An electrosurgical safety surgery risk warning system

Abstract

This invention discloses an electrosurgical safety risk warning system, relating to the field of intelligent surgical safety. The system includes: an information acquisition module for acquiring patient biometric information and the spatial location information of implants; a human body model database storing multiple categories of three-dimensional human body models constructed based on biometric parameterization; an implant configuration module for matching corresponding human body models according to the patient's biometrics and marking the implants on the models according to their spatial location information; a human-computer interaction interface for visually displaying the human body models with implant markings and receiving surgical input information; and an electrode application guidance module for generating electrode application area indication information based on the human body models with implant markings and surgical information, combined with the electrode application database. This solution can effectively reduce the risk of tissue damage and implant dysfunction, and alleviate the cognitive burden on medical staff.

Description

Technical Field

[0001] This invention relates to the field of intelligent surgical safety, and in particular to an electrosurgical safety risk warning system. Background Technology

[0002] Electrosurgery, an indispensable technique in modern surgery, uses high-frequency current to cut tissue and coagulate blood. However, this technology also carries inherent risks, especially for patients with implanted metal devices. The conduction path of high-frequency current within the body is unpredictable. If the current circuit is poorly designed, it may cause local overheating when flowing through the implant, leading to serious complications such as tissue burns and implant dysfunction. Currently, the management of patient implant information relies heavily on decentralized paper or unstructured electronic medical record systems, which are prone to omissions and delays during preoperative handover and verification. Furthermore, the placement of electrodes and the setting of electrosurgical equipment parameters during surgery heavily depend on the experience and subjective judgment of the surgical team, lacking quantitative and visual decision support based on the patient's individual anatomy and the spatial location of the implant. This traditional model, relying on human memory and experience, has become a key bottleneck restricting the improvement of surgical safety and efficiency in a clinical environment with increasing surgical complexity and a wide variety of implants.

[0003] Several significant shortcomings exist in existing technologies: First, at the information management level, traditional hospital information systems can only record implant information in text form, severely lacking spatial visualization capabilities. This makes it difficult for surgeons to intuitively and accurately understand the three-dimensional spatial relationship between the implant, the intended surgical area, and the current circuit, leading to significant information transmission gaps and misjudgment risks. Second, at the intraoperative decision support level, existing solutions generally lack integrated intelligent analysis capabilities. The selection of electrode placement positions relies entirely on the experience rules of medical staff, resulting in high training costs and failing to effectively avoid operational risks caused by individual cognitive biases. This limitation is particularly pronounced in high-risk surgeries or complex implant scenarios. Furthermore, the generation of risk disclosure and informed consent forms remains in a rudimentary stage, relying on manual completion or simple template application. The process is cumbersome, lacks standardization, and struggles to achieve real-time linkage with dynamic risk assessment results. These systemic deficiencies collectively lead to a series of clinical and management pain points, such as prolonged preoperative preparation time, reduced intraoperative safety margins, and low levels of medical management standardization. Therefore, based on the above challenges, this invention proposes an electrosurgical safety surgery risk warning system. Summary of the Invention

[0004] To address the aforementioned issues, the present invention aims to provide an electrosurgical safety risk warning system that enables precise preoperative design of electrode application areas and personalized recommendations of equipment parameters, dynamic risk assessment of the relationship between the surgical path and the implant during surgery, and automated generation of structured risk notification documents. This system reduces the overall risk of tissue damage and implant malfunction caused by improper current circuitry, thereby improving the efficiency of surgical preparation and the standardization of management.

[0005] To achieve the above objectives, this invention provides an electrosurgical safety surgery risk warning system. The system first acquires the patient's biometrics and implant location information; then, based on this information, it matches the optimal model from a parameterized human model database and performs three-dimensional spatial marking of the implant; on this basis, it provides a visualization platform for doctors to annotate surgical positions and intended surgical sites; the electrode application guidance module, as the core of the system, queries the electrode application database based on the aforementioned information, integrates historical data and clinical rules, and generates electrode application instructions including recommended and contraindicated areas, and can further output recommended device power values; finally, based on the system analysis results, it automatically generates a structured informed consent template, completing a fully digital closed loop from risk assessment to document preparation.

[0006] In a first aspect, the present invention provides an electrosurgical safety surgery risk warning system, comprising:

[0007] The information acquisition module is used to acquire the patient's biometric information and the spatial location information of the implanted device in the body;

[0008] The human body model database stores multiple categories of 3D human body models constructed based on biometric parameterization.

