Data processing system based on dynamic adjustment of radio frequency plasma power

Abstract

This invention relates to the field of radiofrequency ablation power control and neurological safety monitoring technology, specifically a data processing system based on dynamic adjustment of radiofrequency plasma power. The system includes: a data acquisition module for real-time acquisition of multimodal input data streams, including instantaneous radiofrequency power, tissue impedance, local tissue surface temperature, and raw signal amplitudes from intraoperative neurophysiological monitoring; an interference identification module for determining interference identification markers; a thermal load estimation module for iteratively estimating the latent thermal load index; a safety verification module for calculating the neurological instability index; an optimal strategy generation module for calculating the optimal power strategy; a conflict detection module for determining model conflict markers; and a power adjustment module for determining the final power output. This invention prevents catastrophic surgical complications, ensures the integrity of biological nerves, and effectively avoids misjudgments caused by relying on only a single data source.

Description

Technical Field

[0001] This invention relates to the field of radiofrequency ablation power control and neural safety monitoring technology, specifically a data processing system based on dynamic adjustment of radiofrequency plasma power. Background Technology

[0002] Radiofrequency plasma ablation is a surgical technique that uses a radiofrequency generator to generate energy to treat tissue. During the procedure, the system needs to dynamically adjust the radiofrequency power to balance surgical efficiency and tissue safety, especially to avoid nerve damage.

[0003] To achieve this regulation, the system typically relies on physical model-based data processing methods. It estimates the accumulated, unmeasurable latent heat load within the tissue by monitoring observable data such as local tissue surface temperature. The system then adjusts its power strategy based on this estimated heat load index. However, in actual surgical environments, data sources are subject to specific interferences, particularly induced hypothermia caused by external cooling operations. This interference can lead to inaccurate local tissue surface temperature data, resulting in a rapid, non-physiological drop. If the data processing system relies solely on this contaminated temperature data to estimate the heat load, the latent heat load index will be severely underestimated. This situation can cause the system to misjudge the current state as safe and potentially continue executing a high-power strategy focused on efficiency. At this point, the actual heat load within the tissue may have already exceeded the safety threshold. Because the thermal model is blinded, the system cannot detect this risk, ultimately leading to uncontrolled latent heat load and potentially causing catastrophic surgical complications such as nerve damage, posing a serious safety hazard. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a data processing system based on dynamic adjustment of radio frequency plasma power. Specifically, the technical solution of this invention includes:

[0005] The data acquisition module is used to acquire multimodal input data streams in real time; the multimodal input data streams include: instantaneous radio frequency power, tissue impedance, local tissue surface temperature, and raw signal amplitudes from intraoperative neurophysiological monitoring;

[0006] An interference identification module is used to determine interference identification markers based on local tissue surface temperature and tissue impedance.

[0007] The heat load estimation module is used to iteratively estimate the latent heat load index based on instantaneous radio frequency power and local tissue surface temperature;

[0008] The safety verification module is used to calculate the neural instability index based on the raw signal amplitude of intraoperative neurophysiological monitoring.

[0009] The optimal strategy generation module is used to calculate the optimal power strategy based on the latent heat load index and organizational impedance.

[0010] The conflict detection module is used to determine the model conflict markers based on interference identification markers, latent heat load index, and neural instability index;

[0011] The power regulation module is used to arbitrate between the optimal power strategy and the preset safe power strategy in response to the model conflict flag, so as to determine the final power output.

[0012] Preferably, the interference identification module is specifically used for:

[0013] Calculate the time rate of change of local tissue surface temperature and the time rate of change of tissue impedance.

[0014] Preferably, the interference identification module is further used for:

[0015] When the time change rate of local tissue surface temperature is less than the preset temperature drop rate threshold and the time change rate of tissue impedance is greater than the preset impedance rise rate threshold, the interference identification flag is determined to be in the interference activation state.

[0016] Otherwise, the interference identification flag is determined to be in a normal state.

