Surgical instrument configuration prediction method and system based on fusion rule engine and machine learning, terminal and storage medium

Abstract

The invention discloses a surgical instrument configuration prediction method and system based on a fusion rule engine and machine learning, a terminal and a storage medium, and the method comprises the steps: obtaining a surgical style and an operator of a target operation, collecting corresponding multi-dimensional data resources, and obtaining structured standard data; the method comprises the following steps: constructing a standard surgical style-instrument configuration table based on clinical expert knowledge, converting the constructed standard surgical style-instrument configuration table by using a rule template language to obtain a rule execution template, and matching the surgical style with the rule execution template to generate a preliminary instrument recommendation list; inputting the structured standard data into the model for deep learning and prediction to obtain an AI prediction result, and comparing the AI prediction result with the preliminary instrument recommendation list to obtain an adjusted recommendation list; and scoring the instruments in the adjusted recommendation list by using a strategy, and obtaining a final recommendation instrument list according to a scoring result. The instrument configuration quality is improved, the preoperative preparation efficiency is improved, and intelligent development of hospital operation is promoted.

Description

Technical Field

[0001] The present invention relates to the field of medical equipment scheduling, and in particular to a surgical instrument configuration prediction method, system, terminal and computer-readable storage medium based on a fusion rule engine and machine learning. Background Art

[0002] Surgical instrument configuration is the process of scientifically planning and preparing the instruments required for surgery based on the type of surgery, operating steps and patient needs. It is the core link of operating room management and directly affects surgical efficiency, patient safety and medical quality. With the continuous growth of hospital surgeries, how to achieve the pre-positioning and intelligentization of surgical preparation has become the focus of industry attention.

[0003] Reasonable instrument configuration is the basis for surgical team collaboration and needs to be continuously optimized in combination with clinical experience and standardized processes. Currently, the configuration of surgical instruments mainly relies on standard procedure-instrument templates. That is, for a certain procedure, the system calls a fixed instrument package configuration. However, the template fixation of this mechanism lacks flexibility, the configuration is prone to redundancy or loss, and the adaptability is weak. Once the preoperative schedule is adjusted, the template system cannot quickly adapt to the new surgeon and procedure combination, and factors such as cost, efficiency, and resource constraints are not comprehensively considered. There is a lack of more accurate, personalized, and economical surgical instrument configuration solutions.

[0004] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0005] The main purpose of the present invention is to provide a surgical instrument configuration prediction method, system, terminal and computer-readable storage medium based on a fusion rule engine and machine learning, aiming to solve the problem in the existing technology that the configuration template of surgical instruments is fixed and lacks flexibility, the configuration is easily redundant or missing, and the cost, efficiency, resource constraints and other factors are not comprehensively considered, resulting in the lack of more accurate, personalized and economical surgical instrument configuration.

[0006] To achieve the above objectives, the present invention provides a method for predicting surgical instrument configuration based on a fusion rule engine and machine learning, the method comprising the following steps:

[0007] Obtain the surgical procedure and surgeon for the target surgery, collect multidimensional data resources corresponding to the surgical procedure and surgeon from various hospital business systems, parse and integrate the multidimensional data resources to obtain structured standard data;

[0008] Constructing a standard procedure-instrument configuration table based on clinical expert knowledge, converting the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, matching the procedure with the rule execution template, and generating a preliminary instrument recommendation list;

[0009] Inputting the structured standard data into multiple machine learning models for deep learning and prediction to obtain AI prediction results, and using the AI ​​prediction results to compare with the preliminary device recommendation list to obtain an adjusted recommendation list;

[0010] A strategy evaluation engine is used and a weight coefficient is introduced to score each device in the adjusted recommended list, and a final recommended device list is obtained based on the scoring results.

[0011] Optionally, in the surgical instrument configuration prediction method based on a fusion rule engine and machine learning, the multidimensional data resources include preoperative scheduling data, surgical procedure definitions, surgeon information, and intraoperative usage behavior;

[0012] The preoperative scheduling data includes procedure number, operator ID, time window and operation room allocation;

[0013] The procedure definition includes the procedure name, classification, complexity, recommended duration, standard instrument template ID and alternative procedures;

[0014] The surgeon information includes the surgeon's department, surgical experience, historical surgical records, and intraoperative instrument behavior vectors;

[0015] The intraoperative usage behaviors include instrument addition and subtraction records, bag change behaviors, actual usage lists and abnormal usage situations.

[0016] Optionally, the surgical instrument configuration prediction method based on the fusion of rule engine and machine learning, wherein the standard surgical procedure-instrument configuration table is constructed based on clinical expert knowledge, and the constructed standard surgical procedure-instrument configuration table is converted using a rule template language to obtain a rule execution template, specifically includes:

[0017] Based on the hospital's clinical pathway, team consensus, and surgical SOP documents, combined with clinical expert experience, define the standard instrument set corresponding to each surgical procedure and the necessity level of each instrument;

[0018] Collecting statistics on the actual use of the devices to obtain statistical data, and using the statistical data to assist in verifying the necessity level of each device to obtain the true necessity level of each device;

[0019] A standard procedure-instrument configuration table is constructed based on the standard instrument set corresponding to each procedure and the actual necessity level of each instrument;

[0020] The constructed standard surgical procedure-instrument configuration table is converted using the Drools rule language to obtain an IF-THEN rule execution template.

[0021] Optionally, in the surgical instrument configuration prediction method based on a fusion rule engine and machine learning, the multiple machine learning models include a preference scoring model, an alternative instrument recommendation model, an instrument usage probability model, and an instrument combination model;

[0022] The AI ​​prediction results include a device recommendation score table, recommended alternative devices, a Top-N combination of high-priority devices, and the probability and confidence index of each device's use.

[0023] Optionally, the surgical instrument configuration prediction method based on the fusion rule engine and machine learning, wherein the structured standard data is input into multiple machine learning models for deep learning and prediction to obtain AI prediction results, specifically includes:

[0024] Inputting the integrated operator information and integrated intraoperative usage behavior into the preference scoring model for learning, obtaining the operator's usage preference scoring matrix for each instrument under a specific surgical procedure, and obtaining an instrument recommendation scoring table based on the usage preference scoring matrix;

[0025] Inputting the integrated surgical procedure definition and the integrated intraoperative usage behavior into the instrument usage probability model for learning, obtaining the usage probability distribution of each instrument under a given surgical procedure, and obtaining the usage probability of each instrument;

[0026] Inputting the integrated operator information, integrated procedure definition, and integrated intraoperative use behavior into the alternative device recommendation model for learning to obtain recommended alternative devices with substitution relationships;

[0027] Inputting the integrated operator information, integrated procedure definition, and integrated intraoperative usage behavior into the instrument combination model for learning to obtain a high-frequency linkage instrument combination that completes the implicit relationship, and outputting the high-frequency linkage instrument combination in a Top-N manner to obtain a Top-N combination of high-priority instruments;

[0028] Obtain historical procedure records of the operator's information, obtain the operator's historical procedure number based on the historical procedure records, and calculate the confidence index based on the historical procedure number and a preset procedure number threshold by combining relative proportion and Sigmoid function.

