baidu to convert the speech into the text "Query the status of the laparoscopy in operating room number three." Then, a Natural Language Understanding (NLU) component based on the BERT model is used to parse the text. This NLU component determines the user's intent as "Query the device status" using an intent recognition model and extracts the entities "operating room number three" and "laparoscopy" using a Named Entity Recognition (NER) model. Next, based on the intent and entities, the module constructs an API request to call the query interface of the device management module. If the user's instruction is "Postpone the first cholecystectomy tomorrow morning by one hour," the NLU parses the intent as "Modify the surgery time," and the entities as "the first cholecystectomy tomorrow morning" and "Postpone by one hour." The module calls the surgical management module's interface to modify the surgery time. After successful modification, it automatically sends a request to the intelligent scheduling module, triggering it to re-execute conflict detection for all related surgeries and equipment schedules that may be affected by the change in surgery time. If a new conflict (such as an equipment time conflict) is detected, the intelligent scheduling module will initiate a new round of optimization and update the global surgical equipment scheduling scheme to ensure that the system as a whole remains in a conflict-free state. The natural language interaction module communicates with the equipment management module and the surgical management module through a defined RESTful API. This application embodiment does not limit the specific type of communication protocol.

[0068] Based on the above, the natural language interaction module provides users with a more intuitive and convenient human-computer interaction method through speech recognition and natural language processing technologies. Furthermore, by intelligently linking natural language commands with the backend core business module and intelligent scheduling module, it ensures that any key business changes triggered by voice can automatically undergo conflict verification and global optimization, maintaining the consistency and optimality of the system scheduling scheme and avoiding new chaos introduced by convenient operation.

[0069] Creating or modifying surgical records is a complex process involving the coordination of multiple resources. Manual operation is cumbersome and prone to overlooking checks on doctor availability and equipment availability, leading to data errors or scheduling conflicts. To address this issue and improve management reliability and surgical scheduling management reliability, as a further optional implementation of this embodiment, based on the aforementioned natural language interaction module, the surgical management module includes: The device list creation unit is used to load the doctor list and the tree-structured list of available devices in parallel in response to requests to create or modify surgical records. The time availability verification unit is used to verify the availability of the surgeon at the selected surgery time in response to the user's selection of the surgeon. The occupancy conflict verification unit, in response to the user's selection of sub-devices through a tree structure, checks the current status of the selected sub-devices and their occupancy conflicts within the selected time period through the intelligent scheduling module; The surgical record management unit is used to save surgical records after verifying surgical information, physician availability, and equipment conflicts.

[0070] The specific implementation process is as follows: When the user clicks "Create Surgery" on the front-end interface, the equipment list creation unit and the occupancy conflict verification unit simultaneously initiate two asynchronous queries: 1) query the human resources system or local doctor database for a list of currently available doctors grouped by department; 2) query the equipment management module for a list of equipment organized in a tree structure with a status of "idle" or "pre-occupied," and display it on the front end as a collapsible tree component. After the user fills in basic information and selects a scheduled time, they select the surgeon from the doctor list. At this time, the time availability verification unit calls a "doctor's schedule service" interface to verify whether the doctor has other scheduled surgeries, meetings, or vacations within the user's selected surgery time slot, and provides real-time feedback on "available" or "unavailable."

[0071] The user then expands the desired primary device node in the device tree and selects one or more sub-devices under it. When the user confirms the device selection, the surgery management module collects all selected sub-device IDs, primary device IDs, and planned usage time periods, encapsulates them into a request, and sends it to the intelligent scheduling module's dedicated "conflict check" API. Upon receiving the request, the intelligent scheduling module quickly checks, based on the current global scheduling relationship graph (maintained by the relationship graph construction unit), whether each selected device is already occupied by other surgeries within the specified time period, and whether its primary device is available. It returns the conflict check results (listing all conflicting devices and their reasons) within milliseconds. Only when all the aforementioned verifications (basic information, doctor availability, and no device conflicts) pass, does the surgery management module persistently save this new surgery record to the database and trigger the intelligent scheduling module to formally incorporate this arrangement into the global scheduling graph. The surgery management module and the intelligent scheduling module achieve the above interaction through function calls or RPC (Remote Procedure Call). This application embodiment does not limit the specific interaction technology.

[0072] Based on the above, this process improves the efficiency of surgical record creation and user experience through parallel loading and real-time verification. By entrusting the complex task of checking for equipment conflicts to a specialized intelligent scheduling module, the authority of the verification and its consistency with the global scheduling status are guaranteed. This design ensures that every new or modified surgical record undergoes multi-dimensional consistency verification before being entered into the database, eliminating data conflicts and guaranteeing the accuracy and reliability of core business data.