[0017] Preferably, the heat load estimation module is specifically used for:

[0018] Based on the latent heat load index of the previous moment, the instantaneous radio frequency power of the current moment, the local tissue surface temperature of the current moment, the preset physiological baseline body temperature, the preset energy accumulation coefficient, and the preset heat dissipation estimation coefficient, the latent heat load index of the current moment is calculated iteratively.

[0019] Preferably, the security verification module is specifically used for:

[0020] Calculate the standard deviation of the raw signal amplitude from intraoperative neurophysiological monitoring within a preset time window;

[0021] And the slope obtained by performing linear regression fitting on the original signal amplitude within a preset time window before the current time is used as the time change rate of the original signal amplitude.

[0022] Preferably, the security verification module is further used for:

[0023] The neural instability index is calculated based on the time rate of change, standard deviation, preset healthy baseline amplitude, preset rate of change weight, and preset signal jitter weight.

[0024] Preferably, the collision detection module is specifically used for:

[0025] When the interference identification flag is in the interference activation state, the latent heat load index is less than the preset latent heat load safety threshold, and the neural instability index is greater than the preset neural instability warning threshold, the model conflict flag is determined to be in the conflict activation state.

[0026] Otherwise, the model conflict flag is set to a non-conflict state.

[0027] Preferably, the power regulation module is specifically used for:

[0028] When the model conflict flag is in a non-conflict state, the optimal power strategy is selected as the final power output.

[0029] Preferably, the power regulation module is further used for:

[0030] When the model conflict flag is in a conflict active state, a preset safe power strategy is selected as the final power output.

[0031] Preferably, the optimal power strategy is a strategy calculated based on a deep reinforcement learning model;

[0032] The preset safe power policy is a preset set of rules.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. This system solves the problem of inaccurate temperature data and blinded thermal models caused by external cooling interference by constructing a dual model of heat load estimation and neuroelectrophysiological verification. The system can identify conflicts between the data model and biological reality, avoiding safety misjudgments;

[0035] 2. This system has established a conflict detection mechanism; when it is detected that the thermal model falsely reports safety due to data contamination, but the independent neuroelectrophysiology has shown danger, the system can proactively abandon the optimal strategy that pursues efficiency and switch to the preset safety strategy to ensure surgical safety.

[0036] 3. This system improves the robustness and antifragility of the data processing system in complex interference environments; the system can proactively manage the latent thermal load that gets out of control due to model blindness, preventing catastrophic surgical complications and ensuring the integrity of biological neural networks;

[0037] 4. By cross-validating the time-varying rate of change of local tissue surface temperature and the time-varying rate of change of tissue impedance, this system can accurately identify specific interference events, providing key prior information for downstream conflict detection and effectively avoiding misjudgments caused by relying on a single data source. Attached Figure Description

[0038] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0039] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0041] Example 1:

[0042] Please see Figure 1 A data processing system based on dynamic adjustment of radio frequency plasma power includes:

[0043] The data acquisition module is used to acquire multimodal input data streams in real time; the multimodal input data streams include: instantaneous radio frequency power, tissue impedance, local tissue surface temperature, and raw signal amplitudes from intraoperative neurophysiological monitoring;

[0044] An interference identification module is used to determine interference identification markers based on local tissue surface temperature and tissue impedance.

[0045] The heat load estimation module is used to iteratively estimate the latent heat load index based on instantaneous radio frequency power and local tissue surface temperature;

[0046] The safety verification module is used to calculate the neural instability index based on the raw signal amplitude of intraoperative neurophysiological monitoring.

[0047] The optimal strategy generation module is used to calculate the optimal power strategy based on the latent heat load index and organizational impedance.

[0048] The conflict detection module is used to determine the model conflict markers based on interference identification markers, latent heat load index, and neural instability index;

[0049] The power regulation module is used to arbitrate between the optimal power strategy and the preset safe power strategy in response to the model conflict flag, so as to determine the final power output.