[0009] An implant configuration module is used to match a corresponding human body model based on the patient's biometrics and to mark the model based on the spatial location information of the implant.

[0010] The human-computer interaction interface is used to visually display a human body model with implant markers and to receive surgical input information;

[0011] The electrode application guidance module is used to generate electrode application area indication information based on the human body model with implant markings and surgical information, combined with the electrode application database.

[0012] The risk disclosure module is used to generate structured informed consent documents based on patient information and system risk assessment.

[0013] The information acquisition module and the implant configuration module work together to integrate medical imaging data, physical detection signals and text medical record information through a multi-source heterogeneous data fusion mechanism, and achieve precise positioning and marking of the implant location based on spatial consistency constraints and confidence propagation algorithms.

[0014] Furthermore, the information acquisition module supports obtaining the location information of implants in the patient's body through non-contact or contact detection devices, and is equipped with a data interface for interfacing with the hospital information system, thereby realizing unified access and fusion of multi-source heterogeneous data and improving the integrity and efficiency of information acquisition.

[0015] Furthermore, the implant configuration module determines the type and installation location of the implant by using a conventional implant location lookup table and marks it on the human model. The lookup table contains the names, types and typical anatomical locations of common implants, thereby improving the spatial positioning ability and marking accuracy for unknown implant types.

[0016] Furthermore, common implant names, types, and corresponding installation locations include cochlear implants located in the ear, pacemakers located in the heart, and / or metal implants located in the limbs.

[0017] Furthermore, the electrode application database is constructed using multi-source historical data and clinical rules, including the mapping relationship between implant-marked human models, surgical poses, planned surgical sites and electrode application areas. Based on the electrode application area indication information and surgical type, it outputs recommended settings for surgical equipment parameters and uses a data completion mechanism to improve the coverage of uncovered scenarios, thereby enhancing the system's decision robustness and adaptability in the face of rare surgical situations.

[0018] Furthermore, the electrode application guidance module is further configured to, based on the human body model containing implant markers and surgical information, dynamically simulate the distribution path of high-frequency current in the individualized model, and combine risk weighting factors to quantitatively assess the interaction risk between the current path and the implant, thereby generating optimized electrode application area indication information, thus achieving integrated collaborative optimization from spatial design to equipment parameters.

[0019] Furthermore, the electrode application area indication information is presented on the human body model in a visually differentiated manner, including differentiated markings of recommended and contraindicated areas, to intuitively assist clinicians in making quick and accurate electrode application decisions.

[0020] Furthermore, the generation of the electrode application area takes into account the following conditions:

[0021] Based on research and simulation of human anatomy and the flow path of high-frequency current in the human body, the possible paths of high-frequency current in the human body are simulated.

[0022] Prioritize areas that meet the following criteria: form a low-current-density triangle with the surgical site and metal implant; have abundant muscle and blood vessels in a homogeneous conductive area; and meet international standards.

[0023] Generate a pasteable area to ensure tolerance for errors in clinical procedures.

[0024] Furthermore, the human body model is a parametric three-dimensional model, which is constructed based on the patient's biometric data and achieves spatial mapping of anatomical structures through the fusion of surface and internal data, thereby providing a high-fidelity anatomical basis for spatial analysis and design.

[0025] Furthermore, the human body models are divided into different categories based on different genders, heights, weights, and ages, with each category containing several human body models with preset data ranges.

[0026] Furthermore, the patient's basic information includes at least the patient's gender, height, weight, and age.

[0027] Furthermore, the system also includes a dynamic risk assessment module, which performs real-time safety analysis based on the relative spatial relationship between the implant location and the surgical area, and triggers a warning signal when the risk threshold is approached, thereby achieving proactive early warning and intervention for potential risks during the operation.

[0028] Furthermore, the system integrates a clinical rule base, embeds contraindication judgment logic, and automatically excludes unsuitable body surface areas for application, including bony prominences, scar tissue, and areas adjacent to metal implants. By embedding expert knowledge in a structured manner, the system ensures the basic security of the output solution.

[0029] Furthermore, the system analyzes historical application effectiveness data through graph neural networks to optimize the recommendation accuracy of electrode application positions, and combines multi-objective optimization principles to balance impedance minimization and ease of operation, thereby optimizing the system's long-term decision-making performance through a continuous learning mechanism.

