Emergency treatment process automatic processing method based on multiple modes

Abstract

The invention provides an emergency treatment process automatic processing method based on multiple modes, and the method comprises the steps: collecting the microcirculation dynamic waveform of the skin of a patient and extracting the directional voice semantic features under the environment noise through a medical robot at an out-of-hospital first-aid site; constructing a three-dimensional space-time compensation field and generating a composite feature flow; when the consciousness state grading index of the composite characteristic flow is lower than a preset threshold value, the tissue perfusion elastic coefficient is calculated through the phase mutation characteristic of the microcirculation waveform; based on the data, a three-layer dynamic decision tree topology network is constructed, the first layer generates a hemostasis intervention intensity spectrum, the second layer generates neural function injury probability distribution, and the third layer corrects robustness parameters of a mechanical arm operation path; a closed-loop emergency treatment scheme is formed by monitoring harmonic component change characteristics and generating an anti-interference control loop; according to the invention, the accuracy and anti-interference capability of out-of-hospital first aid are improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical fields of medical robots, artificial intelligence, signal processing, and automation control, and particularly to a multi-mode-based automated processing method for emergency treatment processes. Background Art

[0002] At the out-of-hospital emergency scene, it is crucial to quickly and accurately assess the patient's physiological state and implement effective first aid measures. Especially in complex environments, such as under conditions of noise interference, insufficient light, or ambulance jolts, traditional first aid methods often struggle to achieve precise operations. Therefore, there is an urgent need for a technical solution that can real-time monitor the dynamic microcirculation waveforms of patients, extract directional speech semantic features under environmental noise, and generate intelligent first aid decisions based on this data. Such a technology not only requires high-precision data processing capabilities but also needs to adaptively adjust in a dynamic environment to ensure the accuracy and stability of first aid operations.

[0003] Currently, some first aid assistance systems based on artificial intelligence and machine learning have been proposed. These systems can collect the patient's physiological data through sensors and conduct preliminary analysis in combination with environmental data. For example, some systems use deep learning algorithms to denoise the patient's speech signals, extract key semantic features, and generate preliminary first aid suggestions in combination with physiological data. In addition, some systems have also introduced robotic arm-assisted operations, which execute first aid tasks in a relatively stable environment through preset path planning algorithms.

[0004] Although the existing solutions have improved the intelligence level of first aid to a certain extent, they still have deficiencies in processing data in complex environments. For example, when existing systems extract directional speech semantic features, they often cannot effectively separate environmental noise from the patient's speech signals, resulting in a decline in the accuracy of semantic analysis and difficulty in maintaining stable operation accuracy in complex environments. These problems severely limit the application effect of existing systems in real first aid scenarios. Summary of the Invention

[0005] The embodiments of the present application provide a multi-mode-based automated processing method for emergency treatment processes to solve the problems of low accuracy and poor anti-interference ability in out-of-hospital first aid in the prior art.

[0006] In a first aspect, the embodiments of the present application provide a multi-mode-based automated processing method for emergency treatment processes, including:

[0007] At the out-of-hospital emergency scene, collect the dynamic microcirculation waveforms of the patient's skin through the medical robot, and extract the directional speech semantic features under environmental noise interference;

[0008] Compare the dynamic microcirculation waveform and the directional voice semantic features in the spatio-temporal dimension to construct a three-dimensional spatio-temporal compensation field, and generate a composite feature stream based on the three-dimensional spatio-temporal compensation field;

[0009] When the consciousness state grading index of the composite feature stream is lower than the preset index threshold, calculate the tissue perfusion elasticity coefficient through the phase mutation feature of the dynamic microcirculation waveform;

[0010] Based on the composite feature stream and the tissue perfusion elasticity coefficient, construct a three-layer dynamic decision tree topology network. The first layer generates a hemostasis intervention intensity spectrum based on the dynamic microcirculation waveform and the composite feature stream. The second layer generates a probability distribution of nerve function damage according to the tissue perfusion elasticity coefficient and the directional voice semantic features. The third layer combines the real-time collected environmental light intensity and the ambulance bump index to correct the robustness parameters of the robotic arm operation path;

[0011] Based on the hemostasis intervention intensity spectrum, monitor the change characteristics of the harmonic components of the microcirculation waveform. Based on the probability distribution of nerve function damage, feedback the robustness parameters to the motion planner of the robotic arm to generate an anti-interference control loop. Based on the change characteristics of the harmonic components and the anti-interference control loop, generate a closed-loop first aid treatment plan.

[0012] Optionally, the first layer generates a hemostasis intervention intensity spectrum based on the dynamic microcirculation waveform and the composite feature stream. The second layer generates a probability distribution of nerve function damage according to the tissue perfusion elasticity coefficient and the directional voice semantic features. The third layer combines the real-time collected environmental light intensity and the ambulance bump index to correct the robustness parameters of the robotic arm operation path, including:

[0013] In the first layer of the dynamic decision tree topology network, extract the microcirculation waveform attenuation rate from the dynamic microcirculation waveform, and analyze the spatial diffusion direction parameter from the trauma diffusion trend vector of the composite feature stream. Dynamically divide the hemostasis pressure level according to the microcirculation waveform attenuation rate and the spatial diffusion direction parameter, and generate a hemostasis intervention intensity spectrum in combination with the current body position stability of the patient;

[0014] In the second layer of the dynamic decision tree topology network, when the semantic interruption frequency in the directional voice semantic features is higher than the preset frequency threshold, non-linearly superimpose the tissue perfusion elasticity coefficient and the number of semantic logic breakpoints in the directional voice semantic features to generate a nerve function damage contradiction index, and generate a probability distribution of nerve function damage within a preset probability interval according to the nerve function damage contradiction index;

[0015] In the third layer of the dynamic decision tree topology network, the bump spectrum characteristics are collected in real time through the vibration sensors installed on the ambulance, the influence coefficient of the ambient light intensity collected in real time on the visual capture sensitivity of the medical robot is determined, the ambulance bump index is calculated based on the bump spectrum characteristics, and the robustness parameter is generated by combining the ambulance bump index and the influence coefficient.

[0016] Optionally, when the semantic interruption frequency in the directional voice semantic feature is higher than the preset frequency threshold, the tissue perfusion elasticity coefficient is non-linearly superimposed with the number of semantic logical break points in the directional voice semantic feature to generate a neurological function damage contradiction index, and a neurological function damage probability distribution is generated within a preset probability interval according to the neurological function damage contradiction index, including:

[0017] The directional voice semantic feature is segmented by a time series window to extract the voice fundamental frequency and the semantic logical connection strength in each window. A dynamic baseline is generated based on the historical data of the voice fundamental frequency. When the variance of the voice fundamental frequency in three consecutive windows exceeds the dynamic baseline and the semantic logical connection strength is lower than the preset connection strength threshold, the end point of the third window is marked as a semantic logical break point. The number of occurrences of the semantic logical break point within a unit time is counted, and the semantic interruption frequency is calculated in combination with the jump distribution density of the incoherent phrases. The incoherent phrases are phrases that are logically incoherent or have a jump strength greater than the preset jump strength threshold in the semantic stream extracted from the directional voice semantic feature;

[0018] It is judged whether the semantic interruption frequency is higher than the preset frequency threshold. If so, the tissue perfusion elasticity coefficient is normalized, and a piecewise exponential weight function is used to perform a coupling operation on the normalized tissue perfusion elasticity coefficient and the semantic interruption frequency to obtain a neurological function damage contradiction index;

[0019] Based on the neurological function damage contradiction index, in combination with the preset probability density response surface, by solving the curvature change rate of the corresponding contour line of the neurological function damage contradiction index on the probability density response surface, the bandwidth parameter of the probability density function corresponding to the neurological function damage contradiction index is dynamically adjusted;

[0020] Based on the adjusted bandwidth parameter, Gaussian kernel density estimation is performed on the neurological function damage contradiction index within a preset probability interval to obtain the slope of the cumulative distribution function corresponding to each sub-segment in the neurological function damage contradiction index within the preset probability interval. The probability weights of the sub-segments are re-allocated according to the position of the sign reversal point in the cumulative distribution function slope to generate a neurological function damage probability distribution.

[0021] Optionally, normalizing the tissue perfusion elasticity coefficient, and performing a coupling operation on the normalized tissue perfusion elasticity coefficient and the semantic interruption frequency using a piecewise exponential weight function to obtain a neurological function damage contradiction index, including:

[0022] Normalizing the tissue perfusion elasticity coefficient based on the baseline stability parameter of the microcirculation dynamic waveform, where the baseline stability parameter is obtained by calculating the variance and phase drift amount of the microcirculation dynamic waveform in the state without pressure stimulation;

[0023] According to the real-time distribution characteristics of the semantic interruption frequency, dividing the semantic interruption frequency into a high-frequency section, a medium-frequency section, and a low-frequency section;

[0024] In each section, calculate the weighted tissue perfusion elasticity coefficient according to the exponential weight function corresponding to the section and the normalized tissue perfusion elasticity coefficient, and use the product result of the weighted tissue perfusion elasticity coefficient and the semantic interruption frequency of the corresponding section as the coupling result of the section;

[0025] Take the sum result of the coupling results of all sections as the neurological function damage contradiction index.

[0026] Optionally, based on the neurological function damage contradiction index, combining with a preset probability density response surface, by solving the curvature change rate of the contour line corresponding to the neurological function damage contradiction index on the probability density response surface, dynamically adjusting the bandwidth parameter of the probability density function corresponding to the neurological function damage contradiction index, including:

[0027] Based on the neurological function damage contradiction index, construct a three-dimensional grid mapping relationship, perform interpolation smoothing processing on the three-dimensional grid mapping relationship to generate a probability density response surface, where the first dimension is the contradiction index, the second dimension is the diagnosis result grading, and the third dimension is the probability density value;

[0028] On the probability density response surface, locate the coordinate point corresponding to the neurological function damage contradiction index, and extract the local curvature parameter of the contour line to which the coordinate point belongs;

[0029] According to the partial derivative direction of the local curvature parameter in the tangent plane corresponding to the probability density response surface, obtain the direction symbol by analyzing the change trend of the partial derivative direction, calculate the curvature change rate of the contour line, and dynamically adjust the bandwidth parameter of the probability density function based on the absolute value of the curvature change rate and the direction symbol.