[0029] Optionally, the surgical instrument configuration prediction method based on a fusion rule engine and machine learning, wherein the use of the AI ​​prediction results to compare the preliminary instrument recommendation list to obtain an adjusted recommendation list specifically includes:

[0030] Obtaining from the device recommendation scoring table devices that have a score higher than a preset score threshold and are not in the preliminary device recommendation list as devices to be added;

[0031] Obtaining from the device recommendation score table devices whose scores are not higher than the preset score threshold, whose usage probabilities are lower than the preset probability threshold, and which exist in the preliminary device recommendation list as devices to be deleted;

[0032] Selecting a device with a score higher than the preset score threshold from the recommended alternative devices as the device to be replaced;

[0033] Rearranging the actual necessity level of each device in the preliminary device recommendation list according to the Top-N combination of high-priority devices to obtain a device recommendation list after sorting optimization;

[0034] The preliminary device recommendation list is adjusted according to the devices to be added, the devices to be deleted, the devices to be replaced, and the sorted and optimized device recommendation list to generate an adjusted recommendation list.

[0035] Optionally, the surgical instrument configuration prediction method based on the fusion of rule engine and machine learning, wherein the strategy evaluation engine is used and a weight coefficient is introduced to score each instrument in the adjusted recommendation list, and a final recommended instrument list is obtained according to the scoring results, specifically including:

[0036] Calculating the cost score, efficiency score, safety performance score, and operator satisfaction matching of the adjusted recommendation list through a strategy optimizer, and calculating the confidence score of the adjusted recommendation list based on the confidence index;

[0037] Setting different weights for the cost score, the efficiency score, the safety performance score, the operator satisfaction match, and the confidence score, respectively, and performing weighted fusion of the cost score, the efficiency score, the safety performance score, the operator satisfaction match, and the confidence score according to the weights to obtain a comprehensive score;

[0038] If the comprehensive score exceeds the preset comprehensive score threshold, the adjusted recommendation list will be used as the final recommended equipment list; if the comprehensive score does not exceed the preset comprehensive score threshold, the preliminary equipment recommendation list will be used as the final recommended equipment list.

[0039] In addition, to achieve the above-mentioned purpose, the present invention further provides a surgical instrument configuration prediction system based on a fusion rule engine and machine learning, wherein the surgical instrument configuration prediction system based on a fusion rule engine and machine learning includes:

[0040] The data integration module is used to obtain the surgical procedure and surgeon of the target surgery, collect the multi-dimensional data resources corresponding to the surgical procedure and surgeon from various business systems of the hospital, and parse and integrate the multi-dimensional data resources to obtain structured standard data;

[0041] A rule engine module is used to construct a standard procedure-instrument configuration table based on clinical expert knowledge, convert the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, match the procedure with the rule execution template, and generate a preliminary instrument recommendation list;

[0042] a machine learning module, configured to input the structured standard data into a plurality of machine learning models for deep learning and prediction to obtain AI prediction results, and compare the AI ​​prediction results with the preliminary device recommendation list to obtain an adjusted recommendation list;

[0043] The decision fusion module is used to use a strategy evaluation engine and introduce a weight coefficient to score each device in the adjusted recommendation list, and obtain a final recommended device list based on the scoring results.

[0044] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a surgical instrument configuration prediction program based on a fusion rule engine and machine learning stored on the memory and runnable on the processor, wherein the surgical instrument configuration prediction program based on a fusion rule engine and machine learning is executed by the processor to implement the steps of the surgical instrument configuration prediction method based on a fusion rule engine and machine learning as described above.

[0045] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a surgical instrument configuration prediction program based on a fusion rule engine and machine learning, and when the surgical instrument configuration prediction program based on a fusion rule engine and machine learning is executed by a processor, the steps of the surgical instrument configuration prediction method based on a fusion rule engine and machine learning as described above are implemented.

[0046] In the present invention, the surgical procedure and surgeon of the target surgery are obtained, the corresponding multi-dimensional data resources are collected, and structured standard data is obtained; based on clinical expert knowledge, a standard surgical procedure-instrument configuration table is constructed, and the constructed standard surgical procedure-instrument configuration table is converted using a rule template language to obtain a rule execution template, and the surgical procedure is matched with the rule execution template to generate a preliminary instrument recommendation list; the structured standard data is input into a machine learning model for deep learning and prediction to obtain AI prediction results, and the AI ​​prediction results are compared with the preliminary instrument recommendation list to obtain an adjusted recommendation list; a strategy evaluation engine is used and a weight coefficient is introduced to score the instruments in the adjusted recommendation list, and the final recommended instrument list is obtained according to the scoring results. The present invention significantly improves the quality of instrument configuration, enhances the efficiency of preoperative preparation, and promotes the intelligent development of hospital operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a flow chart of a preferred embodiment of the surgical instrument configuration prediction method based on the fusion rule engine and machine learning of the present invention;

[0048] Figure 2 This is a schematic diagram of the principle structure of the surgical instrument configuration prediction method based on the fusion rule engine and machine learning of the present invention;

[0049] Figure 3 This is a structural diagram of a preferred embodiment of a surgical instrument configuration prediction system based on a fusion rule engine and machine learning according to the present invention;

[0050] Figure 4 FIG. 4 is a structural diagram of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION

[0051] The present application provides a method and system for predicting surgical instrument configuration based on a fusion rule engine and machine learning, aiming to solve the problem in the prior art that the configuration template of surgical instruments is fixed and lacks flexibility, the configuration is easily redundant or missing, and factors such as cost, efficiency, and resource constraints are not comprehensively considered, resulting in the lack of more accurate, personalized, and economical surgical instrument configuration. In order to make the purpose, technical solutions, and effects of this application clearer and more specific, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit this application.

[0052] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0053] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features specified as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0054] The surgical instrument configuration prediction method based on the fusion rule engine and machine learning described in the preferred embodiment of the present invention is as follows: Figure 1 and Figure 2 As shown, the surgical instrument configuration prediction method based on the fusion rule engine and machine learning includes the following steps:

[0055] Step S10: Obtain the surgical procedure and surgeon of the target surgery, collect multidimensional data resources corresponding to the surgical procedure and surgeon from various business systems of the hospital, parse and integrate the multidimensional data resources, and obtain structured standard data.