[0073] The performance of artificial intelligence models degrades as data distribution changes (concept drift), and statically deployed models cannot maintain optimal performance in the long term, requiring continuous optimization. To address this issue and improve management reliability, as an optional implementation method in this embodiment, the model optimization module includes a data collection unit, an optimization analysis unit, and a model update unit.

[0074] The data collection unit is specifically used to collect prediction deviation data between the demand prediction data output by the equipment usage prediction module and the actual demand data, conflict frequency data of the surgical equipment scheduling scheme output by the intelligent scheduling module in actual execution, and accuracy data of fault warning information output by the equipment fault warning module.

[0075] Specifically, this unit collects data by monitoring system business logs and the results database. For example, it calculates the mean absolute percentage error (MAPE) between the "predicted demand" and "actual usage" for each type of equipment weekly as prediction deviation data. It also calculates the proportion of actual, unforeseen equipment conflicts after all surgeries in the intelligent scheduling scheme have been completed daily, as conflict frequency data. Finally, it calculates the proportion of confirmed equipment fault warnings by engineers monthly, as the warning accuracy rate.

[0076] The optimization analysis unit is specifically used to perform parameter tuning or incremental learning analysis on time series prediction models, graph neural network models, and equipment fault classification models based on prediction deviation data, conflict occurrence frequency data, and early warning accuracy data collected by the data collection unit.

[0077] Specifically, this unit sets performance thresholds (e.g., MAPE > 15%, conflict frequency > 5%, warning accuracy < 85%). When a model's evaluation metrics consistently fall below the threshold, the analysis unit determines that the model needs optimization. For predictive and classification models, the optimization analysis unit initiates a hyperparameter search process (e.g., using grid search or Bayesian optimization) to find better combinations of model parameters on new training datasets. For graph neural networks, in addition to parameter tuning, incremental learning can be initiated, using recently generated surgical scheduling data as new samples to fine-tune the original model.

[0078] The model update unit is specifically used to update the model, which has been optimized or incrementally learned, to the corresponding module based on the analysis results of the optimization analysis unit.

[0079] Specifically, after verifying that the new model's performance meets the standards in the test environment, this unit performs a hot model update operation. For example, on the AI ​​server, the old LSTM model file being loaded by the prediction module is replaced with a new model file, and the module is notified to reload the model by sending a signal or calling an API. The entire process does not affect the continuity of online services. The model update unit and each AI module can transfer and manage model files through shared storage (such as the Network File System NFS) or a model repository service. This application embodiment does not limit the specific update deployment strategy.

[0080] Based on the above, the data collection unit established a quantitative monitoring system for model performance. The optimization and analysis unit automatically determines optimization timing and strategies based on monitoring data, enabling intelligent decision-making for model maintenance. The model update unit achieves automated and seamless deployment of optimization results. These modules together construct a complete monitoring-analysis-optimization-deployment closed loop, enabling the entire AI system to continuously learn and self-evolve, ensuring its intelligence level can adapt to long-term business changes and maintain efficient and reliable operation.

[0081] In summary, the AI-based intelligent management system for surgical equipment in this embodiment, by constructing a collaborative system integrating equipment tree structure management, time-series model-based demand forecasting, intelligent scheduling combining graph neural networks and constraint optimization, classification model-based fault warning, and continuous model self-optimization, enables refined and intelligent management of the entire lifecycle of surgical equipment. This improves the utilization rate and scheduling efficiency of operating room equipment, ensures smooth surgical scheduling through accurate demand forecasting and automated conflict resolution, and enhances equipment reliability and surgical safety through early fault warning. Ultimately, it optimizes the allocation of medical resources and ensures the efficient and safe operation of the operating room.

[0082] Secondly, this embodiment provides an intelligent management method for surgical equipment based on artificial intelligence, which is applied to the aforementioned intelligent management of surgical equipment based on artificial intelligence. For example... Figure 2 As shown, the method includes the following steps in sequence: Obtain historical surgical data, current surgical information, and sensor data from surgical equipment from external medical systems; Surgical equipment is managed using a tree structure, which defines the hierarchical relationship between the main equipment and its child equipment. Manage surgical information and associate surgical information with surgical equipment in a tree structure according to surgical schedule; Based on historical surgical data and historical equipment usage records, equipment demand forecast data is output through a time series forecasting model. Based on the current surgical information and real-time equipment status, and combined with a tree structure, a conflict-free surgical equipment scheduling scheme is output through a graph neural network model. Based on sensor data and maintenance records of surgical equipment, fault warning information of surgical equipment is output through equipment fault classification model; Collect operational performance data corresponding to output device demand forecasting data, output surgical equipment scheduling schemes, and output fault warning information; and optimize time series prediction models, graph neural network models, and equipment fault classification models based on operational performance data. Among them, equipment demand forecast data, surgical equipment scheduling plans, and fault warning information are used to guide equipment scheduling and surgical arrangements.