[0050] This embodiment provides a data processing system based on dynamic adjustment of radio frequency plasma power; in a preferred embodiment, the system can be deployed on the main control computer of the radio frequency ablation device or a parallel data processing unit. In this invention, G06F generally refers to calculation or control logic that conforms to the G06F classification number definition, and the same applies below.

[0051] The system includes: a data acquisition module, an interference identification module, a heat load estimation module, a safety verification module, an optimal strategy generation module, a conflict detection module, and a power regulation module;

[0052] Data acquisition module: The purpose of this module is to provide real-time, multi-dimensional data input to all downstream analysis modules of the system, serving as the basis for the G06F system's analysis and decision-making. In this embodiment, this module communicates with the radio frequency generator, temperature sensor, impedance analyzer, and intraoperative neurophysiological monitoring IONM device in the operating room through a data interface.

[0053] This module is used to acquire multimodal input data streams in real time; the data stream specifically includes: instantaneous radio frequency power. This value is fed back in real time by the main controller of the radio frequency generator, and its dimension is watts (W); tissue impedance This value is measured in real time by the impedance analyzer built into the RF generator, and its dimension is ohms. Local tissue surface temperature This value was acquired by the implanted temperature sensor, with dimensions in degrees Celsius (°C); and the raw signal amplitude from intraoperative neurophysiological monitoring. This value is acquired and output by IONM equipment, and its dimension is microvolts. ;

[0054] Interference Identification Module: This module monitors and identifies specific data contamination events that may cause thermal model inaccuracies in real time, especially induced low-temperature interference; this interference is characterized by being caused by external cooling operations. A rapid, non-physiological decline occurred, at the same time An abnormal rise due to low temperature;

[0055] This module is used for applications based on local tissue surface temperature. and tissue resistance Determine interference identification markers ; It is a binary logic flag, calculated by this module, used to indicate the current status to downstream modules, especially the conflict detection module. The reliability of the data is determined by a value of 0, which represents a normal state, or 1, which represents an active interference state.

[0056] Heat load estimation module: This module estimates a latent heat load that cannot be directly measured but is critical to neural safety; this module is based on observable instantaneous radio frequency power. and reliable local tissue surface temperature under undisturbed conditions The latent heat load index is estimated using a thermal accumulation iterative model. ; It is a dimensionless index, calculated iteratively by this module, used to quantify the level of accumulated heat stress within the tissue that has not yet been dissipated through the surface; it is used as the primary basis for assessing the long-term neural safety threats posed by current power strategies.

[0057] Security verification module: This module provides a separate... The biological truth verification channel, in order to solve The model in The problem of blindness caused by interference; this module does not rely on thermodynamic parameters, but rather analyzes the raw signal amplitudes from intraoperative neurophysiological monitoring. The subtle trend changes can indirectly reflect the true state of neurological health.

[0058] This module is used to calculate the neural instability index. ; It is a dimensionless exponent, based on this module. The signal is obtained through feature engineering calculations and used for quantization. The signal exhibits a weak, unstable trend prior to a catastrophic collapse, i.e., loss of neural function;

[0059] Optimal Strategy Generation Module: The purpose of this module is to calculate a theoretically optimal power output based on the current thermal state and impedance characteristics of the system. In this embodiment, this module is a pre-trained deep reinforcement learning (DRL) model.

[0060] This module is used for applications based on latent heat load index. and tissue resistance Calculate the optimal power strategy ; It is the DRL agent's view at time t that maximizes surgical efficiency while simultaneously Maintain at the safety threshold The following RF power values ​​are output from this DRL model and are expressed in watts (W).

[0061] Conflict Detection Module: This module serves as the core of the system's decision-making arbitration, used to identify hot models. Validation model with IONM A fatal conflict between them;

[0062] This module is used for interference identification flags. Latent heat load index and the neural instability index Determine model conflict flags ; It is a binary logic flag, calculated by this module, used to indicate whether the system's data model has been contaminated and is unreliable, i.e., the thermal model shows safety, but the biological model shows danger; its value is 0 to represent a non-conflict state or 1 to represent a conflict activation state.