[0056] In this example, the surgical procedure refers to the specific method or type of surgical operation. Based on the disease type, patient condition, and medical technology selected, it can be categorized by technique, purpose, or location. For example, the technique classification can be categorized as laparoscopic surgery, open surgery, and robotic-assisted surgery. The surgical procedure determines the instruments, consumables, equipment, and personnel required for the surgery.

[0057] The surgeon refers to the main executor of the operation, usually the surgeon who is responsible for formulating the surgical plan and completing the core operations. The surgeon can include the surgeon, assistant surgeon, etc. It is understandable that different surgeons may have personal preferences in the use of instruments (such as specific instruments or specific instrument brands).

[0058] Specifically, the surgical procedure information and surgeon information of the target surgery are obtained, and the multidimensional data resources corresponding to the surgical procedure and the surgeon are collected and integrated from various business systems of the hospital. The multidimensional data resources include preoperative scheduling data, surgical procedure definition, surgeon information and intraoperative usage behavior.

[0059] Among them, the preoperative scheduling data includes procedure number, surgeon ID, time window and inter-operative room allocation; in this embodiment, the daily surgical scheduling plan is accessed from the HIS (Hospital Information System) to extract basic scheduling information such as procedure number, surgeon ID, time window, and inter-operative room allocation.

[0060] Furthermore, the procedure definition includes the procedure name, classification, complexity, recommended duration, standard instrument template ID and alternative procedures; in this embodiment, the system extracts structured procedure definition information based on procedure coding, including procedure name, classification, complexity, recommended duration, standard instrument template ID, alternative procedures, etc., to provide procedure features for rule matching and model building.

[0061] It can be understood that the procedure definition refers to the structured description of the surgical procedure after standardized modeling for each type of surgery. The fields included are shown in Table 1:

[0062] Table 1: Field table of term definition

[0063]

[0064]

[0065] Furthermore, the surgeon information includes the surgeon's department, surgical experience, historical surgical records, and intraoperative instrument behavior vectors. In this embodiment, the doctor master data management system is called to obtain data such as the surgeon's department, surgical experience, historical surgical records, and intraoperative instrument behavior vectors to construct the input basis for the surgeon's personalized preference model.

[0066] It is understandable that surgeon information is not just name and department, but also includes multidimensional portrait data that can be used for modeling and decision-making. The fields included are shown in Table 2:

[0067] Table 2: Operator information field table

[0068] Field Name meaning use Operator ID The doctor's unique code in the hospital system Used for behavior modeling and recommendation binding Name For display (not involved in modeling) UI layer display Department Such as gastrointestinal surgery, orthopedics, etc. Template distribution and procedure scope limitation Surgical experience level Senior / attending / resident physicians, etc. Used for recommendation model weight setting Historical surgical records Surgical procedures performed and their frequency Constructing a procedure-instrument preference model Equipment addition and subtraction records What instruments are commonly added / removed during surgery? Constructing the operator preference behavior vector Using a preference curve Personalized device scoring data output by AI training Integrate into the fusion recommendation engine

[0069] Furthermore, the intraoperative usage behavior includes instrument addition and subtraction records, package change behavior, actual usage list and abnormal usage; in this embodiment, the system is connected to the surgical nursing record system and the instrument package traceability platform to extract intraoperative instrument addition and subtraction records, package change behavior, actual usage list, abnormal usage, etc., and store them in association with the operator ID and procedure number for model training and rule correction.

[0070] It is understandable that the intraoperative use behavior is associated with the above-mentioned intraoperative use behavior. The sources of the intraoperative use behavior include the intraoperative nursing record system, postoperative instrument usage statistics, and bag change documents, etc. The fields contained therein are shown in Table 3:

[0071] Table 3: Field table of intraoperative behavior

[0072]

[0073] Furthermore, the multi-dimensional data resources are parsed and integrated, converted into a standard format, and a structured data interface is output to obtain structured standard data for call by the rule engine and AI model module, ensuring high-quality data support and closed-loop capabilities of the entire system.

[0074] Step S20: construct a standard procedure-instrument configuration table based on clinical expert knowledge, convert the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, match the procedure with the rule execution template, and generate a preliminary instrument recommendation list.

[0075] The method of constructing a standard surgical procedure-instrument configuration table based on clinical expert knowledge and converting the constructed standard surgical procedure-instrument configuration table using a rule template language to obtain a rule execution template specifically includes:

[0076] Based on the hospital's clinical pathway, team consensus, and surgical SOP documents, combined with clinical expert experience, define the standard instrument set corresponding to each surgical procedure and the necessity level of each instrument;

[0077] Collecting statistics on the actual use of the devices to obtain statistical data, and using the statistical data to assist in verifying the necessity level of each device to obtain the true necessity level of each device;

[0078] A standard procedure-instrument configuration table is constructed based on the standard instrument set corresponding to each procedure and the actual necessity level of each instrument;

[0079] The constructed standard surgical procedure-instrument configuration table is converted using the Drools rule language to obtain an IF-THEN rule execution template.

[0080] It is understandable that the standard procedure-instrument configuration table is constructed by a multidisciplinary clinical expert group based on various procedure SOPs, actual instrument usage experience, and statistical data. The table defines the corresponding instrument item list, recommendation level (the recommendation level includes mandatory, recommended, and optional), quantity recommendations, and usage conditions for each procedure. The standard procedure-instrument configuration table supports multi-dimensional matching logic such as the complexity of the procedure, the level of the surgeon, and the risk level. Furthermore, a rule template language, such as Drools DSL (Domain Specific Language), is used to convert the configuration table into an IF-THEN formal rule. The IF-THEN rule execution template has the capabilities of rule priority control, conditional judgment branching, and abnormal procedure coverage, supporting flexible expansion.

[0081] Among them, the Drools rule language is an open source rule engine language based on the RETE algorithm, which is used to decouple business logic from code and supports the definition, management and execution of complex business rules. It is developed based on Java, compatible with natural language and programming syntax, and is often used for rule management in enterprise-level applications (such as risk control, compliance, automated decision-making, etc.). Its core is to describe business policies through rules, achieve decoupling from application code, and support dynamic rule updates and efficient reasoning. The IF-THEN rule execution template is a process control tool based on logical conditional judgment. It is used to define rules that "trigger corresponding operations when specific conditions are met". The core structure is the causal relationship between conditional judgment and operation execution. The medical field is a key application area of ​​the IF-THEN rule execution template.