[0030] Optionally, extract the attenuation rate of the microcirculation waveform from the microcirculation dynamic waveform, and parse the spatial diffusion direction parameter from the trauma diffusion trend vector of the composite feature stream. Dynamically divide the hemostatic pressure level according to the attenuation rate of the microcirculation waveform and the spatial diffusion direction parameter, and generate a hemostatic intervention intensity spectrum by combining the hemostatic pressure level division result and the current body position stability of the patient, including:

[0031] Perform time-frequency decomposition on the microcirculation dynamic waveform, and extract the energy attenuation slope of the microcirculation dynamic waveform within a preset frequency band as the attenuation rate of the microcirculation waveform;

[0032] Parse the spatial gradient distribution of the blood flow diffusion path from the trauma diffusion trend vector of the composite feature stream. By performing eigenvalue decomposition on the spatial gradient distribution, extract the main change direction and the secondary change direction of the spatial gradient distribution, generate a spatial diffusion direction parameter based on the cosine value of the angle between the main change direction and the secondary change direction, and divide the hemostatic pressure level according to the joint distribution characteristics of the attenuation rate of the microcirculation waveform and the spatial diffusion direction parameter;

[0033] Combine the hemostatic pressure level division result, and collect the inertial sensing data of the patient's body position and posture in real time. Extract the trunk inclination variance and the limb displacement covariance matrix from the inertial sensing data, calculate the sum of the elements on the main diagonal between the trunk inclination variance and the limb displacement covariance matrix, generate the current body position stability of the patient, and construct a hemostatic intervention intensity spectrum based on the product coupling relationship between the hemostatic pressure level division result and the body position stability.

[0034] Optionally, when the consciousness state grading index of the composite feature stream is lower than a preset index threshold, calculate the tissue perfusion elasticity coefficient through the phase mutation feature of the microcirculation dynamic waveform, including:

[0035] Monitor the consciousness state grading index. When the consciousness state grading index is lower than the preset index threshold, trigger the grading pressure control protocol of the variable stiffness actuator of the medical robot to drive the variable stiffness actuator of the medical robot to apply a stepped pressure stimulus through the grading pressure control protocol. The grading pressure control protocol includes the pressure amplitude of the increasing step sequence and the action time window of each step. Construct a pressure stimulus energy spectrum according to the product relationship between the pressure amplitude of the increasing step sequence and the action time window;

[0036] Extract the phase angle change amount and the waveform distortion degree within a preset frequency band from the microcirculation dynamic waveform to generate a phase mutation feature, and calculate the tissue perfusion elasticity coefficient based on the coupling response between the phase mutation feature and the pressure stimulus energy spectrum.

[0037] Optionally, based on the hemostasis intervention intensity spectrum, monitor the change characteristics of the harmonic components of the microcirculation waveform. Based on the probability distribution of nerve function damage, feedback the robustness parameter to the motion planner of the robotic arm to generate an anti-interference control loop. Based on the change characteristics of the harmonic components and the anti-interference control loop, generate a closed-loop first aid treatment plan, including:

[0038] According to the time-varying mapping table of the spatial coordinate sequence and pressure amplitude at the pressurization position in the hemostasis intervention intensity spectrum, drive the robotic arm of the medical robot to perform an adaptive pressurization action to collect the energy distribution ratio and phase synchronization parameter of the microcirculation dynamic waveform within a preset frequency band, and generate the change characteristics of the harmonic components;

[0039] When the proportion of the high-risk sub-segment in the probability distribution of nerve function damage exceeds a preset ratio, perform hierarchical compression coding on the multi-modal sensing data stream of the collected medical robot to generate a compressed data packet, trigger the compressed transmission protocol of the remote consultation data packet, and send the compressed data packet to the motion planner of the robotic arm of the medical robot based on the compressed transmission protocol. Combine the preset gain adaptive adjustment rule to output an anti-interference control loop, and the robustness parameter is carried in the compressed data packet;

[0040] Synchronize the timestamps of the change characteristics of the harmonic components, the transmission status in the compressed data packet, and the robotic arm motion path correction vector in the anti-interference control loop to construct a multi-modal feedback matrix. Based on the priority scheduling strategy of the feature weights in the multi-modal feedback matrix, generate a closed-loop first aid treatment plan.

[0041] In a second aspect, an embodiment of the present application provides a computing device, including a processor and a memory. A computer program is stored in the memory, and the processor is configured to run the computer program to execute a multi-mode-based emergency treatment process automation method according to any one of the first aspects.

[0042] In a third aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, a multi-mode-based emergency treatment process automation method according to any one of the first aspects is implemented.

[0043] In an embodiment of the present application, a method for automated processing of an emergency treatment process based on multiple modes is provided. The method includes: at the out-of-hospital emergency scene, collecting the dynamic microcirculation waveform of the patient's skin through the medical robot, and extracting the directional speech semantic features under environmental noise interference; comparing the dynamic microcirculation waveform and the directional speech semantic features in the spatio-temporal dimension to construct a three-dimensional spatio-temporal compensation field, and generating a composite feature stream based on the three-dimensional spatio-temporal compensation field; when the consciousness state grading index of the composite feature stream is lower than a preset index threshold, calculating the tissue perfusion elasticity coefficient through the phase mutation feature of the dynamic microcirculation waveform; constructing a three-layer dynamic decision tree topology network based on the composite feature stream and the tissue perfusion elasticity coefficient, where the first layer generates a hemostasis intervention intensity spectrum based on the dynamic microcirculation waveform and the composite feature stream, the second layer generates a probability distribution of nerve function damage based on the tissue perfusion elasticity coefficient and the directional speech semantic features, and the third layer corrects the robustness parameters of the manipulator operation path by combining the real-time collected environmental light intensity and the ambulance bump index; monitoring the change characteristics of the harmonic components of the microcirculation waveform based on the hemostasis intervention intensity spectrum, and feeding back the robustness parameters to the motion planner of the manipulator based on the probability distribution of nerve function damage to generate an anti-interference control loop, and generating a closed-loop emergency treatment plan based on the change characteristics of the harmonic components and the anti-interference control loop.

[0044] The technical solution of the present application has the following beneficial effects:

[0045] The present application obtains the patient's physiological state and voice information under environmental interference in real time, providing a multi-modal data basis for subsequent analysis. By comparing in the spatio-temporal dimension, environmental noise interference is eliminated, a more accurate composite feature stream is generated, and the reliability of data analysis is improved. Through phase mutation feature analysis, the tissue perfusion state is quantified, providing a key physiological index for emergency decision-making. The first layer generates a hemostasis intervention intensity spectrum to achieve accurate hemostasis intervention, the second layer generates a probability distribution of nerve function damage to evaluate the risk of nerve damage, and the third layer corrects the robustness parameters of the manipulator operation path to improve the operation stability of the manipulator in a complex environment. Through the change characteristics of the harmonic components and the anti-interference control loop, dynamic adjustment and optimization of emergency operations are realized, and the success rate of emergency treatment is improved.

[0046] Furthermore, in the dynamic decision tree topology network of the embodiment of the present application, in the first layer, the hemostasis pressure level is dynamically divided through the microcirculation waveform attenuation rate and the trauma diffusion trend vector, and a hemostasis intervention intensity spectrum is generated in combination with the patient's body position stability; in the second layer, a nerve function damage contradiction index is generated through the semantic interruption frequency and the tissue perfusion elasticity coefficient, and further a probability distribution of nerve function damage is generated; in the third layer, the robustness parameters are calculated through the bump spectrum characteristics and the environmental light intensity influence coefficient and used to correct the manipulator operation path.

[0047] Through a multi-level dynamic decision tree topology network, precise grading of hemostasis intervention, quantitative assessment of the risk of nerve function injury, and robust optimization of the operating path of the robotic arm are achieved, improving the accuracy, adaptability, and success rate of first aid operations.

[0048] These aspects or other aspects of the present application will be more clearly understood in the description of the following embodiments. Brief Description of the Drawings

[0049] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0050] Figure 1 It is a flowchart of a multi-mode-based automated processing method for emergency treatment processes provided by an embodiment of the present application;

[0051] Figure 2 It is a schematic structural diagram of a computing device provided by an embodiment of the present application. Detailed Embodiments

[0052] To enable those skilled in the art of the present technology to better understand the solutions of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application.

[0053] In some processes described in the specification, claims, and the above-mentioned drawings of the present application, multiple operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. Additionally, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0055] By using a medical robot to collect the dynamic microcirculation waveforms of the patient's skin and the directional speech semantic features under environmental noise interference in the out-of-hospital first-aid scene, a three-dimensional spatio-temporal compensation field is constructed by using spatio-temporal dimension comparison, and a composite feature stream is generated to eliminate noise interference; when the consciousness state grading index of the composite feature stream is lower than a preset threshold, the tissue perfusion elasticity coefficient is calculated through the phase mutation characteristics of the microcirculation waveform, and a three-layer dynamic decision tree topology network is constructed by combining the composite feature stream to respectively generate the hemostasis intervention intensity spectrum, the probability distribution of nerve function damage, and the robustness parameters of the robotic arm operation path; finally, by monitoring the harmonic component change characteristics of the microcirculation waveform and generating an anti-interference control loop, a closed-loop first-aid treatment plan is formed to achieve precise, adaptive and anti-interference automated first-aid treatment.

[0056] Figure 1 The flowchart of a multi-mode based automated emergency treatment process method provided by an embodiment of this application is as Figure 1 shown, and the method includes:

[0057] Step 101: At the out-of-hospital first-aid scene, collect the dynamic microcirculation waveforms of the patient's skin through the medical robot, and extract the directional speech semantic features under environmental noise interference.