[0063] Power Regulation Module: The purpose of this module is to execute the final decision of the G06F system, construct a G06F decision selector, and adjust the power regulation module based on the conflict flag. Switching between optimal and security strategies;

[0064] This module is used in response to model conflict flags. In the optimal power strategy and preset safe power strategy Arbitration switching is performed between the two to determine the final power output. ; It is a pre-defined rule set of G06F, designed with security as the primary goal rather than efficiency; It is the power value that the system ultimately decides and outputs to the RF generator for execution. It is determined by the arbitration of this module, and its unit is watts (W).

[0065] This system constructs a heat load estimation mechanism. and neurophysiological verification The dual model is used, and a conflict detection module is established. This system provides a profound understanding of the internal state of the G06F data processing system; it is also capable of operating within a thermal model. Due to external interference When a data model becomes ineffective due to blindness, it is crucial to promptly identify the fatal conflict between this data model and biological truth; once the conflict is activated... The system immediately abandons the optimal strategy of pursuing short-term efficiency. Switch to security policy This proactively manages inaccurate and invisible latent heat loads, ensuring the integrity of the biological nervous system, avoiding catastrophic surgical complications, and achieving antifragility and robustness of the G06F system in complex interference environments.

[0066] Example 2:

[0067] The interference identification module is specifically used for:

[0068] Calculate the time rate of change of local tissue surface temperature and the time rate of change of tissue impedance;

[0069] The interference identification module is also used for:

[0070] When the time change rate of local tissue surface temperature is less than the preset temperature drop rate threshold and the time change rate of tissue impedance is greater than the preset impedance rise rate threshold, the interference identification flag is determined to be in the interference activation state.

[0071] Otherwise, the interference identification flag is determined to be in a normal state.

[0072] This embodiment is a concretization of the interference identification module in Embodiment 1; it aims to achieve high-precision identification of thermal models. Inaccurate induced low-temperature interference; the interference identification module is specifically used for:

[0073] Calculate the rate of change of local tissue surface temperature over time and the rate of change of tissue impedance over time ; and They are and The first derivative of a signal reflects its trend of change over a short period of time, and can be obtained by analyzing the data input from the data acquisition module. and The module performs real-time differencing or linear fitting within a short sliding window to obtain the slope from the time series data. Then, it executes the following interference identification logic to determine the interference identification flag. : ;

[0074] in: This is an interference identification flag; a value of 1 indicates an active interference state, and 0 indicates a normal state. surface temperature The rate of change over time, in °C / s; Tissue impedance The rate of change over time, with dimensions of / s; This is a preset temperature drop rate threshold; it is a preset positive value used to define the non-physiological temperature drop rate, with the dimension °C / s. This is a preset impedance rise rate threshold; it is a preset positive value used to define the non-physiological impedance rise rate related to low temperature, with dimensions of _____. / s;

[0075] threshold and It is a key calibration parameter of the G06F system; its determination process involves offline calibration of historical surgical data; analysis of multiple data segments in the historical dataset where induced hypothermia operations are known to have occurred, extraction of the time change rate of local tissue surface temperature and the time change rate of tissue impedance in these data segments, and based on their statistical distribution, such as taking the 95th percentile to set a threshold sufficient to distinguish between anesthetic interference and normal physiological fluctuations.

[0076] The rate of change of local tissue surface temperature over time Less than the preset temperature drop rate threshold The negative value, that is This indicates that the temperature is exceeding [a certain value]. The rate of decrease is rapid, and the time-varying rate of tissue impedance is... Greater than the preset impedance rise rate threshold , This indicates that when the impedance is rising rapidly, the system determines that induced cryogenic interference has occurred and identifies the interference flag. In the interference activation state, ;

[0077] Otherwise, in other cases, identify interference identification markers. This is the normal state. ;

[0078] pass Temperature changes and By cross-validating the physical characteristics of impedance changes in two dimensions, this module can accurately identify specific induced cryogenic interferences from complex surgical data, effectively avoiding reliance solely on... Misjudgments caused by absolute values; The generation of the flag provides crucial prior information on data contamination for the downstream conflict detection module, and is a prerequisite for the startup of the entire G06F antifragile system.