[0082] Furthermore, the surgical procedure is matched with the rule execution template to generate a preliminary device recommendation list.

[0083] In this embodiment, the surgical procedures are received from the data integration layer as input of a preliminary instrument recommendation list; a preliminary instrument recommendation list (V1) is generated through a rule matching mechanism, and is accompanied by instrument grade labels for subsequent fusion module optimization.

[0084] Step S30: Input the structured standard data into multiple machine learning models for deep learning and prediction to obtain AI prediction results, and use the AI ​​prediction results to compare with the preliminary device recommendation list to obtain an adjusted recommendation list.

[0085] In this embodiment, the multiple machine learning models include a preference scoring model, an alternative device recommendation model, a device usage probability model, and a device combination model; the AI ​​prediction results include a device recommendation score table, recommended alternative devices, a top-N combination of high-priority devices, and the usage probability and confidence index of each device. It should be noted that in another implementation, a statistical model can also be used instead of a machine learning model.

[0086] Among them, the preference scoring model is a dual-tower cross-attention network, which is a deep learning architecture for modeling multimodal or multi-source data interactions. It is widely used in recommendation systems, search ranking, medical data analysis and other scenarios. The core idea is to extract features from different fields through independent feature encoding towers (Tower), and then dynamically capture the implicit associations between features through a cross-attention mechanism (Cross-Attention), and finally fuse them to generate a high-dimensional semantic representation, independently encode user and device features to avoid feature confusion, including a cross-attention mechanism for capturing the dynamic interactive relationship between features and device characteristics. In the present invention, this network is used because it shows significant advantages in scenarios such as personalized medical recommendations and risk assessment through a balanced design of decoupled encoding and dynamic interaction.

[0087] The alternative device recommendation model is used to recommend alternatives with similar functions that meet the surgeon's preferences when the target device is unavailable. The model architecture of the alternative device recommendation model is a graph neural network combined with contrastive learning. The recommendation strategies used include Top-K recall and re-ranking. Top-K recall selects the top five devices with the highest cosine similarity to the target device in the graph embedding space. Re-ranking combines a preference scoring model with secondary ranking to ensure that the alternative devices meet the surgeon's individual needs.

[0088] The instrument use probability model uses a LightGBM model enhanced with time series features. The LightGBM model is an improved version of the gradient boosting decision tree, which gradually reduces the prediction error by iteratively constructing a series of decision trees. Each step constructs a new decision tree to reduce the error of the previous stage, and ultimately obtains an accurate prediction model. In this embodiment, it is used to predict the probability of a surgeon using a certain surgical instrument. The LightGBM model enhanced with time series features identifies key influencing factors through SHAP values ​​to provide explanations for the decision.

[0089] The instrument combination model is composed of a reinforcement learning PPO algorithm and a physical simulator, and introduces safety constraint mechanisms: physical simulator pre-verification and action mask; physical simulator pre-verification: detecting dangerous actions (such as excessive joint angle movement) through biomechanical simulation; action mask: directly prohibiting instrument strength combinations that may cause injury.

[0090] Specifically, the integrated operator information and the integrated intraoperative usage behavior are input into the preference scoring model for learning to obtain the operator's usage preference scoring matrix for each instrument under a specific surgical procedure, and an instrument recommendation scoring table is obtained based on the usage preference scoring matrix.

[0091] It can be understood that the preference scoring model is used to receive the historical surgical procedure records in the integrated surgeon information, as well as the integrated records of adding and subtracting instruments in the intraoperative usage behavior, and output the surgeon's preference scoring matrix for the use of each instrument under a specific surgical procedure for personalized recommendation sorting.

[0092] Furthermore, the integrated surgical procedure definition and the integrated intraoperative usage behavior are input into the instrument usage probability model for learning, thereby obtaining the usage probability distribution of each instrument under a given surgical procedure and the usage probability of each instrument.

[0093] It can be understood that the instrument usage probability model is used to receive the integrated procedure definition and the actual usage list in the integrated intraoperative usage behavior, and output the probability distribution of instrument usage under a given procedure, which is used to identify low-frequency redundant instruments and mark them as optional items.

[0094] Furthermore, the integrated operator information, the integrated procedure definition and the integrated intraoperative usage behavior are input into the alternative instrument recommendation model for learning to obtain recommended alternative instruments with substitution relationships.

[0095] The integrated operator information, integrated procedure definition and integrated intraoperative usage behavior are input into the instrument combination model for learning to obtain a high-frequency linkage instrument combination that completes the implicit relationship. The high-frequency linkage instrument combination is output in a Top-N manner to obtain a Top-N combination of high-priority instruments.

[0096] Specifically, the integrated surgeon information, integrated procedure definition, and integrated intraoperative use behaviors are first standardized and encoded to obtain target surgeon information. The integrated procedure definition undergoes text embedding and value mapping, and semantic vectors of the surgical type and key steps are extracted using BERT to obtain the target procedure definition. Time series features are constructed for the integrated intraoperative use behaviors, and context vectors of the intraoperative use behaviors are extracted using Transformer to obtain the target intraoperative use behaviors. Next, implicit relationship completion and model training are performed. Multimodal feature fusion is performed on the target surgeon information, the target procedure definition, and the target intraoperative use behaviors. Structured surgeon information and unstructured procedure descriptions are uniformly encoded to achieve multi-source heterogeneous data fusion. Based on the fused multimodal features, a loss function is designed for dynamic weight learning to obtain an optimized model. This optimized model is used to mine high-frequency linkage patterns and obtain high-frequency linkage instrument combinations that complete the implicit relationships. Then, based on the candidate combination generation rules, these high-frequency linkage instrument combinations are output in a top-N manner to obtain the top-N combinations of high-priority instruments.

[0097] Among them, the Top-N method is a data sorting and recommendation strategy, which refers to screening out the top N results with the highest scores and the strongest priorities from a massive candidate set, and outputting them in order. Its core is to sort the candidate objects through quantitative evaluation indicators to ensure that the output results meet both high value and operability. Top: represents the head results sorted by specific rules. N: The number of returns defined by the user (such as Top-5, Top-10), usually determined according to business needs. Scoring basis: comprehensive score calculated based on the algorithm model (such as correlation, prediction effect, user preference, etc.).

[0098] The rule types of the candidate combination generation rules include: explicit rule inheritance, implicit relationship expansion and safety constraint filtering; the specific logic of explicit rule inheritance is: at least one standard instrument package in the procedure definition must be included; the specific logic of implicit relationship expansion is: instruments with a correlation degree > 0.7 automatically generate 2-3 combinations, and the specific logic of safety constraint filtering is: exclude contraindication instruments (real-time verification through knowledge graph).