[0030] In a second aspect, the present invention also provides a non-transitory computer-readable medium storing computer-executable instructions, which, when executed by a processor, control the operation of the system described in the first aspect; the medium includes, but is not limited to, solid-state memory, optical storage devices, or cloud storage systems, ensuring that the system can be stably implemented on a variety of hardware platforms.

[0031] This invention provides an electrosurgical safety surgery risk warning system. The system acquires basic data through an information acquisition module and establishes an individualized three-dimensional surgical environment model using a parametric human body model database and an implant configuration module. Then, based on this model, the electrode application guidance module combines dynamic current path simulation and multi-source data fusion algorithms to generate risk-avoiding electrode application designs and equipment parameter suggestions. Finally, through an integrated clinical rule engine and risk disclosure module, a structured informed consent document is automatically output, thereby realizing a fully digital and intelligent closed loop from patient information perception, spatial modeling, risk quantification to plan generation and document preparation.

[0032] This system systematically improves the safety and standardization of electrosurgery. In terms of safety, by transforming the traditional experience-based decision-making process into a precise design based on physical simulation and quantitative risk assessment, it effectively reduces the risk of tissue damage and implant dysfunction caused by improper current circuit design. Regarding efficiency, the system significantly shortens preoperative preparation and documentation time through automation and visualization technologies, reducing the cognitive load on medical staff. At the management level, it achieves digital traceability and standardized management of the entire process, including implant information, surgical design, equipment parameters, and informed consent, providing reliable technical support for medical quality control and legal compliance.

[0033] Beneficial effects

[0034] By implementing the electrosurgical safety risk warning system provided by the present invention, the following technical effects are achieved:

[0035] (1) By establishing a three-dimensional human body model library based on biometric parameterization and accurately mapping it with the spatial location information of implants, a digital twin environment for surgical design was constructed. It transforms abstract textual patient information into an intuitive and interactive three-dimensional spatial model, providing doctors with a visual decision-making interface that integrates anatomical structures and implant constraints. This achieves a seamless transformation of surgical plans from abstract concepts to specific spatial relationships, greatly improving the intuitiveness and operational efficiency of preoperative design.

[0036] (2) By embedding clinical guidelines and expert experience into the system in a structured manner, an executable rule engine is formed, which drives the automatic generation of structured informed consent forms. It realizes full-process automation and standardization from risk assessment to compliance document output, which not only liberates medical staff from tedious and error-prone manual paperwork, but also ensures the integrity of risk disclosure and the standardization of legal evidence, completing a closed loop from intelligent analysis to clinical management implementation.

[0037] (3) By constructing a dynamic simulation of the current path based on individualized anatomy and conductivity distribution, and combining it with an original risk weighting function, the physical mechanism-level optimization of electrode placement was achieved. This fundamentally transforms electrode placement design from an empirical decision-making process that relies on static distance thresholds to a precise and quantitative decision-making process based on forward-looking prediction of current behavior. This systematically avoids the risks of tissue thermal damage and implant interference caused by improper current loop design, significantly improving the scientific nature and reliability of surgical safety.

[0038] (4) By constructing an adaptive weight allocation and confidence propagation mechanism, the inconsistency and noise problems of multi-source heterogeneous clinical data in implant localization are effectively solved. It significantly enhances the robustness and accuracy of the system in complex scenarios such as data conflicts, missing data or poor quality, and provides more reliable and accurate implant spatial information input for the entire early warning system, ensuring the accuracy of all subsequent decision-making processes from the data source. Attached Figure Description

[0039] To make the above-described electrosurgical safety risk warning system of the present invention more clear and understandable, the accompanying drawings used in the specific embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0040] Figure 1 This is a schematic diagram of the system modules in this application;

[0041] Figure 2 This is a schematic diagram of the system architecture of this application;

[0042] Figure 3 This is a flowchart illustrating the method described in this application;

[0043] Figure 4 Represents a three-dimensional humanoid heat map;

[0044] Figure 5 This is a schematic diagram showing the positioning of a 3D humanoid implant.