[0058] In this step, the dynamic microcirculation waveform is the dynamic change data of the patient's skin microvascular blood flow collected by the medical robot, which reflects the tissue perfusion state, including the change of the patient's blood flow, blood flow velocity, vasomotor state, etc. The directional speech semantic feature is the semantic information in the patient's speech extracted through speech recognition technology under environmental noise interference, and is used to evaluate the patient's consciousness state and nerve function.

[0059] In actual operation, the flexible epidermal contact sensing unit of the medical robot collects the dynamic microcirculation waveforms of the patient's skin through an optical sensor (such as a laser Doppler flowmeter) to obtain parameters such as blood flow velocity and vascular elasticity. At the same time, the robot collects the patient's speech through a directional microphone array, uses a speech enhancement algorithm (such as beamforming) to eliminate environmental noise interference, and extracts semantic features (such as semantic coherence, keyword frequency) through natural language processing technology. Finally, the dynamic microcirculation waveform and the directional speech semantic feature are used as multi-modal data inputs to provide a basis for subsequent analysis.

[0060] For example, at the first-aid scene, the medical robot detects that the skin microcirculation waveform of a car accident patient shows an abnormal decrease in blood flow velocity, and at the same time, through speech recognition, it is found that the patient's semantic expression is frequently interrupted, and it is initially judged that the patient may have hemorrhagic shock and consciousness disorder.

[0061] Step 102: Compare the microcirculation dynamic waveform and the directional speech semantic features in the spatio-temporal dimension to construct a three-dimensional spatio-temporal compensation field. Based on the three-dimensional spatio-temporal compensation field, generate a composite feature stream.

[0062] In this step, the three-dimensional spatio-temporal compensation field is used to eliminate environmental interference and enhance data reliability. The composite feature stream contains the fusion information of the microcirculation dynamic waveform and the directional speech semantic features.

[0063] In actual operation, align the microcirculation dynamic waveform and the directional speech semantic features in the spatio-temporal dimension, and use time series analysis algorithms (such as dynamic time warping) and spatial matching techniques (such as feature point mapping) to construct a three-dimensional spatio-temporal compensation field. In the compensation field, eliminate noise interference through data fusion algorithms (such as Kalman filtering) to generate a composite feature stream, which contains the time variation trend of the microcirculation waveform and the spatial distribution characteristics of the speech semantics.

[0064] For example, a medical robot aligns the microcirculation waveform of a patient with the speech semantic features, and finds that the microcirculation waveform shows obvious fluctuations during speech interruption. Eliminate the ambulance engine noise interference through the three-dimensional spatio-temporal compensation field to generate a composite feature stream, providing reliable data for subsequent analysis.

[0065] Step 103: When the consciousness state grading index of the composite feature stream is lower than the preset index threshold, calculate the tissue perfusion elasticity coefficient through the phase mutation characteristics of the microcirculation dynamic waveform.

[0066] In this step, the consciousness state grading index is a quantitative index generated based on the composite feature stream, used to evaluate the patient's consciousness state. The tissue perfusion elasticity coefficient is used to reflect the elasticity and recovery ability of tissue perfusion. The phase mutation characteristics refer to the phase angle change points in the microcirculation dynamic waveform due to sudden changes in blood flow status (such as vasoconstriction or dilation). These mutation points reflect the dynamic changes of the tissue perfusion state and can be used to evaluate vascular elasticity and tissue recovery ability.

[0067] In actual operation, extract features from the composite feature stream, and use a machine learning model (such as a support vector machine) to calculate the consciousness state grading index. When the index is lower than the preset threshold, drive the variable stiffness actuator of the medical robot to apply a stepped pressure stimulus, and calculate the tissue perfusion elasticity coefficient through the phase mutation characteristics of the microcirculation waveform (such as the mutation points extracted by wavelet transform) to quantify the tissue perfusion state.

[0068] For example, the composite feature stream shows that the consciousness state grading index of the patient is 3 (lower than the threshold of 5). The medical robot calculates the tissue perfusion elasticity coefficient to be 0.6 by analyzing the phase mutation characteristics of the microcirculation waveform, indicating that the patient's tissue perfusion ability is poor.

[0069] Step 104: Based on the composite feature stream and the tissue perfusion elasticity coefficient, construct a three-layer dynamic decision tree topology network. The first layer generates a hemostasis intervention intensity spectrum based on the microcirculation dynamic waveform and the composite feature stream. The second layer generates a probability distribution of neurological function damage according to the tissue perfusion elasticity coefficient and the directional speech semantic features. The third layer combines the ambient light intensity collected in real time and the ambulance bump index to correct the robustness parameters of the robotic arm operation path.

[0070] In this step, the hemostasis intervention intensity spectrum includes different pressure levels and intervention times. The probability distribution of neurological function damage is used to predict the possibility and severity of the patient's neurological function damage, including the neurological function damage contradiction index and its distribution in different probability intervals. The robustness parameter is a quantitative index used to describe the stability and adaptability of the robotic arm operation path in a complex environment (such as ambulance bumps and light changes), which reflects the anti-interference ability and operation accuracy of the robotic arm under external interference.

[0071] In actual operation, the first layer: Extract the attenuation rate from the microcirculation waveform, analyze the trauma diffusion trend vector from the composite feature stream, and generate a hemostasis intervention intensity spectrum in combination with the patient's body position stability. The second layer: When the semantic interruption frequency in the speech semantic features is higher than the threshold, non-linearly superimpose the tissue perfusion elasticity coefficient and the number of semantic logical breakpoints to generate a probability distribution of neurological function damage. The third layer: Collect the ambulance bump frequency spectrum features through a vibration sensor, and combine the influence coefficient of the ambient light intensity on the visual capture sensitivity to generate the robustness parameter.

[0072] For example, the medical robot generates a hemostasis intervention intensity spectrum based on the microcirculation waveform attenuation rate and the trauma diffusion trend vector, calculates the probability of neurological function damage as 45% in combination with the speech semantic features, and corrects the robotic arm operation path through the ambulance bump index and the ambient light intensity to ensure operation stability.

[0073] Step 105: Based on the hemostasis intervention intensity spectrum, monitor the change characteristics of the harmonic components of the microcirculation waveform. Based on the probability distribution of neurological function damage, feedback the robustness parameter to the motion planner of the robotic arm to generate an anti-interference control loop. Based on the change characteristics of the harmonic components and the anti-interference control loop, generate a closed-loop first aid treatment plan.

[0074] In this step, the change characteristics of the harmonic components refer to the characteristics in the microcirculation waveform that reflect the periodic changes of blood flow and are used to evaluate the hemostasis intervention effect. The anti-interference control loop is used to resist environmental interference.

[0075] In actual operation, based on the hemostasis intervention intensity spectrum, the harmonic component change characteristics of the microcirculation waveform are extracted through Fourier transform to evaluate the hemostasis effect. The probability distribution of nerve function injury and the robustness parameter are fed back to the robotic arm motion planner, and an anti-interference control loop is generated using a control algorithm. Finally, combining the harmonic component change characteristics and the anti-interference control loop, a closed-loop first aid treatment plan is generated.

[0076] For example, when the medical robot monitors that the harmonic components of the microcirculation waveform tend to be stable, it indicates that the hemostasis intervention is effective. At the same time, the operation path of the robotic arm is adjusted through the anti-interference control loop, and the first aid treatment is successfully completed.

[0077] Through multi-modal data acquisition, three-dimensional spatio-temporal compensation field construction, dynamic decision tree network generation, and closed-loop first aid treatment plan optimization, accurate, adaptive, and anti-interference automated first aid treatment in out-of-hospital first aid scenarios is achieved, improving the success rate of first aid and the survival rate of patients.

[0078] To solve the complexity problems of multi-modal data fusion and dynamic decision-making in out-of-hospital first aid scenarios and further improve the accuracy and adaptability of first aid operations, in the first layer of the dynamic decision tree topology network, the attenuation rate of the microcirculation waveform is extracted from the microcirculation dynamic waveform, and the spatial diffusion direction parameter is parsed from the trauma diffusion trend vector of the composite feature stream. Combining the patient's body position stability, the hemostasis pressure level is dynamically divided to generate a hemostasis intervention intensity spectrum; in the second layer, when the semantic interruption frequency in the directional speech semantic features is higher than the preset threshold, the tissue perfusion elasticity coefficient and the number of semantic logic breakpoints are non-linearly superimposed to generate a nerve function injury contradiction index, and based on this index, the probability distribution of nerve function injury is generated within the preset probability interval; in the third layer, the bump spectrum characteristics are collected in real time through the ambulance vibration sensor, and combined with the influence coefficient of the environmental light intensity on the visual capture sensitivity of the medical robot, the ambulance bump index is calculated, and a robustness parameter is generated to optimize the stability and accuracy of the robotic arm operation. Overall, through a multi-layer dynamic decision tree network, the collaborative processing of hemostasis intervention, nerve function injury assessment, and robotic arm operation optimization is realized, improving the first aid efficiency and success rate.

[0079] In some embodiments, in step 104, the first layer generates a hemostasis intervention intensity spectrum based on the microcirculation dynamic waveform and the composite feature stream, the second layer generates a probability distribution of nerve function injury according to the tissue perfusion elasticity coefficient and the directional speech semantic features, and the third layer corrects the robustness parameter of the robotic arm operation path by combining the environmental light intensity collected in real time and the ambulance bump index, including:

[0080] Step 201: In the first layer of the dynamic decision tree topology network, extract the attenuation rate of the microcirculation waveform from the microcirculation dynamic waveform, and parse the spatial diffusion direction parameter from the trauma diffusion trend vector of the composite feature stream. Dynamically divide the hemostatic pressure level according to the attenuation rate of the microcirculation waveform and the spatial diffusion direction parameter, and generate a hemostatic intervention intensity spectrum in combination with the current body position stability of the patient.