[0099] Furthermore, the historical procedure records of the operator's information are obtained, and the number of historical procedure sessions of the operator is obtained based on the historical procedure records. According to the historical procedure session number and a preset procedure session number threshold, a confidence index is calculated by combining relative proportion and Sigmoid function.

[0100] In this embodiment, the stability and coverage of the model output are scored. If the data of a certain operator is too little, the model will prompt a low confidence index. In this case, the rule will take priority in the decision. Specifically, the historical records of the operator's information are obtained, and the number of historical procedures of the operator G is counted. Assuming the preset threshold of the number of procedures is T, the confidence index C is calculated based on the relative ratio of the historical number of procedures to the preset threshold of the number of procedures, and the Sigmoid function is combined. It can be expressed as:

[0101]

[0102] In the above formula, when G is close to T, the confidence level changes smoothly in the middle range, and considering the relative ratio of G to T rather than the absolute difference, the confidence level is calculated by Normalizing G as a ratio relative to the threshold T eliminates dimensionality and enhances the robustness of the formula. Overall, this formula, by combining relative ratios with the Sigmoid function, avoids the limitations of piecewise or linear formulas, balances relative ratios with exponential decay, and implements confidence calculation for dynamic responses to thresholds.

[0103] Furthermore, as shown in Table 4, Table 4 shows the input content, processing logic and output content of the above-mentioned multiple machine learning models.

[0104] Table 4: Machine Learning Model Comparison

[0105]

[0106]

[0107] Furthermore, the AI ​​prediction results are compared with the preliminary device recommendation list to obtain an adjusted recommendation list, which specifically includes:

[0108] Obtaining from the device recommendation scoring table devices that have a score higher than a preset score threshold and are not in the preliminary device recommendation list as devices to be added;

[0109] Obtaining from the device recommendation score table devices whose scores are not higher than the preset score threshold, whose usage probabilities are lower than the preset probability threshold, and which exist in the preliminary device recommendation list as devices to be deleted;

[0110] Selecting a device with a score higher than the preset score threshold from the recommended alternative devices as the device to be replaced;

[0111] Rearranging the actual necessity level of each device in the preliminary device recommendation list according to the Top-N combination of high-priority devices to obtain a device recommendation list after sorting optimization;

[0112] The preliminary device recommendation list is adjusted according to the devices to be added, the devices to be deleted, the devices to be replaced, and the sorted and optimized device recommendation list to generate an adjusted recommendation list.

[0113] In this embodiment, the AI ​​prediction results are compared with the preliminary device recommendation list, so that the preliminary device recommendation list is subjected to multi-strategy evaluation and intelligent optimization through the AI ​​prediction results to form an adjusted final device recommendation list V2.

[0114] As shown in Table 5, the system first receives the standard recommendation list (V1) output by the rule engine; at the same time, it receives the instrument scoring matrix, recommended combination list, confidence score and other results output by the machine learning module; the system compares the two and performs the following adjustment operations: Add: items with high model scores but missing templates; Delete: items with low model scores and extremely low probability of use; Replace: items with a substitution relationship and the model recommends a better instrument; Sorting: sorting based on a comprehensive combination of scores and surgeon historical behavior; Grading: grading each instrument in the list sequence according to the score and the preset grading threshold range, and the levels include mandatory items, recommended items, and optional items.

[0115] Table 5: AI prediction result adjustment correspondence table

[0116]

[0117] Step S40: Use a strategy evaluation engine and introduce a weight coefficient to score each device in the adjusted recommendation list, and obtain a final recommended device list based on the scoring results.

[0118] Specifically, the cost score, efficiency score, safety performance score, and operator satisfaction matching degree of the adjusted recommendation list are calculated by the strategy optimizer, and the confidence score of the adjusted recommendation list is calculated based on the confidence index.

[0119] Calculating scores using the Policy Optimizer is a process that combines algorithm design, objective function optimization, and dynamic feedback. As shown in Table 6, in this embodiment, the Policy Evaluation Engine uses a scoring formula to score each recommendation, comprehensively considering cost, efficiency, safety, and surgeon preference. The confidence score for the adjusted recommendation list is then calculated based on the confidence index for each device.

[0120] Table 6: Strategic plan scoring table

[0121]

[0122] Furthermore, different weights are set for the cost score, the efficiency score, the safety performance score, the surgeon satisfaction match and the confidence score, respectively. The cost score, the efficiency score, the safety performance score, the surgeon satisfaction match and the confidence score are weightedly fused according to the weights to obtain a comprehensive score.

[0123] It is understood that the present invention introduces a weighted coefficient control recommendation strategy, assigning different weights to the cost score, efficiency score, safety performance score, operator satisfaction match, and confidence score (the specific values ​​can be set based on actual conditions, for example, setting the weight of the safety performance score dimension to take precedence over the weight of the cost score dimension). A comprehensive score is obtained based on the weighted scores. Weighted addition is a quantitative assessment method used to comprehensively consider the impact of multiple factors on a specific goal or outcome, and can reflect the differences in the importance of each data.

[0124] Furthermore, if the comprehensive score exceeds the preset comprehensive score threshold, the adjusted recommendation list will be used as the final recommended equipment list; if the comprehensive score does not exceed the preset comprehensive score threshold, the preliminary equipment recommendation list will be used as the final recommended equipment list.

[0125] In this embodiment, a comprehensive scoring threshold is preset. In the standard procedure-instrument configuration table, the preset comprehensive scoring threshold is calculated in advance in the preliminary instrument recommendation list corresponding to each procedure. Whether the adjusted recommendation list can be used as the final recommended instrument list also requires comparing the comprehensive score corresponding to the adjusted recommendation list with the comprehensive scoring threshold. Only when the comprehensive score exceeds the preset comprehensive scoring threshold will the adjusted recommendation list be used as the final recommended instrument list. Otherwise, it means that the rationality of the adjusted recommendation list is not as good as that of the preliminary instrument recommendation list after comprehensive consideration, and therefore the preliminary instrument recommendation list will be used as the final recommended instrument list.

[0126] Furthermore, a scheduling task list is generated based on the final recommended equipment list, and the scheduling task list is pushed to the execution system. At the same time, the postoperative usage data is collected as feedback records and input into the machine learning model to participate in model optimization, so as to feed back the model training and realize a complete data closed loop.