[0079] Example 3:

[0080] The heat load estimation module is specifically used for:

[0081] Based on the latent heat load index of the previous moment, the instantaneous radio frequency power of the current moment, the local tissue surface temperature of the current moment, the preset physiological baseline body temperature, the preset energy accumulation coefficient, and the preset heat dissipation estimation coefficient, the latent heat load index of the current moment is calculated iteratively.

[0082] This embodiment is a concretization of the heat load estimation module in Embodiment 1; this module aims to estimate the invisible latent heat load index based on observable physical parameters. This module constructs a G06F thermal accumulation iterative model based on first-order energy balance logic: the thermal load at time t is equal to the load at time t−1, plus the accumulation term introduced by power, minus the heat dissipation term estimated based on surface temperature.

[0083] The heat load estimation module is used to execute the following iterative estimation formula to iteratively calculate the current time. Latent heat load index : ;

[0084] in: For the current moment The estimated latent heat load index is the result of this iteration and is dimensionless; For the previous moment The latent heat load index is the output of the previous calculation cycle of the G06F system and is dimensionless. For the current moment The instantaneous radio frequency power is provided by the data acquisition module and is measured in W. For the current moment The local tissue surface temperature is provided by the data acquisition module and is measured in °C. The preset physiological baseline body temperature is either a preset biological routine reference value or a value determined based on the patient's body temperature upon entering the operating room, and its dimension is °C. This is a preset energy accumulation coefficient; its function is to convert the input power... Converted to a dimensionless cumulative heat load index contribution, with dimensions of ; This is a preset heat dissipation estimation coefficient; its function is to estimate the surface temperature difference. Converted to a dimensionless thermal expansion contribution, with dimensions of ;

[0085] parameter and It is based on simplified model parameters of the biological heat transfer equation; its calibration is performed by fitting data from in vitro phantom experiments or animal experiments; a set of known power curves are applied in the calibration experiment. Simultaneously, a deep tissue thermal stress index that reflects true thermal damage was measured. ,in The experimental time was adjusted by optimizing the algorithm, such as least squares regression analysis. and The value of makes it possible to use this formula and... Calculated as input Curve and experimentally measured The fitting error between curves is minimized;

[0086] It should be emphasized that this G06F model heavily relies on The accuracy of the data; although the formula

[56] uses The function ensures that the heat dissipation term is in It will not be incorrectly calculated as a heat source, thus improving the physical fidelity of the model; when disturbances are activated hour, The temperature was artificially lowered; this caused the heat dissipation item to be affected. The estimation error means that the heat dissipation was overestimated. The value was severely underestimated, causing the G06F system to misjudge security.

[0087] This embodiment provides a computationally lightweight and physically clear method for estimating latent heat load; it quantifies the risk of invisible tissue thermal damage into a value that can be used in DRL models. Input and security thresholds Comparison index This provides a core basis for realizing closed-loop power control of G06F based on thermal state.

[0088] Example 4:

[0089] The security verification module is specifically used for:

[0090] Calculate the standard deviation of the raw signal amplitude from intraoperative neurophysiological monitoring within a preset time window;

[0091] And the slope obtained by performing linear regression fitting on the original signal amplitude within a preset time window before the current moment is used as the time change rate of the original signal amplitude.

[0092] The security verification module is also used for:

[0093] The neural instability index is calculated based on the time rate of change, standard deviation, preset healthy baseline amplitude, preset rate of change weight, and preset signal jitter weight.