[0081] In step 201, the attenuation rate of the microcirculation waveform refers to the speed at which the blood flow velocity in the microcirculation dynamic waveform decays over time, usually calculated by analyzing the slope of the descending segment of the waveform, used to evaluate the dynamic changes in tissue perfusion status, reflect the degree of vasoconstriction or dilation, and provide a basis for hemostatic intervention. The trauma diffusion trend vector is vector data parsed from the composite feature stream that reflects the direction and magnitude of trauma diffusion in tissues, used to judge the diffusion direction and range of trauma, and provide a spatial reference for hemostatic intervention. The spatial diffusion direction parameter is a quantitative parameter extracted from the trauma diffusion trend vector that reflects the trauma diffusion direction, used to dynamically divide the hemostatic pressure level and determine the key areas of hemostatic intervention. The hemostatic pressure levels include three levels: low, medium, and high, and the division basis is the magnitude of the attenuation rate (such as low: <0.5% / s, medium: 0.5 - 1.0% / s, high: >1.0% / s) and the severity of the diffusion direction parameter (such as the size of the diffusion range). The current body position stability refers to the stability of the patient's body position, calculated from the body position change data collected by an inertial sensor (such as an accelerometer), including the amplitude of body position change (such as angle change) and frequency (such as the number of changes per minute).

[0082] In the embodiment of the present application, the attenuation rate is extracted from the microcirculation dynamic waveform through signal processing techniques (such as wavelet transform) to reflect the dynamic changes in the blood flow state. The trauma diffusion trend vector is parsed from the composite feature stream, and the spatial analysis algorithm (such as gradient calculation) is used to determine the diffusion direction parameter. In combination with the patient's body position stability (collected by an inertial sensor), the dynamic programming algorithm is used to divide the hemostatic pressure level and generate a hemostatic intervention intensity spectrum.

[0083] Step 202: In the second layer of the dynamic decision tree topology network, when the semantic interruption frequency in the directional speech semantic feature is higher than the preset frequency threshold, non-linearly superimpose the tissue perfusion elasticity coefficient and the number of semantic logical breakpoints in the directional speech semantic feature to generate a neurological function injury contradiction index, and generate a neurological function injury probability distribution within a preset probability interval according to the neurological function injury contradiction index.

[0084] In step 202, the semantic interruption frequency refers to the number of times of semantic expression interruption in the patient's speech, usually expressed as the number of interruptions per unit time (such as times / minute), and is used to evaluate the patient's consciousness state and the possibility of neurological function damage. The semantic logic break point refers to the specific position where the semantic logic in the patient's speech is incoherent or interrupted, usually identified by natural language processing techniques (such as semantic analysis), and is used to quantify the abnormality degree of the speech semantic features, providing a basis for the evaluation of neurological function damage. The neurological function damage contradiction index is used to generate the probability distribution of neurological function damage and evaluate the risk of nerve damage.

[0085] In the embodiment of the present application, when the semantic interruption frequency in the directional speech semantic features is higher than the preset threshold, the tissue perfusion elasticity coefficient is combined with the number of semantic logic break points by using a non-linear superposition algorithm (such as weighted summation or fuzzy logic) to generate a neurological function damage contradiction index. According to the preset probability interval (such as 0 - 100%), a probability distribution model (such as Gaussian distribution) is used to generate the probability distribution of neurological function damage and evaluate the risk of nerve damage.

[0086] Step 203: In the third layer of the dynamic decision tree topology network, the bump spectrum characteristics are collected in real time through the vibration sensor installed on the ambulance, the influence coefficient of the ambient light intensity collected in real time on the visual capture sensitivity of the medical robot is determined, the ambulance bump index is calculated based on the bump spectrum characteristics, and a robustness parameter is generated by combining the ambulance bump index and the influence coefficient.

[0087] In step 203, the bump spectrum characteristics are spectrum data reflecting the vehicle bump state collected in real time through the ambulance vibration sensor. The ambulance bump index is used to reflect the vehicle bump intensity. The influence coefficient of sensitivity refers to the degree of influence of the ambient light intensity on the visual capture sensitivity of the medical robot, usually calculated from the light intensity data collected by the light sensor, and is used to correct the robustness parameter of the manipulator operation path to ensure the operation stability of the manipulator in a complex light environment.

[0088] In the embodiment of the present application, the bump spectrum characteristics are collected through the vibration sensor, the main frequency components are extracted by using Fourier transform, and the ambulance bump index is calculated. Combining the influence coefficient of the ambient light intensity on the visual capture sensitivity of the medical robot (collected by the light sensor), a robustness parameter is generated by using a weighted fusion algorithm for correcting the manipulator operation path.

[0089] The following is a specific example:

[0090] In an out-of-hospital emergency scenario, a medical robot detects that the attenuation rate of the microcirculation waveform of a car accident patient is 0.8 (unit: % / s), the trauma diffusion trend vector shows that the trauma spreads to the right, and the patient's body position stability is 70%. Based on these data, the first level of the dynamic decision tree topology network generates a hemostasis intervention intensity spectrum, recommends an intermediate hemostasis pressure level, and an intervention time of 5 minutes. At the same time, the patient's directional speech semantic features show that the semantic interruption frequency is 12 times per minute (higher than the preset threshold of 10 times per minute), the tissue perfusion elasticity coefficient is 0.6, the number of semantic logic breakpoints is 5, and the neurological function damage contradiction index is generated as 8.2 through non-linear superposition. The neurological function damage probability distribution shows that the damage probability is 45%. In the third level, the ambulance bump index is 3.5 (unit: m / s 2 ), the environmental light intensity influence coefficient is 0.8, and the generated robustness parameter is 2.8, which is used to correct the robotic arm operation path to ensure operation stability. Finally, the medical robot successfully completes the first aid treatment according to the hemostasis intervention intensity spectrum and the robustness parameter.

[0091] Through the multi-level collaborative processing of the dynamic decision tree topology network, the precise grading of hemostasis intervention, the quantitative assessment of the risk of neurological function damage, and the robustness optimization of the robotic arm operation path are realized, improving the operation accuracy, adaptability, and success rate in out-of-hospital emergency scenarios.

[0092] To solve the problem of the accuracy of the risk assessment of neurological function damage in out-of-hospital emergency scenarios and further improve the reliability and dynamic adaptability of the assessment results, the directional speech semantic features are segmented by time series windows, the voice fundamental frequency and the semantic logic connection strength within each window are extracted, a dynamic baseline is generated based on the historical data of the voice fundamental frequency. When the variance of the voice fundamental frequency in three consecutive windows exceeds the dynamic baseline and the semantic logic connection strength is lower than the preset threshold, semantic logic breakpoints are marked, the number of breakpoints appearing within a unit time is counted, and the semantic interruption frequency is calculated in combination with the jump distribution density of non-coherent phrases; if the semantic interruption frequency is higher than the preset threshold, the tissue perfusion elasticity coefficient is normalized, and a piecewise exponential weight function is used to couple and operate it with the semantic interruption frequency to generate a neurological function damage contradiction index; based on this index, combined with the probability density response surface, the bandwidth parameter of the probability density function is dynamically adjusted by solving the curvature change rate, and the probability weights are reallocated using Gaussian kernel density estimation and the slope of the cumulative distribution function, and finally the neurological function damage probability distribution is generated to realize the precise assessment of the risk of neurological function damage.

[0093] In some embodiments, when the semantic interruption frequency in the directional speech semantic features is higher than a preset frequency threshold in step 202, the tissue perfusion elastic coefficient is non-linearly superimposed with the number of semantic logical breakpoints in the directional speech semantic features to generate a neurological function damage contradiction index, and a neurological function damage probability distribution is generated within a preset probability interval, including:

[0094] Step 301: Perform temporal window segmentation on the directional speech semantic features to extract the speech fundamental frequency and semantic logical connection strength within each window. Generate a dynamic baseline based on the historical data of the speech fundamental frequency. When the variance of the speech fundamental frequency in three consecutive windows exceeds the dynamic baseline and the semantic logical connection strength is lower than the preset connection strength threshold, mark the end point of the third window as a semantic logical breakpoint. Count the number of occurrences of the semantic logical breakpoints within a unit time, and calculate the semantic interruption frequency in combination with the jump distribution density of incoherent phrases. The incoherent phrases are phrases that are logically incoherent or have a jump strength greater than the preset jump strength threshold in the semantic stream extracted from the directional speech semantic features.

[0095] In step 301, the temporal window segmentation divides the directional speech semantic features into multiple windows according to time, and each window contains speech data of a certain time length (such as 1 second). The speech fundamental frequency represents a parameter in the speech signal that reflects the vocal cord vibration frequency and is used to evaluate the stability and continuity of the speech. The semantic logical connection strength is a parameter used to reflect the semantic logical coherence of the speech and is used to evaluate the integrity of semantic expression. The dynamic baseline is used to determine whether the current speech fundamental frequency is abnormal.

[0096] In the embodiments of the present application, temporal window segmentation is performed on the directional speech semantic features to extract the speech fundamental frequency and semantic logical connection strength within each window. A dynamic baseline is generated based on the historical data of the speech fundamental frequency, and the variance analysis algorithm is used to determine whether the speech fundamental frequency of the current window is abnormal. When the variance of the speech fundamental frequency in three consecutive windows exceeds the dynamic baseline and the semantic logical connection strength is lower than the preset threshold, mark the semantic logical breakpoint. Count the number of occurrences of the semantic logical breakpoints within a unit time, and calculate the semantic interruption frequency in combination with the jump distribution density of incoherent phrases.

[0097] Step 302: Determine whether the semantic interruption frequency is higher than the preset frequency threshold. If so, perform normalization processing on the tissue perfusion elastic coefficient, and perform a coupling operation on the normalized tissue perfusion elastic coefficient and the semantic interruption frequency using a piecewise exponential weight function to obtain a neurological function damage contradiction index.

[0098] In step 302, the normalization processing converts the tissue perfusion elastic coefficient into a standardized value within the range of 0-1, which is convenient for subsequent calculations.

[0099] In the embodiment of the present application, it is determined whether the semantic interruption frequency is higher than a preset threshold. If it is higher than the threshold, the tissue perfusion elasticity coefficient is normalized. A piecewise exponential weight function is used to perform a coupling operation on the normalized tissue perfusion elasticity coefficient and the semantic interruption frequency to generate a neurological function injury contradiction index.