[0127] Specifically, the system receives the final recommended equipment list (including rating labels such as required items, recommended items, and alternative items), combines it with preoperative scheduling information, and generates a standardized scheduling task list, including task ID, interoperative goals, time window, and tracking code; scheduling tasks are sent to the SPD supply system or equipment packaging system through the API for picking, packaging, and intelligent distribution; after surgery, the system collects actual equipment data (intraoperative nursing records, equipment recovery lists, etc.), compares the differences with the recommended list, and generates structured feedback data; and uses difference items (such as unused items, supplementary items, and alternative items) as labels and inputs them into the model training pipeline to improve the predictive capabilities of preference models, probability models, and combination models; it supports setting feedback cycles and retraining frequencies to realize the platform's continuous iteration mechanism of "prediction → execution → feedback → relearning."

[0128] It can be seen that the present invention proposes a surgical instrument configuration prediction method based on a fusion rule engine and machine learning. It can access the hospital's preoperative scheduling data, historical surgical procedure execution data, surgeon behavior characteristics, and instrument inventory status, and automatically generate more accurate, personalized, and economical surgical instrument configuration plans by combining a "rule-driven + data-driven" dual-engine architecture. The present invention supports the dual-path fusion of a standard surgical instrument rule library and a machine learning algorithm model, introducing a personalized correction mechanism into the rule framework; it can be trained based on historical data to form a surgeon preference model, a surgical procedure evolution model, and a consumables redundancy identification model; it can realize adaptive strategy selection in different scheduling scenarios and continuously iterate and optimize through a feedback mechanism.

[0129] This invention upgrades the traditional static instrument template configuration method to a dynamic intelligent recommendation mechanism driven by the fusion of a rule engine and a machine learning model, which has the following beneficial effects:

[0130] 1. Realize personalized, adaptive and data-driven preoperative instrument configuration;

[0131] 2. Effectively solve the problem of template failure caused by the evolution of surgical procedures and replacement of surgeons;

[0132] 3. Significantly reduce device redundancy and missing risks, and improve recommendation accuracy;

[0133] 4. Reduce the burden of manual maintenance of the rule base and continuously optimize the configuration model through self-learning capabilities;

[0134] 5. Improve the dispatch system's ability to respond to procedure changes and equipment conflicts without affecting the scheduling rhythm;

[0135] 6. Build a complete closed-loop path to support the intelligent and standardized upgrade of hospital surgical preparations.

[0136] Furthermore, if Figure 3As shown, based on the above-mentioned surgical instrument configuration prediction method based on the fusion rule engine and machine learning, the present invention also provides a surgical instrument configuration prediction system based on the fusion rule engine and machine learning, wherein the surgical instrument configuration prediction system based on the fusion rule engine and machine learning includes:

[0137] The data integration module 51 is used to obtain the surgical procedure and surgeon of the target surgery, collect multi-dimensional data resources corresponding to the surgical procedure and surgeon from various business systems of the hospital, and parse and integrate the multi-dimensional data resources to obtain structured standard data;

[0138] A rule engine module 52 is configured to construct a standard procedure-instrument configuration table based on clinical expert knowledge, convert the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, match the procedure with the rule execution template, and generate a preliminary instrument recommendation list;

[0139] a machine learning module 53 for inputting the structured standard data into a plurality of machine learning models for deep learning and prediction to obtain AI prediction results, and using the AI ​​prediction results to compare with the preliminary device recommendation list to obtain an adjusted recommendation list;

[0140] The decision fusion module 54 is configured to use a strategy evaluation engine and introduce a weight coefficient to score each device in the adjusted recommendation list, and obtain a final recommended device list based on the scoring results.

[0141] Furthermore, if Figure 4 As shown, based on the above-mentioned surgical instrument configuration prediction method and system based on fusion rule engine and machine learning, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 4 Only some of the components of the terminal are shown, but it should be understood that implementation of all of the shown components is not required, and more or fewer components may be implemented instead.

[0142] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Furthermore, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a surgical instrument configuration prediction program 40 based on a fusion rule engine and machine learning is stored on the memory 20. The surgical instrument configuration prediction program 40 based on a fusion rule engine and machine learning can be executed by the processor 10, thereby realizing the surgical instrument configuration prediction method based on a fusion rule engine and machine learning in this application.

[0143] In some embodiments, the processor 10 can be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the surgical instrument configuration prediction method based on the fusion rule engine and machine learning.

[0144] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0145] In one embodiment, when the processor 10 executes the surgical instrument configuration prediction program 40 based on the fusion rule engine and machine learning in the memory 20, the following steps are implemented:

[0146] Obtain the surgical procedure and surgeon for the target surgery, collect multidimensional data resources corresponding to the surgical procedure and surgeon from various hospital business systems, parse and integrate the multidimensional data resources to obtain structured standard data;

[0147] Constructing a standard procedure-instrument configuration table based on clinical expert knowledge, converting the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, matching the procedure with the rule execution template, and generating a preliminary instrument recommendation list;

[0148] Inputting the structured standard data into multiple machine learning models for deep learning and prediction to obtain AI prediction results, and using the AI ​​prediction results to compare with the preliminary device recommendation list to obtain an adjusted recommendation list;

[0149] A strategy evaluation engine is used and a weight coefficient is introduced to score each device in the adjusted recommended list, and a final recommended device list is obtained based on the scoring results.

[0150] The multi-dimensional data resources include preoperative scheduling data, procedure definition, operator information, and intraoperative usage behavior;

[0151] The preoperative scheduling data includes procedure number, operator ID, time window and operation room allocation;

[0152] The procedure definition includes the procedure name, classification, complexity, recommended duration, standard instrument template ID and alternative procedures;

[0153] The surgeon information includes the surgeon's department, surgical experience, historical surgical records, and intraoperative instrument behavior vectors;

[0154] The intraoperative usage behaviors include instrument addition and subtraction records, bag change behaviors, actual usage lists and abnormal usage situations.

[0155] The construction of a standard surgical procedure-instrument configuration table based on clinical expert knowledge and the conversion of the constructed standard surgical procedure-instrument configuration table using a rule template language to obtain a rule execution template specifically include:

[0156] Based on the hospital's clinical pathway, team consensus, and surgical SOP documents, combined with clinical expert experience, define the standard instrument set corresponding to each surgical procedure and the necessity level of each instrument;

[0157] Collecting statistics on the actual use of the devices to obtain statistical data, and using the statistical data to assist in verifying the necessity level of each device to obtain the true necessity level of each device;

[0158] A standard procedure-instrument configuration table is constructed based on the standard instrument set corresponding to each procedure and the actual necessity level of each instrument;

[0159] The constructed standard surgical procedure-instrument configuration table is converted using the Drools rule language to obtain an IF-THEN rule execution template.