[0094] This embodiment is a concretization of the security verification module in Embodiment 1; the G06F verification model aims to analyze IONM signals. The subtle trend changes are used to indirectly reflect the true state of neural health. The core logic is that neural instability is equal to the weighted sum of the short-term rate of change of the signal relative to the baseline and the short-term jitter of the signal relative to the baseline.

[0095] To achieve this goal, the security verification module needs to... The signal undergoes feature engineering to calculate two key intermediate metrics:

[0096] signal jitter :

[0097] This value reflects The signal is The degree of short-term volatility or jitter around a given time;

[0098] The module calculates the raw signal amplitude of intraoperative neurophysiological monitoring. exist A preset time window near the time The standard deviation within the range, as ;

[0099] Signal change rate :

[0100] This value reflects The signal is A short-term downward trend near the specified time;

[0101] To resist transient noise interference, this module uses the current time... The same preset time window before Internal to the original signal amplitude The slope obtained by linear regression fitting is used as the rate of change of the original signal amplitude over time. This method is more robust than simple single-step differencing.

[0102] Preset time window This is a calibration parameter, the specific value of which is preset based on the signal sampling rate and noise level, such as... Second;

[0103] After obtaining the above two features, the security verification module is also used to verify the time change rate. Standard deviation Preset health baseline amplitude Preset rate of change weights and preset signal jitter weights Calculate the final neural instability index The calculation formula is as follows: ;

[0104] in: For the current moment The neural instability index is the result of this calculation and is dimensionless. The rate of change over time, i.e., the slope, is obtained from the linear regression fitting described above, and its dimensions are... ; The standard deviation, or jitter, is calculated above and has the following dimensions: ; The preset healthy baseline amplitude is measured by collecting stable data before surgery, i.e., when the nerves are not subjected to thermal stress. The signal is calibrated, with dimensions of ; Preset weights for rate of change; used to adjust the rate of change. right The contribution, measured in seconds (Time), ensures the first item It is dimensionless; Preset signal jitter weights; used to adjust jitter. right The contribution is dimensionless, ensuring the second term. It is dimensionless; This is a preset numerical stability parameter; it is a very small positive constant, such as... Used to prevent when The model crashes due to division by zero when calibrated to 0 or close to 0, ensuring the computational robustness of the G06F model;

[0105] Weight and The values ​​were trained using machine learning methods on an intraoperative IONM historical dataset containing a large number of known clinical states, such as 9 / 10 risk states and safe states of impending nerve injury; for each signal in this training set... Extract its corresponding features and and give it a clinical label. For example, if risk = 1 and safety = 0, the goal of optimization is to find a set of... and This makes the calculation based on these features In distinguishing different When labeling, the maximum discriminative power can be achieved by evaluating the area under the ROC curve.

[0106] This embodiment utilizes G06F feature engineering, specifically the linear regression slope. and standard deviation rather than original Value, effectively quantified The subtle instability of the signal before it collapses avoids misjudgment caused by transient noise. Completely independent of Data, making it in Contaminated It is the only reliable source of biological truth at the time, providing an indispensable verification signal for conflict detection.

[0107] Example 5:

[0108] The collision detection module is specifically used for:

[0109] When the interference identification flag is in the interference activation state, the latent heat load index is less than the preset latent heat load safety threshold, and the neural instability index is greater than the preset neural instability warning threshold, the model conflict flag is determined to be in the conflict activation state.

[0110] Otherwise, the model conflict flag is set to a non-conflict state.

[0111] This embodiment is a concretization of the conflict detection module in Embodiment 1; the G06F module is the core of the system's decision arbitration, and it identifies a high-risk blinding state: namely, interference. Leading to thermal models Misleading, showing security, while independent biometric verification However, it indicates danger;

[0112] To achieve this objective, the conflict detection module specifically executes the following G06F model conflict detection logic to determine model conflict flags. : ;

[0113] A model conflict flag is determined if and only if the following three conditions are met simultaneously. Conflict activated state :

[0114] Interference identification flag In the interference activation state, ,show The data source has been contaminated;

[0115] Latent heat load index Less than the preset latent heat load safety threshold This indicates that the contaminated thermal model is safe from false alarms.