[0100] Step 303: Based on the neurological function injury contradiction index, in combination with a preset probability density response surface, by solving the curvature change rate of the contour line corresponding to the neurological function injury contradiction index on the probability density response surface, dynamically adjust the bandwidth parameter of the probability density function corresponding to the neurological function injury contradiction index.

[0101] In step 303, the bandwidth parameter refers to the parameter used to control the smoothness of the probability density function in Gaussian kernel density estimation. The probability density response surface is a three-dimensional surface that reflects the probability density distribution corresponding to different contradiction index values. The abscissa represents the value range of the neurological function injury contradiction index (such as 0 - 10). The ordinate represents the value range of the probability density (such as 0 - 1). The surface height represents the probability density value corresponding to a specific contradiction index value. The probability density function is used to describe the probability distribution of the neurological function injury contradiction index within a preset probability interval, indicating the probability density near each value point. The curvature change rate is extracted by analyzing the contour line of the probability density response surface through mathematical methods (such as second derivative calculation) and is used to dynamically adjust the bandwidth parameter to ensure that the probability density function can accurately reflect the distribution characteristics of the neurological function injury contradiction index.

[0102] In the embodiment of the present application, based on the neurological function injury contradiction index, in combination with a preset probability density response surface, the bandwidth parameter is dynamically adjusted by solving the curvature change rate.

[0103] Step 304: Based on the adjusted bandwidth parameter, perform Gaussian kernel density estimation on the neurological function injury contradiction index within a preset probability interval to obtain the slope of the cumulative distribution function corresponding to each sub-segment in the neurological function injury contradiction index within the preset probability interval. Reallocate the probability weights of each sub-segment according to the position of the sign reversal point in the slope of the cumulative distribution function to generate a neurological function injury probability distribution.

[0104] In step 304, the slope of the cumulative distribution function is used to describe the cumulative probability of the neurological function injury contradiction index from the minimum value to a specific value, which is the integral of the probability density function. The preset probability interval refers to a predefined range used to limit the value range of the neurological function injury contradiction index. This interval is usually determined based on historical case data or clinical experience, such as [0,1] or [-1,1], and the specific range depends on the distribution characteristics of the normalized neurological function injury contradiction index. The sign reversal point refers to the key point where the slope of the cumulative distribution function changes from positive to negative or from negative to positive, reflecting the turning position of the change trend of the probability density function.

[0105] In the embodiment of the present application, based on the adjusted bandwidth parameter, Gaussian kernel density estimation is performed on the neurological function injury contradiction index to obtain the slope of the cumulative distribution function. According to the position of the sign reversal point in the slope, the probability weights of each sub-segment are reallocated to generate the probability distribution of neurological function injury.

[0106] The following is a specific example:

[0107] In the out-of-hospital first aid scenario, the medical robot performs time-series window segmentation on the directional voice semantic features of the patient, and extracts the voice fundamental frequency and semantic logical connection strength of each window. After generating a dynamic baseline based on historical data, it is found that the variance of the voice fundamental frequency in three consecutive windows exceeds the baseline and the semantic logical connection strength is lower than the threshold. The semantic logical break point is marked, and the number of break points occurring per unit time is counted as 8 times. Combining with the jump distribution density of non-coherent phrases, the semantic interruption frequency is calculated as 12 times per minute. Since the semantic interruption frequency is higher than the preset threshold (10 times per minute), the tissue perfusion elasticity coefficient (0.6) is normalized, and a piecewise exponential weight function is used to perform a coupling operation with the semantic interruption frequency to generate a neurological function injury contradiction index of 7.5. Based on this index, combined with the probability density response surface, the bandwidth parameter is dynamically adjusted, and the probability distribution of neurological function injury is generated through Gaussian kernel density estimation, showing an injury probability of 50%.

[0108] Through time-series window segmentation, dynamic baseline generation, semantic logical break point marking, piecewise exponential weight function coupling operation, and Gaussian kernel density estimation, the accurate quantitative assessment of the risk of neurological function injury is realized, the dynamic adaptability and reliability of the assessment results are improved, and a scientific basis is provided for first aid decision-making.

[0109] To solve the problem of accurately calculating the contradiction index of nerve function injury in out-of-hospital first-aid scenarios, based on the baseline stability parameters of the microcirculation dynamic waveform (obtained by calculating the variance and phase drift amount in the state of no pressure stimulation), the tissue perfusion elasticity coefficient is normalized; according to the real-time distribution characteristics of the semantic interruption frequency, it is divided into three sections: high frequency, medium frequency, and low frequency; within each section, the normalized tissue perfusion elasticity coefficient is weighted and calculated using the corresponding exponential weight function, and the weighted result is multiplied by the semantic interruption frequency of the corresponding section to obtain the coupling result of the section; finally, the coupling results of all sections are added together to generate the nerve function injury contradiction index, thereby realizing the accurate quantitative assessment of the risk of nerve function injury.

[0110] In some embodiments, in step 302, when normalizing the tissue perfusion elasticity coefficient, a piecewise exponential weight function is used to perform a coupling operation on the normalized tissue perfusion elasticity coefficient and the semantic interruption frequency to obtain the nerve function injury contradiction index, including:

[0111] Step 401: Based on the baseline stability parameters of the microcirculation dynamic waveform, normalize the tissue perfusion elasticity coefficient.

[0112] In step 401, the baseline stability parameter is a parameter obtained by calculating the variance and phase drift amount of the microcirculation dynamic waveform in the state of no pressure stimulation, reflecting the stability of the microcirculation waveform.

[0113] In the embodiments of the present application, the variance and phase drift amount of the microcirculation dynamic waveform in the state of no pressure stimulation are calculated through signal processing techniques (such as wavelet transform) to generate the baseline stability parameter. The tissue perfusion elasticity coefficient is normalized using the baseline stability parameter to obtain a standardized value.

[0114] Step 402: According to the real-time distribution characteristics of the semantic interruption frequency, divide the semantic interruption frequency into a high-frequency section, a medium-frequency section, and a low-frequency section.

[0115] In step 402, the real-time distribution characteristics are used to reflect the distribution of the semantic interruption frequency within different time windows.

[0116] In the embodiments of the present application, statistical analysis is performed on the semantic interruption frequency to determine its distribution characteristics (such as mean, variance). According to preset thresholds (such as high frequency: > 10 times / minute, medium frequency: 5 - 10 times / minute, low frequency: < 5 times / minute), the semantic interruption frequency is divided into three sections: high frequency, medium frequency, and low frequency.

[0117] Step 403: Within each section, calculate the weighted tissue perfusion elasticity coefficient according to the exponential weight function corresponding to the section and the normalized tissue perfusion elasticity coefficient, and use the product of the weighted tissue perfusion elasticity coefficient and the semantic interruption frequency of the corresponding section as the coupling result of the section.

[0118] In step 403, the coupling result of the section is used to reflect the contribution of the section to the neurological function injury contradiction index.

[0119] In the embodiment of the present application, within each section, the normalized tissue perfusion elasticity coefficient is weighted and calculated using the corresponding exponential weight function. Multiply the weighted result by the semantic interruption frequency of the corresponding section to obtain the coupling result of the section.

[0120] Step 404: Use the sum of the coupling results of all sections as the neurological function injury contradiction index.

[0121] In the embodiment of the present application, add the coupling results of the high-frequency, medium-frequency, and low-frequency sections to generate the neurological function injury contradiction index.

[0122] The following is a specific example:

[0123] In an out-of-hospital first aid scenario, the medical robot detects that the baseline stability parameter of the patient's microcirculation dynamic waveform is 0.8 (variance is 0.2, phase drift amount is 0.1), and normalizes the tissue perfusion elasticity coefficient (0.6) to obtain a standardized value of 0.75. At the same time, the real-time distribution characteristics of the semantic interruption frequency show that the high-frequency section (>10 times / minute) accounts for 30%, the medium-frequency section (5-10 times / minute) accounts for 50%, and the low-frequency section (<5 times / minute) accounts for 20%. Within the high-frequency section, the weighted tissue perfusion elasticity coefficient calculated using the exponential weight function is 0.9, and multiplying it by the semantic interruption frequency of the high-frequency section (12 times / minute) gives a coupling result of 10.8; within the medium-frequency section, the weighted tissue perfusion elasticity coefficient is 0.7, and multiplying it by the semantic interruption frequency of the medium-frequency section (8 times / minute) gives a coupling result of 5.6; within the low-frequency section, the weighted tissue perfusion elasticity coefficient is 0.5, and multiplying it by the semantic interruption frequency of the low-frequency section (3 times / minute) gives a coupling result of 1.5. Finally, add the coupling results of the three sections to generate a neurological function injury contradiction index of 17.9.

[0124] Through baseline stability parameter normalization, semantic interruption frequency section division, weighted tissue perfusion elasticity coefficient calculation, and addition of section coupling results, the accurate quantification of the neurological function injury contradiction index is achieved, improving the scientificity and practicality of the evaluation results and providing a reliable basis for first aid decision-making.

[0125] To solve the problem of the accuracy of generating the probability density response surface in the out-of-hospital first aid scenario and further improve the dynamic adaptability and reliability of adjusting the bandwidth parameter of the probability density function, based on the neurological function injury contradiction index, a three-dimensional grid mapping relationship is constructed (the first dimension is the contradiction index, the second dimension is the diagnosis result grading, and the third dimension is the probability density value), and the mapping relationship is interpolated and smoothed to generate a probability density response surface; the coordinate point corresponding to the contradiction index is located on the surface, and the local curvature parameter of the contour line to which it belongs is extracted; according to the partial derivative direction of the local curvature parameter in the tangent plane, the change trend of the direction symbol is analyzed and the curvature change rate of the contour line is calculated; based on the absolute value and direction symbol of the curvature change rate, the bandwidth parameter of the probability density function is dynamically adjusted, so as to optimize the accuracy and adaptability of the probability density estimation.