[0160] The multiple machine learning models include a preference scoring model, an alternative device recommendation model, a device usage probability model, and a device combination model;

[0161] The AI ​​prediction results include a device recommendation score table, recommended alternative devices, a Top-N combination of high-priority devices, and the probability and confidence index of each device's use.

[0162] The step of inputting the structured standard data into a variety of machine learning models for deep learning and prediction to obtain AI prediction results specifically includes:

[0163] Inputting the integrated operator information and integrated intraoperative usage behavior into the preference scoring model for learning, obtaining the operator's usage preference scoring matrix for each instrument under a specific surgical procedure, and obtaining an instrument recommendation scoring table based on the usage preference scoring matrix;

[0164] Inputting the integrated surgical procedure definition and the integrated intraoperative usage behavior into the instrument usage probability model for learning, obtaining the usage probability distribution of each instrument under a given surgical procedure, and obtaining the usage probability of each instrument;

[0165] Inputting the integrated operator information, integrated procedure definition, and integrated intraoperative use behavior into the alternative device recommendation model for learning to obtain recommended alternative devices with substitution relationships;

[0166] Inputting the integrated operator information, integrated procedure definition, and integrated intraoperative usage behavior into the instrument combination model for learning to obtain a high-frequency linkage instrument combination that completes the implicit relationship, and outputting the high-frequency linkage instrument combination in a Top-N manner to obtain a Top-N combination of high-priority instruments;

[0167] Obtain historical procedure records of the operator's information, obtain the operator's historical procedure number based on the historical procedure records, and calculate the confidence index based on the historical procedure number and a preset procedure number threshold by combining relative proportion and Sigmoid function.

[0168] The AI ​​prediction results are used to compare the preliminary device recommendation list to obtain an adjusted recommendation list, which specifically includes:

[0169] Obtaining from the device recommendation scoring table devices that have a score higher than a preset score threshold and are not in the preliminary device recommendation list as devices to be added;

[0170] Obtaining from the device recommendation score table devices whose scores are not higher than the preset score threshold, whose usage probabilities are lower than the preset probability threshold, and which exist in the preliminary device recommendation list as devices to be deleted;

[0171] Selecting a device with a score higher than the preset score threshold from the recommended alternative devices as the device to be replaced;

[0172] Rearranging the actual necessity level of each device in the preliminary device recommendation list according to the Top-N combination of high-priority devices to obtain a device recommendation list after sorting optimization;

[0173] The preliminary device recommendation list is adjusted according to the devices to be added, the devices to be deleted, the devices to be replaced, and the sorted and optimized device recommendation list to generate an adjusted recommendation list.

[0174] The strategy evaluation engine is used to introduce a weight coefficient to score each device in the adjusted recommendation list, and a final recommended device list is obtained based on the scoring results, specifically including:

[0175] Calculating the cost score, efficiency score, safety performance score, and operator satisfaction matching of the adjusted recommendation list through a strategy optimizer, and calculating the confidence score of the adjusted recommendation list based on the confidence index;

[0176] Setting different weights for the cost score, the efficiency score, the safety performance score, the operator satisfaction match, and the confidence score, respectively, and performing weighted fusion of the cost score, the efficiency score, the safety performance score, the operator satisfaction match, and the confidence score according to the weights to obtain a comprehensive score;

[0177] If the comprehensive score exceeds the preset comprehensive score threshold, the adjusted recommendation list will be used as the final recommended equipment list; if the comprehensive score does not exceed the preset comprehensive score threshold, the preliminary equipment recommendation list will be used as the final recommended equipment list.

[0178] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a surgical instrument configuration prediction program based on a fusion rule engine and machine learning, and when the surgical instrument configuration prediction program based on a fusion rule engine and machine learning is executed by a processor, the steps of the surgical instrument configuration prediction method based on a fusion rule engine and machine learning as described above are implemented.

[0179] In summary, the present invention proposes a surgical instrument configuration prediction method, system, terminal and storage medium based on the fusion of rule engine and machine learning, the method comprising: obtaining the surgical procedure and surgeon of the target surgery, collecting corresponding multidimensional data resources, and obtaining structured standard data; constructing a standard surgical procedure-instrument configuration table based on clinical expert knowledge, converting the constructed standard surgical procedure-instrument configuration table using a rule template language to obtain a rule execution template, matching the surgical procedure with the rule execution template, and generating a preliminary instrument recommendation list; inputting the structured standard data into a machine learning model for deep learning and prediction to obtain an AI prediction result, comparing the AI ​​prediction result with the preliminary instrument recommendation list to obtain an adjusted recommendation list; using a strategy evaluation engine and introducing a weight coefficient to score the instruments in the adjusted recommendation list, and obtaining a final recommended instrument list based on the scoring result. The present invention can access the hospital's preoperative scheduling data, historical surgical procedure execution data, surgeon behavior characteristics, and instrument inventory status, and by combining a "rule-driven + data-driven" dual-engine architecture, automatically generates a more accurate, personalized and economical surgical instrument configuration plan. It supports dual-path fusion of the standard surgical instrument rule library and the machine learning algorithm model, and introduces a personalized correction mechanism into the rule framework. It can be trained based on historical data to form a surgeon preference model, a surgical procedure evolution model, and a consumables redundancy identification model. It can realize adaptive strategy selection in different scheduling scenarios and continuously iterate and optimize through a feedback mechanism.

[0180] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or terminal comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or terminal comprising the element.

[0181] Of course, those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0182] It should be understood that the application of the present invention is not limited to the above examples. For those skilled in the art, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A surgical instrument configuration prediction method based on a fusion rule engine and machine learning, characterized in that: The surgical instrument configuration prediction method based on the fusion rule engine and machine learning includes: Obtain the surgical procedure and surgeon for the target surgery, collect multidimensional data resources corresponding to the surgical procedure and surgeon from various hospital business systems, parse and integrate the multidimensional data resources to obtain structured standard data; Constructing a standard procedure-instrument configuration table based on clinical expert knowledge, converting the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, matching the procedure with the rule execution template, and generating a preliminary instrument recommendation list; Inputting the structured standard data into multiple machine learning models for deep learning and prediction to obtain AI prediction results, and using the AI ​​prediction results to compare with the preliminary device recommendation list to obtain an adjusted recommendation list; A strategy evaluation engine is used and a weight coefficient is introduced to score each device in the adjusted recommended list, and a final recommended device list is obtained based on the scoring results.

2. The method for predicting surgical instrument configuration based on a fusion rule engine and machine learning according to claim 1, characterized in that: The multidimensional data resources include preoperative scheduling data, procedure definitions, operator information, and intraoperative usage behavior; The preoperative scheduling data includes procedure number, operator ID, time window and operation room allocation; The procedure definition includes the procedure name, classification, complexity, recommended duration, standard instrument template ID and alternative procedures; The surgeon information includes the surgeon's department, surgical experience, historical surgical records, and intraoperative instrument behavior vectors; The intraoperative usage behaviors include instrument addition and subtraction records, bag change behaviors, actual usage lists and abnormal usage situations.