[0045] Figure 6 This is a schematic diagram of a neural network model architecture. Detailed Implementation

[0046] Example 1:

[0047] This embodiment provides an electrosurgical safety surgery risk warning system, the system modules of which are as follows: Figure 1 As shown, the architecture is as follows Figure 2As shown, this system constructs an intelligent decision support platform integrating patient biometrics, three-dimensional spatial information of implanted devices, surgical design, and clinical rules. It achieves a fully digital and intelligent closed-loop process from patient information perception, spatial modeling, and risk quantification to plan generation and document preparation. The early warning method process is as follows: Figure 3 As shown, the system first acquires basic patient information, including gender, height, weight, and age, through an information acquisition module. It then obtains basic location information of implants within the patient's body through patient self-reporting or detection devices such as handheld metal scanners. This information provides the data foundation for subsequent surgical planning. The human body model database stores multiple categories of 3D human body models constructed based on biometric parameters. These models are categorized according to different genders, heights, weights, and ages, with each category containing several human body models within preset data ranges to ensure the model's compatibility with the patient's anatomical structure. The implant configuration module retrieves the most suitable human body model from the database based on the patient information and marks the implant on the human body model based on the implant's basic location information. This module integrates an implant location mapping database, used to infer the type and typical anatomical location of implants based on the acquired implant location information and spatially mark them on the human body model, thereby establishing an individualized surgical environment model containing implant spatial constraints. During the marking process, the system determines the type and installation location of the implant based on the acquired basic implant location information and a conventional implant location lookup table, and then marks the implant on the human body model accordingly based on the implant's type and installation location. The implant placement reference table is generated based on historical experience data and includes the names, types and corresponding placement locations of common implants, such as cochlear implants in the ear, pacemakers in the heart, and metal implants in the limbs.

[0048] The human-computer interface is used to visually display a human body model marked with implants, and to receive surgical pose and surgical site annotation information, enabling the visual and interactive design of the surgical plan. A 3D human body heatmap is shown below. Figure 4 As shown, the implant is positioned as follows Figure 5As shown, doctors can intuitively see the implant's location on the interface and mark the surgical posture and intended surgical site. The system then uses this information for subsequent analysis and decision-making. The electrode application guidance module generates electrode application area instructions based on the implant-marked human model, surgical posture, and intended surgical site marking information, and displays them through the human-computer interaction interface. The electrode application database is constructed by integrating multi-source historical data and clinical rules to establish a mapping relationship between the implant marking model, surgical information, and electrode application areas. A data completion mechanism is used to supplement uncovered scenarios, thereby enhancing the system's robustness and adaptability in rare surgical situations. Historical data includes implant-marked human models, surgical posture, and intended surgical site marking information, as well as corresponding electrode application area information that has been verified in practice or manually reviewed. The system builds a mapping data table based on this historical data, supplementing missing historical data using artificial intelligence models or expert experience, ultimately constructing a complete electrode application database.

[0049] The electrode application guidance module is also configured to generate recommended power values ​​for surgical equipment based on electrode application area indication information and surgical type, thereby achieving integrated collaborative optimization from spatial design to equipment parameters. Electrode application area indication information is presented visually on the human model, including differentiated markings for recommended and contraindicated areas, intuitively assisting clinicians in making quick and accurate electrode application decisions. For example, the system uses green to indicate recommended electrode application areas and red to highlight prohibited areas on the human model. The system also includes a dynamic risk assessment unit, which performs real-time safety analysis based on the relative spatial relationship between the implant and the surgical area, triggering warning signals when approaching risk thresholds, thus enabling proactive early warning and intervention for potential intraoperative risks. The risk notification module generates a structured informed consent template based on basic patient information and pre-set risk warning information, automating the output from risk assessment to compliant documentation. Through the collaborative work of these modules, the system effectively reduces the risk of tissue damage and implant malfunction caused by improper current circuitry, improving the efficiency of surgical preparation and the standardization of management.

[0050] Example 2:

[0051] This embodiment further optimizes the data fusion and real-time inference process based on the aforementioned embodiments. By introducing multimodal data fusion and clinical rule embedding technology, it significantly improves the system's positioning accuracy and decision-making efficiency. The system first acquires the patient's biometric information and the spatial location information of the implanted device through the information acquisition module, providing a data foundation for personalized surgical design. The neural network model architecture is as follows: Figure 6As shown. The information acquisition module supports acquiring the location information of implants in the patient's body through non-contact or contact detection devices, and has a data interface for interfacing with the hospital information system, thereby achieving unified access and fusion of multi-source heterogeneous data and improving the completeness and efficiency of information acquisition. The human body model is a parametric 3D model constructed based on the fusion of patient biometrics and medical imaging data, which can reflect individualized anatomical structural features, thus providing a high-fidelity anatomical basis for spatial analysis and design. The implant configuration module matches the corresponding human body model according to the patient's biometrics and marks the model according to the spatial location information of the implant, thereby establishing an individualized surgical environment model containing implant spatial constraints. This module integrates an implant location mapping relationship library, which is used to infer the type and typical anatomical location of the implant based on the acquired implant location information, and to spatially mark it in the human body model, thereby improving the spatial positioning ability and marking accuracy of unknown implant types.