[0126] In some embodiments, in step 303, based on the neurological function injury contradiction index, combined with a preset probability density response surface, by solving the curvature change rate of the contour line corresponding to the neurological function injury contradiction index on the probability density response surface, the bandwidth parameter of the probability density function corresponding to the neurological function injury contradiction index is dynamically adjusted, including:

[0127] Step 501: Based on the neurological function injury contradiction index, construct a three-dimensional grid mapping relationship, and perform interpolation and smoothing processing on the three-dimensional grid mapping relationship to generate a probability density response surface, where the first dimension is the contradiction index, the second dimension is the diagnosis result grading, and the third dimension is the probability density value.

[0128] In step 501, the three-dimensional grid mapping relationship is a three-dimensional data structure based on the neurological function injury contradiction index, the diagnosis result grading, and the probability density value, and is used to describe the mathematical relationship among the three. The interpolation and smoothing processing means that the three-dimensional grid mapping relationship is smoothed through an interpolation algorithm (such as bicubic interpolation).

[0129] In the embodiment of the present application, a three-dimensional grid mapping relationship is constructed based on the neurological function injury contradiction index, the diagnosis result grading, and the probability density value. The mapping relationship is smoothed by using an interpolation algorithm to generate a continuous and smooth probability density response surface.

[0130] Step 502: On the probability density response surface, locate the coordinate point corresponding to the neurological function injury contradiction index, and extract the local curvature parameter of the contour line to which the coordinate point belongs.

[0131] In step 502, the coordinate points represent the points on the probability density response surface corresponding to the nerve function injury contradiction index, including three coordinates: the contradiction index, the diagnosis result grade, and the probability density value. The contour lines represent the set of points on the probability density response surface with equal probability density values, and are used to describe the local characteristics of the surface. The local curvature parameter is a parameter used to reflect the degree of bending of the contour lines near the coordinate points, and is used to analyze the local shape of the surface.

[0132] In the embodiment of the present application, the coordinate points corresponding to the nerve function injury contradiction index are located on the probability density response surface. The local curvature parameter of the contour line to which the coordinate points belong is extracted for subsequent curvature change rate analysis.

[0133] Step 503: According to the partial derivative direction of the local curvature parameter in the tangent plane corresponding to the probability density response surface, obtain a direction symbol by analyzing the change trend of the partial derivative direction, calculate the curvature change rate of the contour line, and dynamically adjust the bandwidth parameter of the probability density function based on the absolute value of the curvature change rate and the direction symbol.

[0134] In step 503, the partial derivative direction represents the change direction of the local curvature parameter in the tangent plane, and is used to analyze the curvature change trend of the contour line. The direction symbol represents the change trend of the partial derivative direction (such as positive and negative directions), and is used to determine the directionality of the curvature change.

[0135] In the embodiment of the present application, according to the partial derivative direction of the local curvature parameter in the tangent plane, analyze the change trend of the direction symbol. Calculate the curvature change rate of the contour line, and dynamically adjust the bandwidth parameter based on its absolute value and the direction symbol.

[0136] The following is a specific example:

[0137] In the out-of-hospital first aid scenario, the medical robot constructs a three-dimensional grid mapping relationship based on the nerve function injury contradiction index (7.5), the diagnosis result grade (moderate injury), and the probability density value (0.6), and generates a smooth probability density response surface through bicubic interpolation. Locate the coordinate points corresponding to the contradiction index 7.5 on the surface, and extract the local curvature parameter of the contour line to which they belong as 0.3. According to the partial derivative direction of the local curvature parameter in the tangent plane, analyze that the direction symbol is negative, and calculate the curvature change rate as -0.2. Based on the absolute value of the curvature change rate (0.2) and the direction symbol (negative), dynamically adjust the bandwidth parameter of the probability density function to 1.5, thereby optimizing the accuracy of the probability density estimation.

[0138] Through the construction of three-dimensional grid mapping relationship, interpolation smoothing processing, local curvature parameter extraction, and curvature change rate analysis, the accurate generation of the probability density response surface and the dynamic optimization of the bandwidth parameter are realized, improving the accuracy and adaptability of probability density estimation and providing a scientific basis for the risk assessment of nerve function injury.

[0139] To solve the problem of the accuracy of generating the hemostasis intervention intensity spectrum in the out-of-hospital first aid scenario and further improve the dynamic adaptability and reliability of hemostasis intervention, the time-frequency decomposition of the microcirculation dynamic waveform is performed, and the energy attenuation slope within the preset frequency band is extracted as the attenuation rate of the microcirculation waveform; the spatial gradient distribution of the blood flow diffusion path is analyzed from the trauma diffusion trend vector of the composite feature flow, the main change direction and the secondary change direction are extracted through eigenvalue decomposition, and the spatial diffusion direction parameter is generated based on the cosine value of the included angle between the two; according to the joint distribution characteristics of the microcirculation waveform attenuation rate and the spatial diffusion direction parameter, the hemostasis pressure level is divided; combined with the result of the hemostasis pressure level division, the inertial sensing data of the patient's body position and posture are collected in real time, the variance of the trunk inclination angle and the covariance matrix of the limb displacement are extracted, and the sum of the elements on the main diagonal is calculated to generate the body position stability; based on the product coupling relationship between the hemostasis pressure level and the body position stability, the hemostasis intervention intensity spectrum is constructed, thereby realizing the accurate grading and dynamic optimization of hemostasis intervention.

[0140] In some embodiments, in step 201, extracting the attenuation rate of the microcirculation waveform from the microcirculation dynamic waveform, analyzing the spatial diffusion direction parameter from the trauma diffusion trend vector of the composite feature flow, dynamically dividing the hemostasis pressure level according to the attenuation rate of the microcirculation waveform and the spatial diffusion direction parameter, and generating the hemostasis intervention intensity spectrum in combination with the result of the hemostasis pressure level division and the current body position stability of the patient includes:

[0141] Step 601: Perform time-frequency decomposition on the microcirculation dynamic waveform, and extract the energy attenuation slope of the microcirculation dynamic waveform within the preset frequency band as the attenuation rate of the microcirculation waveform.

[0142] In step 601, time-frequency decomposition means converting the microcirculation dynamic waveform from the time domain to the frequency domain and analyzing its energy distribution at different frequencies. The preset frequency band is a frequency range set according to the characteristics of the microcirculation waveform (such as 0.01 - 0.1 Hz) for extracting the energy attenuation characteristics of the key frequency band. The energy attenuation slope represents the attenuation rate of the energy with frequency change within the preset frequency band, reflecting the dynamic change trend of the microcirculation waveform.

[0143] In the embodiment of the present application, time-frequency decomposition (such as wavelet transform) is performed on the microcirculation dynamic waveform, and the energy attenuation slope within the preset frequency band is extracted. The energy attenuation slope is used as the attenuation rate of the microcirculation waveform for subsequent hemostasis pressure level division.

[0144] Step 602: Analyze the spatial gradient distribution of the blood flow diffusion path from the trauma diffusion trend vector of the composite feature stream. By performing eigenvalue decomposition on the spatial gradient distribution, extract the main change direction and the secondary change direction of the spatial gradient distribution. Generate a spatial diffusion direction parameter based on the cosine value of the angle between the main change direction and the secondary change direction. Divide the hemostasis pressure level according to the joint distribution characteristics of the microcirculation waveform attenuation rate and the spatial diffusion direction parameter.

[0145] In step 602, the spatial gradient distribution represents the gradient change characteristics of the trauma diffusion path in space, reflecting the intensity and direction of diffusion. Distinguish the main direction and the secondary direction: eigenvalue magnitude, in eigenvalue decomposition, the main direction corresponds to the largest eigenvalue, and the secondary direction corresponds to the second largest eigenvalue. Contribution rate, the contribution rate of the main direction (i.e., the proportion of the eigenvalue in the total eigenvalue) is usually higher than that of the secondary direction. Spatial directivity, the main direction usually points to the main path of blood flow diffusion (such as the core organ area), while the secondary direction reflects the secondary path of diffusion or local changes. The joint distribution characteristic means that when the attenuation rate is higher than the critical threshold and the spatial diffusion direction parameter points to the core organ area, the hemostasis pressure level is increased. When the attenuation rate is lower than the critical threshold and the spatial diffusion direction parameter deviates from the core organ area, the hemostasis pressure level is decreased.

[0146] In the embodiment of the present application, the trauma diffusion trend vector is analyzed from the composite feature stream, and the spatial gradient distribution is calculated. Eigenvalue decomposition is performed on the spatial gradient distribution to extract the main change direction and the secondary change direction. Calculate the cosine value of the angle between the two to generate a spatial diffusion direction parameter. Divide the hemostasis pressure level according to the joint distribution characteristics of the microcirculation waveform attenuation rate and the spatial diffusion direction parameter.

[0147] Step 603: Combine the hemostasis pressure level division result, and collect the inertial sensing data of the patient's body position and posture in real time. Extract the trunk inclination variance and the limb displacement covariance matrix from the inertial sensing data. Calculate the sum of the elements on the main diagonal between the trunk inclination variance and the limb displacement covariance matrix to generate the current body position stability of the patient. Based on the product coupling relationship between the hemostasis pressure level division result and the body position stability, construct a hemostasis intervention intensity spectrum.

[0148] In step 603, the trunk inclination variance is used to reflect the variance of the trunk inclination angle change of the patient, and is used to evaluate the trunk stability. The limb displacement covariance matrix is used to reflect the covariance matrix of the limb displacement change of the patient, and is used to evaluate the limb stability. The inertial sensing data is the dynamic change data of the patient's body position and posture collected by inertial sensors (such as accelerometers, gyroscopes), and is used to evaluate the stability and motion state of the patient's body position, including acceleration data, angular velocity data, attitude angle data and displacement data.

[0149] In the embodiments of the present application, inertial sensing data of the patient's body position and posture is collected in real time, and the variance of the trunk inclination angle and the covariance matrix of limb displacements are extracted. The sum of the elements on the main diagonal of the covariance matrix is calculated to generate the body position stability. Based on the product coupling relationship between the hemostatic pressure level and the body position stability, a hemostatic intervention intensity spectrum is constructed.