3. The method for predicting surgical instrument configuration based on a fusion rule engine and machine learning according to claim 2, characterized in that: The method of constructing a standard surgical procedure-instrument configuration table based on clinical expert knowledge and converting the constructed standard surgical procedure-instrument configuration table using a rule template language to obtain a rule execution template specifically includes: Based on the hospital's clinical pathway, team consensus, and surgical SOP documents, combined with clinical expert experience, define the standard instrument set corresponding to each surgical procedure and the necessity level of each instrument; Collecting statistics on the actual use of the devices to obtain statistical data, and using the statistical data to assist in verifying the necessity level of each device to obtain the true necessity level of each device; A standard procedure-instrument configuration table is constructed based on the standard instrument set corresponding to each procedure and the actual necessity level of each instrument; The constructed standard surgical procedure-instrument configuration table is converted using the Drools rule language to obtain an IF-THEN rule execution template.

4. The method for predicting surgical instrument configuration based on a fusion rule engine and machine learning according to claim 3, characterized in that: The multiple machine learning models include a preference scoring model, an alternative device recommendation model, a device usage probability model, and a device combination model; The AI ​​prediction results include a device recommendation score table, recommended alternative devices, a Top-N combination of high-priority devices, and the probability and confidence index of each device's use.

5. The method for predicting surgical instrument configuration based on a fusion rule engine and machine learning according to claim 4, characterized in that: The structured standard data is input into a variety of machine learning models for deep learning and prediction to obtain AI prediction results, specifically including: Inputting the integrated operator information and integrated intraoperative usage behavior into the preference scoring model for learning, obtaining the operator's usage preference scoring matrix for each instrument under a specific surgical procedure, and obtaining an instrument recommendation scoring table based on the usage preference scoring matrix; Inputting the integrated surgical procedure definition and the integrated intraoperative usage behavior into the instrument usage probability model for learning, obtaining the usage probability distribution of each instrument under a given surgical procedure, and obtaining the usage probability of each instrument; Inputting the integrated operator information, integrated procedure definition, and integrated intraoperative use behavior into the alternative device recommendation model for learning to obtain recommended alternative devices with substitution relationships; Inputting the integrated operator information, integrated procedure definition, and integrated intraoperative usage behavior into the instrument combination model for learning to obtain a high-frequency linkage instrument combination that completes the implicit relationship, and outputting the high-frequency linkage instrument combination in a Top-N manner to obtain a Top-N combination of high-priority instruments; Obtain historical procedure records of the operator's information, obtain the operator's historical procedure number based on the historical procedure records, and calculate the confidence index based on the historical procedure number and a preset procedure number threshold by combining relative proportion and Sigmoid function.

6. The method for predicting surgical instrument configuration based on a fusion rule engine and machine learning according to claim 5, characterized in that: The AI ​​prediction results are used to compare the preliminary device recommendation list to obtain an adjusted recommendation list, specifically including: Obtaining from the device recommendation scoring table devices that have a score higher than a preset score threshold and are not in the preliminary device recommendation list as devices to be added; Obtaining from the device recommendation score table devices whose scores are not higher than the preset score threshold, whose usage probabilities are lower than the preset probability threshold, and which exist in the preliminary device recommendation list as devices to be deleted; Selecting a device with a score higher than the preset score threshold from the recommended alternative devices as the device to be replaced; Rearranging the actual necessity level of each device in the preliminary device recommendation list according to the Top-N combination of high-priority devices to obtain a device recommendation list after sorting optimization; The preliminary device recommendation list is adjusted according to the devices to be added, the devices to be deleted, the devices to be replaced, and the sorted and optimized device recommendation list to generate an adjusted recommendation list.

7. The method for predicting surgical instrument configuration based on a fusion rule engine and machine learning according to claim 6, characterized in that: The strategy evaluation engine is used to introduce a weight coefficient to score each device in the adjusted recommendation list, and a final recommended device list is obtained based on the scoring results, specifically including: Calculating the cost score, efficiency score, safety performance score, and operator satisfaction matching of the adjusted recommendation list through a strategy optimizer, and calculating the confidence score of the adjusted recommendation list based on the confidence index; Setting different weights for the cost score, the efficiency score, the safety performance score, the operator satisfaction match, and the confidence score, respectively, and performing weighted fusion of the cost score, the efficiency score, the safety performance score, the operator satisfaction match, and the confidence score according to the weights to obtain a comprehensive score; If the comprehensive score exceeds the preset comprehensive score threshold, the adjusted recommendation list will be used as the final recommended equipment list; if the comprehensive score does not exceed the preset comprehensive score threshold, the preliminary equipment recommendation list will be used as the final recommended equipment list.

8. A surgical instrument configuration prediction system based on a fusion rule engine and machine learning, characterized in that: The surgical instrument configuration prediction system based on the fusion rule engine and machine learning includes: The data integration module is used to obtain the surgical procedure and surgeon of the target surgery, collect the multi-dimensional data resources corresponding to the surgical procedure and surgeon from various business systems of the hospital, and parse and integrate the multi-dimensional data resources to obtain structured standard data; A rule engine module is used to construct a standard procedure-instrument configuration table based on clinical expert knowledge, convert the constructed standard procedure-instrument configuration table using a rule template language to obtain a rule execution template, match the procedure with the rule execution template, and generate a preliminary instrument recommendation list; a machine learning module, configured to input the structured standard data into a plurality of machine learning models for deep learning and prediction to obtain AI prediction results, and compare the AI ​​prediction results with the preliminary device recommendation list to obtain an adjusted recommendation list; The decision fusion module is used to use a strategy evaluation engine and introduce a weight coefficient to score each device in the adjusted recommendation list, and obtain a final recommended device list based on the scoring results.

9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a surgical instrument configuration prediction program based on a fusion rule engine and machine learning stored in the memory and runnable on the processor. When the surgical instrument configuration prediction program based on a fusion rule engine and machine learning is executed by the processor, the steps of the surgical instrument configuration prediction method based on a fusion rule engine and machine learning as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a surgical instrument configuration prediction program based on a fusion rule engine and machine learning. When the surgical instrument configuration prediction program based on a fusion rule engine and machine learning is executed by a processor, the steps of the surgical instrument configuration prediction method based on a fusion rule engine and machine learning as described in any one of claims 1 to 7 are implemented.

Images