[0116] And the neural instability index Greater than the preset neural instability warning threshold This indicates that the independent biological validation model reports a risk;

[0117] Otherwise, in any other case, such as, without interference Or both models consistently show safety. Or both models consistently show danger. Determine model conflict flags It is in a non-conflict state. ;

[0118] in: This is a conflict marker for the G06F model, and is the result of this calculation; it is dimensionless. This is an interference identification marker, originating from the interference identification module, and is dimensionless. The latent heat load index is derived from the heat load estimation module and is dimensionless. The neural instability index is derived from the security verification module and is dimensionless. The latent heat load safety threshold is one of the core safety boundaries of the G06F system; its value is based on... The offline calibration data is determined, for example, by analyzing historical data corresponding to a risk status of 9 / 10. The value distribution, taking its 80th percentile, is dimensionless; The warning threshold for neural instability is the second core security boundary of the G06F system; its value is based on... ROC curve analysis confirmed that this ROC curve is also based on the aforementioned... The training dataset was generated by selecting datasets with high sensitivity before 10 / 10 failure, such as 95%. The value is dimensionless;

[0119] pass , and The AND logic of the three G06F logic indicators, this module precisely defines and captures the most dangerous model-blinding states of the system. ; The generation of the flag is the G06F decision trigger for the system to switch from pursuing the optimal to ensuring survival. It identifies the fatal conflict between pursuing the G06F optimal solution on the short-term algorithm model and maintaining the biological integrity on which the long-term system survival depends, and provides a clear arbitration instruction for subsequent power regulation.

[0120] Example 6:

[0121] The power regulation module is specifically used for:

[0122] When the model conflict flag is in a non-conflict state, the optimal power strategy is selected as the final power output;

[0123] The power regulation module is also used for:

[0124] When the model conflict flag is in a conflict activation state, a preset safe power strategy is selected as the final power output; the optimal power strategy is a strategy calculated based on the deep reinforcement learning model.

[0125] The preset safe power policy is a preset set of rules.

[0126] This embodiment is a concretization of the power regulation module in Embodiment 1, and incorporates the definition of the optimal strategy generation module; this module is the final actuator of the G06F system, and its purpose is to determine the arbitration result of the conflict detection module. Switching between two distinct power strategies;

[0127] Define two input strategies:

[0128] Optimal power strategy This strategy is based on the Deep Reinforcement Learning (DRL) model, i.e., the strategy of technique A computation.

[0129] Its technical considerations are Model credibility means Under the premise of maximizing surgical efficiency index A and hemostasis effect index B;

[0130] Its source is DRL agent according to and Calculated;

[0131] Safe power policy This strategy is a pre-defined set of rules whose sole objective is safety, i.e., to reduce [the risk] as quickly as possible. It's about efficiency, not efficiency;

[0132] In this embodiment, It can be a fixed value, such as That is, a forced cooling cycle; or a simple negative feedback controller, such as ,Right now The higher, The lower the value, the faster its design goal is to achieve... Falling back to the following;

[0133] Its source is the G06F preset rule;

[0134] The power regulation module executes a G06F decision arbitration switch, i.e., a 2-to-1 multiplexer, to determine the final power output. : ;

[0135] in: The final power output is the execution value sent to the RF generator after system decision, and its dimension is W; This is the conflict flag for the G06F model, with a value of 0 or 1. The optimal power strategy calculated for DRL, in W; The suboptimal safety power strategy preset for G06F is in the dimension of W;

[0136] The logical derivation of the G06F arbitration formula is as follows:

[0137] When model conflict flag Non-conflict state ( When the formula becomes ), it indicates that the system model is consistent and reliable; at this point, the formula becomes The system selects the optimal power strategy. As the final power output, it aims to achieve the highest surgical efficiency;