[0052] The human-computer interface (HCI) is used to visually display a human body model with implant markings and receive surgical input information, enabling the visual interactive design of surgical plans. The electrode application guidance module generates electrode application area indication information based on the human body model with implant markings and surgical information, combined with the electrode application database. Based on this, it outputs a current loop design scheme to avoid implant interference. The electrode application database is constructed by integrating multi-source historical data and clinical rules to establish a mapping relationship between implant marking models, surgical information, and electrode application areas. A data completion mechanism is used to improve the coverage of uncovered scenarios, thereby enhancing the system's decision-making robustness and adaptability in rare surgical situations. The system integrates a clinical rule engine, embedding contraindication area judgment logic based on medical safety standards to automatically exclude unsuitable body surface areas for electrode application. By embedding structured expert knowledge, the basic safety of the system's output scheme is ensured. The clinical rule engine constructs a clinical knowledge graph based on graph neural networks, representing clinical rules as a knowledge graph composed of entities and relationships. For example, there is a "prohibited application" relationship between "bone prominence" and "application location". By learning the embedded representation of knowledge graphs through graph convolutional networks or graph attention networks, complex relationships between entities can be captured, enabling real-time early warning and prevention functions.

[0053] The system also includes an intelligent optimization module, which uses graph-based data analysis to model the effectiveness of historical electrode placement and combines multi-objective decision-making principles to improve the recommendation accuracy and clinical adaptability of electrode placement positions. Furthermore, it optimizes the system's long-term decision-making performance through a continuous learning mechanism. The intelligent optimization module uses a multi-objective optimization algorithm to define a loss function to find the optimal placement position, considering factors such as contact impedance and minimum distance from moving joints. It uses graph convolutional networks to analyze the effectiveness of historical electrode positions, improving recommendation accuracy. The system also employs data augmentation strategies to address the scarcity of training data. Through geometric transformations, body shape synthesis, texture transformation, pathological feature synthesis, and surgical position-specific enhancement techniques, it generates a large number of diverse training samples, improving the model's adaptability to different body types, skin conditions, and surgical positions. The real-time inference process includes data acquisition and preprocessing, feature extraction, multimodal fusion, clinical rule validation, and placement position generation. The end-to-end latency of the entire process is controlled between 130-220 milliseconds, meeting the needs of real-time clinical applications. Through optimization strategies such as model quantization, model pruning, batch processing optimization, memory optimization, and hardware acceleration, the system can achieve real-time inference on ordinary medical-grade computing devices, providing clinicians with immediate suggestions on patch placement and improving surgical efficiency and safety. The risk disclosure module generates structured informed consent documents based on patient information and system risk assessments, automating the output from risk assessment to compliant documentation. Through the integration and optimization of the above technical solutions, the system achieves end-to-end digital traceability, and the system's decision-making basis can be retained as legal evidence, complying with electronic signature laws and significantly improving the safety and standardization of electrosurgical procedures.

[0054] Example 3:

[0055] This embodiment illustrates a specific case of generating system electrode application area indication information based on the aforementioned embodiments:

[0056] Case 1: Supine thyroid surgery + left lower extremity metal implant

[0057] Surgical position: neck extended, electrocautery applied to the anterior neck region, metal implant marking: left femoral titanium alloy plate, corresponding contraindication area: left lower limb (current tends to concentrate in metal).

[0058] After consulting the electrode application database, the recommended electrode application areas are:

[0059] Optimal area: Right anterolateral thigh (away from the surgical area + high conductivity on the same side); Alternative areas: Right upper arm posterior side / Right subscapular region; Absolutely forbidden area: Any part of the left lower limb.

[0060] Human-computer interaction interface demonstration:

[0061] [Pasteable Area] Right anterolateral thigh (15×10cm range) (dark green); [Alternative Area] Right upper arm posterior (10×8cm range) (light green); [Restricted Area] Left lower limb (red highlighted warning).

[0062] Case 2: Left lateral decubitus position for right lung surgery + cochlear implant

[0063] Surgical position: Right thoracotomy, electrocautery applied to the right hilum; Metal implant marking: Left intracranial cochlear implant electrode array; Contraindication area: Head and neck (risk of cochlear electrode exposure).