[0150] The following is a specific example:

[0151] In an out-of-hospital first-aid scenario, the medical robot performs time-frequency decomposition on the microcirculation dynamic waveform of the patient, and extracts the energy attenuation slope within a preset frequency band (0.01 - 0.1 Hz) to be 0.5 as the microcirculation waveform attenuation rate. At the same time, the trauma diffusion trend vector is parsed from the composite feature stream, the spatial gradient distribution is calculated and eigenvalue decomposition is performed, the main change direction and the secondary change direction are extracted, and the cosine value of the included angle between the two is calculated to be 0.8 to generate the spatial diffusion direction parameter. According to the joint distribution characteristics of the attenuation rate (0.5) and the spatial diffusion direction parameter (0.8), the hemostatic pressure level is classified as medium. The inertial sensing data of the patient's body position and posture is collected in real time, the variance of the trunk inclination angle is extracted to be 0.2, and the sum of the elements on the main diagonal of the limb displacement covariance matrix is 0.3, generating a body position stability of 0.5. Based on the product coupling relationship between the hemostatic pressure level (medium) and the body position stability (0.5), a hemostatic intervention intensity spectrum is constructed, recommending a hemostatic pressure level of medium and an intervention time of 5 minutes.

[0152] Through the extraction of the microcirculation waveform attenuation rate, the generation of the spatial diffusion direction parameter, the classification of the hemostatic pressure level, and the calculation of the body position stability, the accurate construction of the hemostatic intervention intensity spectrum is realized, improving the scientificity and effectiveness of the hemostatic intervention, and providing reliable hemostatic decision-making support for out-of-hospital first aid.

[0153] To solve the problem of the accuracy of tissue perfusion state assessment in out-of-hospital first-aid scenarios and further improve the dynamic adaptability and reliability of the coupling analysis of pressure stimulation and microcirculation response, by monitoring the consciousness state grading index, when it is lower than the preset threshold, the grading pressure control protocol of the variable stiffness actuator of the medical robot is triggered to drive the actuator to apply stepped pressure stimulation; the grading pressure control protocol includes the pressure amplitude and the action time window of the increasing step sequence, and a pressure stimulation energy spectrum is constructed based on their product relationship; at the same time, the phase angle change amount and the waveform distortion degree within the preset frequency band are extracted from the microcirculation dynamic waveform to generate the phase mutation feature; based on the coupling response between the phase mutation feature and the pressure stimulation energy spectrum, the tissue perfusion elasticity coefficient is calculated, thereby realizing the accurate assessment and dynamic regulation of the tissue perfusion state.

[0154] In some embodiments, when the consciousness state grading index of the composite feature stream is lower than the preset index threshold in step 103, calculating the tissue perfusion elasticity coefficient through the phase mutation feature of the microcirculation dynamic waveform includes:

[0155] Step 701: Monitor the consciousness state grading index. When the consciousness state grading index is lower than the preset index threshold, trigger the grading pressure control protocol of the variable stiffness actuator of the medical robot to drive the variable stiffness actuator of the medical robot to apply a stepped pressure stimulus through the grading pressure control protocol. The grading pressure control protocol includes the pressure amplitude of the increasing ladder sequence and the action time window of each ladder. According to the product relationship between the pressure amplitude of the increasing ladder sequence and the action time window, construct a pressure stimulus energy spectrum.

[0156] In step 701, the variable stiffness actuator refers to an actuator in the medical robot that can dynamically adjust the stiffness and pressure and is used to apply a controllable pressure stimulus. The grading pressure control protocol includes a control scheme for the pressure amplitude of the increasing ladder sequence and the action time window and is used to drive the variable stiffness actuator to apply a stepped pressure stimulus. The pressure stimulus energy spectrum is used to reflect the intensity and duration of the pressure stimulus.

[0157] In the embodiments of the present application, the consciousness state grading index is monitored in real time. When the index is lower than the preset threshold, the grading pressure control protocol is triggered. The grading pressure control protocol includes the pressure amplitude of the increasing ladder sequence (such as 10 mmHg, 20 mmHg, 30 mmHg) and the action time window (such as 5 seconds, 10 seconds, 15 seconds). According to the product relationship between the pressure amplitude and the action time window, construct a pressure stimulus energy spectrum to reflect the energy distribution of the pressure stimulus.

[0158] Step 702: Extract the phase angle change amount and waveform distortion degree within a preset frequency band from the microcirculation dynamic waveform to generate a phase mutation feature. Based on the coupling response between the phase mutation feature and the pressure stimulus energy spectrum, calculate the tissue perfusion elasticity coefficient.

[0159] In step 702, the phase angle change amount represents the change amplitude of the phase angle of the microcirculation dynamic waveform within a preset frequency band and reflects the dynamic change of the blood flow state. The waveform distortion degree represents the distortion degree of the waveform shape of the microcirculation dynamic waveform within a preset frequency band and reflects the abnormal change of the blood flow state. The phase mutation feature is used to reflect the mutation feature of the microcirculation waveform.

[0160] In the embodiments of the present application, the phase angle change amount and waveform distortion degree within a preset frequency band are extracted from the microcirculation dynamic waveform to generate a phase mutation feature. Based on the coupling response between the phase mutation feature and the pressure stimulus energy spectrum, use a mathematical model (such as linear regression or neural network) to calculate the tissue perfusion elasticity coefficient.

[0161] The following is a specific example:

[0162] In an out-of-hospital first-aid scenario, the medical robot monitors that the consciousness state grading index of the patient is 2 (lower than the preset threshold of 5), triggers the grading pressure control protocol of the variable stiffness actuator, applies stepped pressure stimulation (10 mmHg / 5 s, 20 mmHg / 10 s, 30 mmHg / 15 s), and constructs a pressure stimulation energy spectrum. At the same time, the change amount of the phase angle within the preset frequency band (0.01 - 0.1 Hz) in the microcirculation dynamic waveform is extracted as 15°, and the waveform distortion degree is 0.3, generating a phase mutation feature. Based on the coupling response between the phase mutation feature and the pressure stimulation energy spectrum, the tissue perfusion elasticity coefficient is calculated as 0.7, indicating that the patient's tissue perfusion ability is good.

[0163] Through triggering stepped pressure stimulation by the grading pressure control protocol, constructing a pressure stimulation energy spectrum, extracting phase mutation features, and coupling response analysis, the accurate calculation of the tissue perfusion elasticity coefficient is achieved, improving the scientificity and effectiveness of tissue perfusion state assessment, and providing a reliable basis for first-aid decision-making.

[0164] Optionally, in step 105, based on the hemostasis intervention intensity spectrum, monitoring the change characteristics of the harmonic components of the microcirculation waveform, and based on the probability distribution of nerve function damage, feeding back the robustness parameter to the motion planner of the robotic arm to generate an anti-interference control loop, and generating a closed-loop first-aid treatment plan based on the change characteristics of the harmonic components and the anti-interference control loop, including:

[0165] Step 801: According to the time-varying mapping table of the spatial coordinate sequence and pressure amplitude at the pressurization position in the hemostasis intervention intensity spectrum, driving the robotic arm of the medical robot to perform an adaptive pressurization action to collect the energy distribution ratio and phase synchronization parameter of the microcirculation dynamic waveform within the preset frequency band, and generating the change characteristics of the harmonic components.

[0166] In step 801, the energy distribution ratio represents the energy proportion of each frequency component of the microcirculation dynamic waveform within the preset frequency band, reflecting the energy distribution characteristics of blood flow. The phase synchronization parameter represents the phase consistency of each frequency component of the microcirculation dynamic waveform within the preset frequency band, reflecting the coordination of blood flow.

[0167] In the embodiment of the present application, according to the spatial coordinate sequence and the time-varying mapping table in the hemostasis intervention intensity spectrum, the robotic arm is driven to perform an adaptive pressurization action. The energy distribution ratio and phase synchronization parameter of the microcirculation dynamic waveform within the preset frequency band are collected, and the change characteristics of the harmonic components are generated.

[0168] Step 802: When the proportion of the high-risk sub-segment in the probability distribution of nerve function injury exceeds a preset ratio, perform hierarchical compression encoding on the multi-modal sensing data stream collected by the medical robot to generate compressed data packets, trigger the compression transmission protocol for remote consultation data packets, and send the compressed data packets to the robotic arm motion planner of the medical robot based on the compression transmission protocol. Combine with a preset gain adaptive adjustment rule to output an anti-interference control loop, and the robustness parameter is carried in the compressed data packet.

[0169] In step 802, the high-risk sub-segment refers to the interval in the probability distribution of nerve function injury where the probability value exceeds a preset high-risk threshold. For example, if the preset high-risk threshold is 70%, the interval where the probability value is greater than 70% is the high-risk sub-segment. The multi-modal sensing data stream represents the real-time stream of various types of data collected by the medical robot, including the dynamic waveform of microcirculation, voice semantic features, inertial sensing data, and environmental interference data. Hierarchical compression encoding means compressing the multi-modal sensing data stream hierarchically according to priority to reduce the amount of data transmitted. The compression transmission protocol is a communication protocol for efficiently transmitting compressed data packets, ensuring the real-time and reliable transmission of data.

[0170] In the embodiment of the present application, when the proportion of the high-risk sub-segment in the probability distribution of nerve function injury exceeds a preset ratio, perform hierarchical compression encoding on the multi-modal sensing data stream to generate compressed data packets. Trigger the compression transmission protocol for remote consultation data packets and send the compressed data packets to the robotic arm motion planner. Combine with the gain adaptive adjustment rule to output an anti-interference control loop to ensure the stability and accuracy of the robotic arm operation.

[0171] Step 803: Synchronize the timestamps of the harmonic component change characteristics, the transmission status in the compressed data packet, and the robotic arm motion path correction vector in the anti-interference control loop to construct a multi-modal feedback matrix. Generate a closed-loop first-aid disposal plan based on the priority scheduling strategy of the feature weights in the multi-modal feedback matrix.