[0138] When model conflict flag Conflict-activated state ( When this condition is met, it indicates that the system model has been contaminated and is unreliable; at this point, the formula becomes... The system selects a preset safe power strategy. As the final power output;

[0139] This embodiment implements the antifragility of the G06F system; when the system detects a model conflict... At that time, the G06F arbitration logic enables the system to proactively relinquish the arbitration by the DRL model. The provided data is based on erroneous data and is therefore low. Instead of finding the spurious optimal solution, a conservative but reliable approach is adopted. Although this approach sacrifices resection efficiency in the short term, it ensures the integrity of the biological neural network, prevents catastrophic failures caused by model blindness, and greatly improves the robustness and security of the G06F data processing system.

[0140] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A data processing system based on dynamic adjustment of radio frequency plasma power, characterized in that, include: The data acquisition module is used to acquire multimodal input data streams in real time; the multimodal input data streams include: instantaneous radio frequency power, tissue impedance, local tissue surface temperature, and raw signal amplitudes from intraoperative neurophysiological monitoring; An interference identification module is used to determine interference identification markers based on local tissue surface temperature and tissue impedance. The heat load estimation module is used to iteratively estimate the latent heat load index based on instantaneous radio frequency power and local tissue surface temperature; The safety verification module is used to calculate the neural instability index based on the raw signal amplitude of intraoperative neurophysiological monitoring. The optimal strategy generation module is used to calculate the optimal power strategy based on the latent heat load index and tissue impedance. The conflict detection module is used to determine the model conflict markers based on interference identification markers, latent heat load index, and neural instability index. The power regulation module is used to arbitrate between the optimal power strategy and the preset safe power strategy in response to the model conflict flag, so as to determine the final power output.

2. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 1, characterized in that, The interference identification module is specifically used for: Calculate the time rate of change of local tissue surface temperature and the time rate of change of tissue impedance.

3. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 2, characterized in that, The interference identification module is also used for: When the time change rate of local tissue surface temperature is less than the preset temperature drop rate threshold and the time change rate of tissue impedance is greater than the preset impedance rise rate threshold, the interference identification flag is determined to be in the interference activation state. Otherwise, the interference identification flag is determined to be in a normal state.

4. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 1, characterized in that, The heat load estimation module is specifically used for: Based on the latent heat load index of the previous moment, the instantaneous radio frequency power of the current moment, the local tissue surface temperature of the current moment, the preset physiological baseline body temperature, the preset energy accumulation coefficient, and the preset heat dissipation estimation coefficient, the latent heat load index of the current moment is calculated iteratively.

5. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 1, characterized in that, The security verification module is specifically used for: Calculate the standard deviation of the raw signal amplitude from intraoperative neurophysiological monitoring within a preset time window; And the slope obtained by linear regression fitting of the original signal amplitude within a preset time window before the current moment is used as the time change rate of the original signal amplitude.

6. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 5, characterized in that, The security verification module is also used for: The neural instability index is calculated based on the time rate of change, standard deviation, preset healthy baseline amplitude, preset rate of change weight, and preset signal jitter weight.

7. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 1, characterized in that, The collision detection module is specifically used for: When the interference identification flag is in the interference activation state, the latent heat load index is less than the preset latent heat load safety threshold, and the neural instability index is greater than the preset neural instability warning threshold, the model conflict flag is determined to be in the conflict activation state. Otherwise, the model conflict flag is set to a non-conflict state.

8. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 1, characterized in that, The power regulation module is specifically used for: When the model's conflict flag is in a non-conflict state, the optimal power strategy is selected as the final power output.

9. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 8, characterized in that, The power regulation module is also used for: When the model conflict flag is in a conflict active state, a preset safe power strategy is selected as the final power output.

10. The data processing system based on dynamic adjustment of radio frequency plasma power according to claim 1, characterized in that, The optimal power strategy is a strategy calculated based on a deep reinforcement learning model; The preset safe power policy is a preset set of rules.

Images