[0064] After consulting the electrode application database, the recommended electrode application areas are:

[0065] Optimal area: Left posterolateral thigh (opposite to surgery + away from the cochlea); Alternative area: Left buttock; Absolutely forbidden areas: Head and neck / shoulder and neck.

[0066] Human-computer interaction interface demonstration:

[0067] [Pasteable area] Left posterolateral thigh (avoiding the iliac crest protrusion area) (displayed in dark green);

[0068] [Special Warning] Do not wrap the negative electrode wire across the torso (to prevent the formation of an inductive loop).

[0069] Case 3: Prone position neurosurgery + pacemaker

[0070] Surgical position: posterior fossa craniotomy, electrocautery applied to the occipital region; metal implant marker: left chest subcutaneous pacemaker generator; contraindications: chest / abdomen (risk of pacemaker circuit interference).

[0071] After consulting the electrode application database, the recommended electrode application areas are:

[0072] Optimal area: right calf gastrocnemius muscle area; alternative area: left calf (must be far away from the pacemaker lead path); absolute no-go zone: any part of the torso.

[0073] Human-computer interaction interface demonstration:

[0074] [Pasteable area] Muscle area on the back of both calves (displayed in dark green);

[0075] [Warning Area] The distance between the negative electrode plate and the pacemaker generator is greater than 30cm (red dynamic ranging indicator).

[0076] Example 4:

[0077] Building upon the aforementioned embodiments, this method optimizes electrode placement by simulating the dynamic distribution path of high-frequency current in a personalized 3D human body model and introducing a risk weighting factor to quantify the potential hazards of current interaction with implants. Traditional methods rely on static rules or empirical distance thresholds, which cannot accurately reflect the actual behavior of current in complex anatomical structures, especially when multiple implants or heterogeneous tissues are present. This method constructs a physical model based on conductivity distribution and spatial constraints to simulate the current density field from the surgical point to the candidate electrode placement area and uses an original risk assessment function to dynamically adjust the recommended area. This ensures that the current loop avoids implants as much as possible while considering tissue conductivity uniformity, fundamentally reducing the risk of thermal damage and device interference.

[0078] The information acquisition module obtains the patient's biometrics and the spatial location information of the implant, and matches them with a parametric 3D human body model.

[0079] Based on the human tissue conductivity database and implant material parameters, conductivity values ​​are assigned to each voxel in the 3D model to construct an individualized conductivity distribution map.

[0080] The Poisson equation was solved using the finite element method to simulate the flow path of high-frequency current from the intended surgical site to the candidate electrode application point. A space-dependent damping term was introduced into the equation to reflect the attraction effect of the implant on the current.

[0081] Calculate the current density distribution corresponding to each candidate application area and extract the current density value passing through the implant area.

[0082] Define a risk assessment function to quantify the risk level of each candidate application site:

[0083]

[0084] In the formula, This is a risk score; the lower the value, the safer the application site. It is a dimensionless adjustable parameter used to balance the contribution ratio of current path risk terms in the total risk score; To pass the first Current density of the implant; Position the candidate electrode at the first... The Euclidean distance between the implants; The characteristic attenuation length; It is a dimensionless, adjustable parameter used to balance the contribution of the organizational homogeneity risk item to the total risk score; The coefficient of variation of electrical conductivity of the local tissue in the candidate application area.

[0085] The optimization objective is to minimize the risk score while constraining the application area to be located in an anatomically feasible area with abundant muscle and far from joints.

[0086] Candidate regions are sorted according to risk scores, and the region with the lowest score is selected as the recommended application site.

[0087] The risk distribution is displayed in the form of a heat map on the human-computer interaction interface. Green indicates low-risk recommended areas, and red indicates high-risk prohibited areas, with dynamic prompts.

[0088] Compared to traditional methods based on fixed safety distances or experience, this method significantly improves the safety of electrode placement through dynamic simulation and risk optimization. In simulating 200 surgical scenarios involving multiple implants, the error rate in electrode placement selection decreased from 7.2% with traditional methods to 0.8%, and the incidence of tissue carbonization decreased by 68%. The results demonstrate that this method can systematically avoid potential thermal damage areas caused by current concentration in implants or heterogeneous tissues, significantly reducing the risk of surgical complications due to improper circuit design. Furthermore, the optimization process is highly automated, transforming electrode placement selection from a subjective decision heavily reliant on expert experience into an objective, reproducible, and standardized process based on physical models and quantified risk. This effectively reduces the uncertainty introduced by human cognitive biases and improves the reliability and standardization of the overall surgical plan.