[0083] Thirdly, this embodiment provides a computer-readable storage medium storing a computer program that can be executed by a processor. When the computer program is executed by the processor, the intelligent management method for surgical equipment based on artificial intelligence, as described above, is implemented.

[0084] It should be noted that the AI-based intelligent management method and storage medium for surgical equipment in this embodiment correspond to the aforementioned AI-based intelligent management system for surgical equipment. Therefore, any parts of the AI-based intelligent management method and storage medium that are not described in detail (including but not limited to specific technical means and effects) can be referred to the relevant descriptions in the aforementioned AI-based intelligent management system for surgical equipment, and will not be repeated here.

[0085] In the embodiments provided in this application, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any suitable combination thereof. For hardware implementation, the processor may be implemented in one or more of the following: application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, other electronic units designed to implement the functions described herein, or combinations thereof. For software implementation, some or all of the processes of the embodiments may be performed by a computer program instructing the associated hardware. During implementation, the program may be stored in a computer-readable storage medium or transmitted as one or more instructions or code on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of a computer program from one place to another. Storage media may be any available medium accessible to a computer. Computer-readable storage media may include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code having the form of instructions or data structures and accessible to a computer.

[0086] Finally, it should be noted that the above description is only a preferred embodiment of this application and is not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An intelligent management system for surgical equipment based on artificial intelligence, characterized in that, The system includes: The equipment management module is used to manage surgical equipment in a tree structure, which defines the hierarchical relationship between the main equipment and its sub-equipment. The surgery management module is used to manage surgical information and associate the surgical information with the surgical equipment in the equipment management module according to the surgical schedule; The equipment usage prediction module is used to output equipment demand prediction data based on historical surgical data and historical equipment usage records through a time series prediction model; The intelligent scheduling module is used to output a conflict-free surgical equipment scheduling scheme based on the current surgical information and the real-time status of the equipment, combined with the tree structure, through a graph neural network model. The equipment fault early warning module is used to output fault early warning information of surgical equipment based on sensor data and maintenance records of surgical equipment and through an equipment fault classification model. The data interface module is used to obtain the historical surgical data, current surgical information, and sensor data from external medical systems; The model optimization module is used to collect the operational performance data of the equipment usage prediction module, the intelligent scheduling module, and the equipment fault early warning module, and to optimize the model based on the operational performance data. The demand forecast data output by the device prediction module, the surgical equipment scheduling plan output by the intelligent scheduling module, and the fault warning information output by the device fault warning module are synchronized to the device management module and the surgical management module to guide device scheduling and surgical arrangements.

2. The intelligent management system for surgical equipment based on artificial intelligence according to claim 1, characterized in that, The device usage prediction module includes: The feature extraction unit is used to perform time-series alignment and feature extraction on the surgical type, duration and department information in the historical surgical data and the equipment usage sequence in the historical equipment usage record according to the timestamp, and to construct the equipment usage training dataset. The prediction model training unit is used to train a long short-term memory network model using the training dataset on the device to obtain a trained time series prediction model. The prediction execution unit is used to input the current surgical scheduling data into the time series prediction model and output the predicted demand data for each surgical device in the future time period.

3. The intelligent management system for surgical equipment based on artificial intelligence according to claim 1, characterized in that, The intelligent scheduling module includes: The relationship graph construction unit is used to construct a scheduling relationship graph based on the tree structure, current surgical information and real-time equipment status, wherein the nodes of the scheduling relationship graph represent surgery or equipment, and the edges represent equipment occupancy, time conflict or equipment hierarchical dependency relationship. The conflict analysis unit is used to input the scheduling relationship graph into a pre-trained graph neural network model, aggregate node information through graph convolution operations, and strengthen the weight of key edges using an attention mechanism to identify device occupancy conflicts and time conflicts. The scheme generation unit is used to generate the surgical equipment scheduling scheme by adjusting the allocation and timing of surgical equipment through a constraint optimization algorithm based on the identified equipment occupancy conflicts and time conflicts, with the optimization goal of minimizing equipment idle time and balancing departmental load.

4. The intelligent management system for surgical equipment based on artificial intelligence according to claim 1, characterized in that, The equipment fault early warning module includes: The feature fusion unit is used to perform feature fusion on the physical parameters in the sensor data and the maintenance and fault history in the maintenance record to construct a feature sample set of equipment health status; The early warning model training unit is used to train a random forest classification model or a gradient boosting decision tree classification model using the equipment health status feature sample set to obtain a trained equipment fault classification model. The early warning output unit is used to input the real-time collected sensor data into the equipment fault classification model. When the output fault probability exceeds a preset threshold, it generates fault early warning information with equipment identification and early warning level.