[0089] Example 5:

[0090] Building upon the aforementioned embodiments, this paper addresses the inconsistencies and noise issues associated with multi-source heterogeneous data in implant identification and localization. An adaptive fusion framework is designed to dynamically evaluate the reliability of each data source and utilizes a graphical model for confidence propagation, achieving high-precision implant location estimation. Traditional methods typically rely on a single data source or simple weighted averaging, making them susceptible to device errors or human input mistakes. This method introduces a confidence aggregation function to integrate multimodal features and optimizes the localization results based on spatial consistency constraints, thus maintaining robustness even with missing or conflicting data.

[0091] To address the risk assessment blind spots caused by missing implant information due to patient concealment or forgetfulness, a structured electronic preoperative questionnaire is introduced as a proactive information capture mechanism. This questionnaire is not simply a list of questions, but dynamically constructed based on a knowledge base covering common concealed, non-metallic, and cosmetic implants. The questions are designed with clear anatomical and clinical relevance, such as explicitly asking patients about their history of facial fillers, silicone implants, or orthopedic polymer implants. The questionnaire results are directly integrated into the system database in a structured data format, holding equal status with data sources such as medical images and text medical records, and participating in subsequent multi-source data fusion and confidence assessment processes. When the questionnaire feedback indicates the presence of a particular implant, while other test data are negative or missing, the system automatically increases the initial weight of that implant in the confidence fusion function and may generate a "Declared Implant Warning Area" on the 3D model requiring focused verification. Simultaneously, it drives the risk disclosure module to generate corresponding mandatory confirmation clauses in the informed consent document. This approach transforms subjective and easily overlooked patient self-reported information into objective and systematically processed structured data through systematic preoperative questioning. This effectively expands the coverage of implant risks from the source of information and enhances the robustness of the system in real clinical scenarios.

[0092] Metal artifact regions are extracted from CT images, and a deep learning segmentation network is used to output bounding boxes and confidence scores for implant candidate regions.

[0093] Signal intensity maps are acquired from a handheld metal scanner and generated as a signal heatmap through Gaussian filtering. Confidence scores for peak regions are then extracted.

[0094] The implant description is parsed from the patient's medical record text, and a natural language processing model is used to match it against a predefined implant library to output a text-based confidence score.

[0095] A graph neural network is constructed, where nodes represent candidate implant locations and edges represent spatial adjacency relationships. Adaptive weights are assigned to each data source and dynamically adjusted based on its signal-to-noise ratio and historical accuracy. The fusion confidence function is shown below:

[0096]

[0097] In the formula, To integrate confidence levels; These are adaptive weighting coefficients, corresponding to the real-time weights of CT during the fusion process; The confidence score is derived from CT image data; These are adaptive weighting coefficients, corresponding to the real-time weights of the scanner during the fusion process; The confidence score is the signal from the handheld scanner. The coefficient for inconsistent attenuation. This represents the spatial offset distance. These are adaptive weighting coefficients, corresponding to the real-time weights of the text data source during the fusion process; This represents the confidence score from the text medical record data.

[0098] The confidence level is propagated among candidate nodes using GNN, and the fusion result is optimized based on spatial smoothness constraints.

[0099] When outputting the final implant location, select the region with a fusion confidence score higher than the threshold and the largest spatial cluster, and mark it on the 3D human body model.

[0100] If the fusion confidence level is below the threshold, the system triggers a manual review process and records conflicting data sources for subsequent model optimization.

[0101] Compared to existing single-data-source or simple weighted fusion methods, this method significantly improves the accuracy and robustness of implant localization through adaptive weighting and confidence propagation, reducing the average localization error from 12 mm to 4 mm and the information omission rate from 10% to 1.5%. The results demonstrate that this method effectively addresses the noise, missing data, and spatial inconsistencies commonly found in multi-source heterogeneous data in clinical practice. By dynamically evaluating the reliability of each data source and performing optimal confidence fusion, it achieves precise localization not only when data quality is good but also maintains stable and reliable localization output even in challenging scenarios with significant errors in a single data source or conflicting data from different sources. This greatly enhances the ability to perceive and judge the spatial location information of implants within the body, providing a solid and reliable data foundation for all subsequent decision-making modules that rely on precise localization, ensuring the accuracy and safety of the entire decision-making process from the system input end.

Claims

1. A risk warning system for safe electrosurgical procedures, characterized in that, include: The information acquisition module is used to acquire the patient's biometric information and the spatial location information of the implant in the body, and a structured electronic preoperative questionnaire is introduced in advance as an active information capture mechanism. The questionnaire is dynamically constructed based on a knowledge base covering concealed, non-metallic and cosmetic implants. The questionnaire results are on an equal footing with medical images and text medical record data sources and participate in the multi-source data fusion and confidence assessment process. The human body model database stores multiple categories of 3D human body models constructed based on biometric parameterization. An implant configuration module is used to match a corresponding human body model based on the patient's biometrics and to mark the model based on the spatial location information of the implant. The human-computer interaction interface is used to visually display a human body model with implant markers and to receive surgical input information; The electrode application guidance module is used to generate electrode application area indication information based on the human body model with implant markings and surgical information, combined with the electrode application database. The risk disclosure module is used to generate structured informed consent documents based on patient information and system risk assessment. The information acquisition module and the implant configuration module work together to integrate medical imaging data, physical detection signals and text medical record information through a multi-source heterogeneous data fusion mechanism, and achieve precise positioning and marking of the implant location based on spatial consistency constraints and confidence propagation algorithms. The multi-source heterogeneous data fusion mechanism and confidence propagation algorithm specifically include: extracting metal artifact regions from CT images, using a deep learning segmentation network to output bounding boxes and confidence scores for implant candidate regions; acquiring signal intensity maps from scanners, generating signal heatmaps through Gaussian filtering, and extracting confidence scores for peak regions; parsing implant descriptions from patient medical record texts, using a natural language processing model to match a predefined implant library, and outputting text-based confidence scores; constructing a graph neural network where nodes represent candidate implant locations and edges represent spatial adjacency relationships, assigning adaptive weights to each data source, dynamically adjusting them based on their signal-to-noise ratio and historical accuracy, and the fusion confidence function is shown below: In the formula, To integrate confidence levels; These are adaptive weighting coefficients; Confidence score from CT image data; These are adaptive weighting coefficients; The confidence score is the signal from the handheld scanner. The coefficient for inconsistent attenuation. This represents the spatial offset distance. These are adaptive weighting coefficients; The confidence score is derived from the text medical record data; the confidence score is propagated among candidate nodes using GNN, and the fusion result is optimized based on spatial smoothness constraints; when outputting the final implant location, the region with the fusion confidence score higher than the threshold and the largest spatial cluster is selected and marked on the human model.

2. The system according to claim 1, characterized in that: The information acquisition module supports obtaining the location information of implants in the patient's body through non-contact or contact detection devices.

3. The system according to claim 1, characterized in that: The implant configuration module determines the type and installation location of the implant by using an implant location lookup table and marks it on the human model. The lookup table contains the correspondence between the name and type of the implant and its typical anatomical location.

4. The system according to claim 1, characterized in that: The electrode application database is constructed using multi-source historical data and clinical rules. It includes the mapping relationship between implant-marked human models, surgical poses, planned surgical sites and electrode application areas. Based on the electrode application area indication information and surgical type, it outputs recommended settings for surgical equipment parameters and uses a data completion mechanism to improve the coverage of uncovered scenarios.

5. The system according to claim 4, characterized in that: The electrode application guidance module is further configured to, based on the human body model containing implant markers and surgical information, dynamically simulate the distribution path of high-frequency current in the individualized model, and combine a risk weighting factor to quantitatively assess the interaction risk between the current path and the implant, thereby generating optimized electrode application area indication information.

6. The system according to claim 5, characterized in that: The electrode application area indication information is presented on the human body model in a visually distinguishable manner, including differentiated markings for recommended and prohibited areas.

7. The system according to claim 1, characterized in that: The human body model is a parametric three-dimensional model, which is constructed based on the patient's biometric data and achieves spatial mapping of anatomical structures by fusing surface and internal data.

8. The system according to claim 1, characterized in that: The system also includes a dynamic risk assessment module, which performs real-time safety analysis based on the relative spatial relationship between the implant location and the surgical area, and triggers a warning signal when the risk threshold is approached.

9. The system according to claim 1, characterized in that: The system integrates a clinical rule base and embeds contraindication judgment logic to automatically exclude unsuitable body surface areas for application, including areas with bony prominences, scar tissue, and adjacent to metal implants.

10. The system according to claim 1, characterized in that: The system analyzes historical application effectiveness data through graph neural networks to optimize the recommended accuracy of electrode application positions, and combines multi-objective optimization principles to balance impedance minimization and ease of operation.

Images