5. The intelligent management system for surgical equipment based on artificial intelligence according to claim 1, characterized in that, The system also includes a natural language interaction module; the natural language interaction module is used for: It receives voice commands from users and converts them into text commands using automatic speech recognition technology; Natural language processing is performed on the text instructions to parse the user's intent and the associated device or surgical entity; Based on the parsed user intent, the device management module is invoked to query the device status, or the surgical management module is invoked to operate the surgical record; wherein, when the surgical management module is invoked to perform an operation involving surgical time or participating in equipment changes, or when the device management module is invoked to perform an operation to modify the device status, the natural language interaction module triggers the intelligent scheduling module to perform conflict detection and update the surgical equipment scheduling scheme.

6. The intelligent management system for surgical equipment based on artificial intelligence according to claim 5, characterized in that, The surgical management module includes: The device list creation unit is used to load the doctor list and the available device list based on the tree structure in parallel in response to a request to create or modify a surgical record; A time availability verification unit is used to verify the availability of the surgeon at the selected surgery time in response to the user's selection of the surgeon. The occupancy conflict verification unit, in response to the user's selection of a sub-device through the tree structure, checks the current status of the selected sub-device and its occupancy conflicts within a selected time period through the intelligent scheduling module; The surgical record management unit is used to save surgical records after verifying surgical information, physician availability, and equipment conflicts.

7. The intelligent management system for surgical equipment based on artificial intelligence according to claim 1, characterized in that, The model optimization module includes: The data collection unit is used to collect the prediction deviation data between the demand prediction data output by the equipment prediction module and the actual demand data, the conflict frequency data of the surgical equipment scheduling scheme output by the intelligent scheduling module in actual execution, and the accuracy data of the fault warning information output by the equipment fault warning module. The optimization analysis unit is used to perform parameter tuning or incremental learning analysis on the time series prediction model, the graph neural network model, and the equipment fault classification model based on the prediction deviation data, conflict occurrence frequency data, and early warning accuracy data collected by the data collection unit. The model update unit is used to update the model after parameter tuning or incremental learning to the corresponding module based on the analysis results of the optimization analysis unit.

8. The intelligent management system for surgical equipment based on artificial intelligence according to claim 3, characterized in that, The constrained optimization algorithm includes: Step 1: Define decision variables , indicating surgery Is it in the device? time Start execution; Step 2: Define the objective function, the expression of which is: in, For the total scheduling duration, For surgery Duration, For the total number of devices, Total number of departments; Indicates equipment The average conflict score during the scheduling period. The device output by the conflict analysis unit In time Conflict score; Indicates department Total surgical load, This indicates the average departmental workload. Belonging to the department A collection of surgical procedures; Indicates department The average conflict score of the scheduled surgeries; and The preset weighting coefficient is positive; Step 3: When the constrained optimization algorithm solves the objective function, the following constraints must be satisfied: Constraint 1: Each surgery must be scheduled and only once. ; Constraint 2, Equipment Time Exclusivity Constraint: The same equipment can only be used by one surgery at a time; Constraint 3, Conflict Avoidance Constraint: If a time constraint exists... make Surgery is prohibited if... In time Start using the equipment ,in The preset conflict threshold; Constraint 4, Equipment Tree Structure Dependency Constraint: If surgery Sub-device used Then its main device must be used simultaneously. And the usage time is the same: .

9. A method for intelligent management of surgical equipment based on artificial intelligence, characterized in that, The method includes the following steps: Obtain historical surgical data, current surgical information, and sensor data from surgical equipment from external medical systems; Surgical equipment is managed using a tree structure, which is used to define the hierarchical relationship between the main equipment and its sub-equipment. Manage surgical information and associate the surgical information with the surgical equipment in the tree structure according to the surgical schedule; Based on the historical surgical data and historical equipment usage records, equipment demand prediction data is output through a time series prediction model. Based on the current surgical information and real-time equipment status, and combined with the tree structure, a conflict-free surgical equipment scheduling scheme is output through a graph neural network model. Based on the sensor data and maintenance records of the surgical equipment, fault warning information of the surgical equipment is output through the equipment fault classification model; Collect the output device demand forecast data, output surgical equipment scheduling scheme, and output fault early warning information corresponding to the operation effect data, and optimize the time series prediction model, graph neural network model, and equipment fault classification model based on the operation effect data; The equipment demand forecast data, the surgical equipment scheduling plan, and the fault warning information are used to guide equipment scheduling and surgical arrangements.

10. A computer-readable storage medium, characterized in that, It stores a computer program that can be executed by a processor. When the computer program is executed by the processor, it implements the intelligent management method for surgical equipment based on artificial intelligence as described in claim 9.

Images