[0172] In step 803, the multi-modal feedback matrix is a matrix constructed based on the multi-modal data after timestamp synchronization alignment, which is used for comprehensive analysis and decision-making. It integrates data of different modalities (such as harmonic component change characteristics, transmission status in the compressed data packet, robotic arm motion path correction vector) into a unified mathematical structure for subsequent analysis and processing. The priority scheduling strategy is a rule for scheduling according to the priorities of the feature weights in the multi-modal feedback matrix (such as harmonic component change characteristic weight, transmission status weight, robotic arm motion path correction vector weight), which is used to determine the importance of different features in decision-making.

[0173] In the embodiments of the present application, timestamp synchronization alignment is performed on the harmonic component change characteristics, the transmission status in the compressed data packet, and the robotic arm motion path correction vector in the anti-interference control loop. A multimodal feedback matrix is constructed, and a closed-loop first aid treatment plan is generated based on the priority scheduling strategy of each feature weight.

[0174] The following is a specific example:

[0175] In an out-of-hospital first aid scenario, the medical robot drives the robotic arm to perform an adaptive pressurization action according to the spatial coordinate sequence (such as x = 10 cm, y = 5 cm, z = 2 cm) in the hemostasis intervention intensity spectrum and the time-varying mapping table (such as 10 mmHg / 5 seconds, 20 mmHg / 10 seconds), collects the energy distribution ratio (such as the low-frequency ratio is 60%) and the phase synchronization parameter (such as the phase consistency is 0.8) of the microcirculation dynamic waveform within a preset frequency band (0.01 - 0.1 Hz), and generates harmonic component change characteristics. At the same time, the probability distribution of neurological function injury shows that the proportion of high-risk sub-segments is 60% (exceeding the preset threshold of 50%). Hierarchical compression coding is performed on the multimodal sensing data stream, a compressed data packet is generated and the compressed transmission protocol of the remote consultation data packet is triggered, and the compressed data packet is sent to the robotic arm motion planner. Combining the gain adaptive adjustment rule, an anti-interference control loop is output. Finally, timestamp synchronization alignment is performed on the harmonic component change characteristics, the transmission status in the compressed data packet, and the robotic arm motion path correction vector in the anti-interference control loop, a multimodal feedback matrix is constructed, and a closed-loop first aid treatment plan is generated based on the priority scheduling strategy, recommending continued pressurization intervention and adjusting the robotic arm path.

[0176] Through the adaptive pressurization action drive, multimodal data compression transmission, anti-interference control loop generation, and multimodal feedback matrix construction, the accurate generation of the closed-loop first aid treatment plan is realized, the anti-interference and dynamic adaptability of the first aid operation are improved, and scientific and reliable decision-making support is provided for out-of-hospital first aid.

[0177] In a possible design, Figure 2 is a schematic structural diagram of an electronic device provided by the embodiments of the present application, as Figure 2 shown. The electronic device may include a storage component 21 and a processing component 22.

[0178] The storage component 21 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 22.

[0179] The processing component 22 is the above Figure 1 A multi-mode based automated emergency treatment process method of the above embodiments.

[0180] Among them, the processing component 22 may include one or more processors to execute computer instructions to complete all or part of the steps in the above methods. Of course, the processing component may also be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above methods.

[0181] The storage component 21 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0182] Of course, the electronic device may also necessarily include other components, such as an input / output interface, a display component, a communication component, etc.

[0183] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0184] The communication component is configured to facilitate communication between the electronic device and other devices in a wired or wireless manner, etc.

[0185] Among them, the electronic device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the electronic device may refer to a cloud server, and the above processing component, storage component, etc. may be basic server resources leased or purchased from the cloud computing platform.

[0186] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 shown embodiment of a multi-mode based emergency treatment process automated processing method.

[0187] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0188] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0189] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0190] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A multi-mode based emergency treatment process automation method, characterized in that: Applications in medical robots include: At the out-of-hospital emergency scene, the medical robot collects the patient's skin microcirculation dynamic waveform and extracts the directional speech semantic features under the interference of environmental noise; Comparing the microcirculation dynamic waveform and the directional speech semantic features in time and space dimensions to construct a three-dimensional time and space compensation field, and generating a composite feature flow based on the three-dimensional time and space compensation field; When the consciousness state classification index of the composite characteristic flow is lower than a preset index threshold, the tissue perfusion elasticity coefficient is calculated through the phase mutation characteristics of the microcirculation dynamic waveform; Based on the composite feature flow and the tissue perfusion elasticity coefficient, a three-layer dynamic decision tree topology network is constructed, wherein the first layer generates a hemostatic intervention intensity spectrum based on the microcirculation dynamic waveform and the composite feature flow, the second layer generates a probability distribution of neurological function damage according to the tissue perfusion elasticity coefficient and the directional speech semantic feature, and the third layer combines the real-time collected ambient light intensity and the ambulance bump index to correct the robustness parameters of the robot arm operation path; Based on the hemostatic intervention intensity spectrum, the harmonic component change characteristics of the microcirculation waveform are monitored, and based on the probability distribution of neurological function damage, the robustness parameters are fed back to the motion planner of the robotic arm to generate an anti-interference control loop, and based on the harmonic component change characteristics and the anti-interference control loop, a closed-loop emergency treatment plan is generated.

2. The method according to claim 1, characterized in that: The first level generates a hemostatic intervention intensity spectrum based on the microcirculation dynamic waveform and the composite characteristic flow, the second level generates a probability distribution of neurological function damage according to the tissue perfusion elasticity coefficient and the directional speech semantic feature, and the third level combines the real-time collected ambient light intensity and the ambulance bump index to correct the robustness parameters of the robot arm operation path, including: In the first level of the dynamic decision tree topological network, the microcirculation waveform attenuation rate is extracted from the microcirculation dynamic waveform, and the spatial diffusion direction parameter is parsed from the trauma diffusion trend vector of the composite characteristic flow, the hemostatic pressure level is dynamically divided according to the microcirculation waveform attenuation rate and the spatial diffusion direction parameter, and the hemostatic intervention intensity spectrum is generated in combination with the patient's current body position stability; In the second level of the dynamic decision tree topological network, when the semantic interruption frequency in the directional speech semantic feature is higher than a preset frequency threshold, the tissue perfusion elasticity coefficient is nonlinearly superimposed with the number of semantic logic breakpoints in the directional speech semantic feature to generate a neurological function damage contradiction index, and a neurological function damage probability distribution is generated within a preset probability interval according to the neurological function damage contradiction index; In the third level of the dynamic decision tree topological network, the vibration sensor installed on the ambulance collects the bump spectrum characteristics in real time, determines the influence coefficient of the real-time collected ambient light intensity on the visual capture sensitivity of the medical robot, calculates the ambulance bump index based on the bump spectrum characteristics, and generates robustness parameters by combining the ambulance bump index and the influence coefficient.

3. The method according to claim 2, characterized in that When the semantic interruption frequency in the directional speech semantic feature is higher than a preset frequency threshold, the tissue perfusion elasticity coefficient is nonlinearly superimposed with the number of semantic logic breakpoints in the directional speech semantic feature to generate a neurological function damage contradiction index, and a neurological function damage probability distribution is generated within a preset probability interval according to the neurological function damage contradiction index, including: The directional speech semantic feature is segmented into time series windows to extract the speech fundamental frequency and semantic logic connection strength in each window, a dynamic baseline is generated based on the historical data of the speech fundamental frequency, when the variance of the speech fundamental frequency of three consecutive windows exceeds the dynamic baseline and the semantic logic connection strength is lower than the preset connection strength threshold, the end point of the third window is marked as a semantic logic breakpoint, the number of occurrences of the semantic logic breakpoint per unit time is counted, and the semantic interruption frequency is calculated in combination with the jump distribution density of incoherent phrases, the incoherent phrases are phrases that are logically incoherent or have a jump strength greater than a preset jump strength threshold in the semantic flow extracted from the directional speech semantic feature; Determine whether the semantic interruption frequency is higher than a preset frequency threshold; if so, normalize the tissue perfusion elasticity coefficient, and use a piecewise exponential weight function to couple the normalized tissue perfusion elasticity coefficient with the semantic interruption frequency to obtain a neurological function damage contradiction index; Based on the neurological function damage contradiction index, combined with a preset probability density response surface, by solving the curvature change rate of the contour line corresponding to the neurological function damage contradiction index on the probability density response surface, the bandwidth parameter of the probability density function corresponding to the neurological function damage contradiction index is dynamically adjusted; Based on the adjusted bandwidth parameters, the Gaussian kernel density estimation of the neurological function impairment contradiction index is performed within the preset probability interval to obtain the cumulative distribution function slope corresponding to each sub-segment in the neurological function impairment contradiction index within the preset probability interval, and the probability weights of the sub-segments are redistributed according to the positions of the sign reversal points in the slope of the cumulative distribution function to generate the probability distribution of neurological function impairment.

4. The method according to claim 3, characterized in that The normalization process is performed on the tissue perfusion elasticity coefficient, and a segmented exponential weight function is used to couple the normalized tissue perfusion elasticity coefficient with the semantic interruption frequency to obtain a neurological function damage contradiction index, including: Based on the baseline stability parameter of the microcirculation dynamic waveform, the tissue perfusion elasticity coefficient is normalized, and the baseline stability parameter is obtained by calculating the variance and phase drift of the microcirculation dynamic waveform in a state without pressure stimulation; According to the real-time distribution characteristics of the semantic interruption frequency, the semantic interruption frequency is divided into a high-frequency segment, a medium-frequency segment and a low-frequency segment; In each segment, a weighted tissue perfusion elasticity coefficient is calculated according to the exponential weight function corresponding to the segment and the normalized tissue perfusion elasticity coefficient, and the product of the weighted tissue perfusion elasticity coefficient and the semantic interruption frequency of the corresponding segment is used as the coupling result of the segment; The sum of the coupling results of all segments was used as the neurological function damage contradiction index.

5. The method according to claim 3, characterized in that: The method of dynamically adjusting the bandwidth parameter of the probability density function corresponding to the neurological function damage contradiction index based on the neurological function damage contradiction index and combining with a preset probability density response surface by solving the curvature change rate of the contour line corresponding to the neurological function damage contradiction index on the probability density response surface comprises: Based on the contradiction index of neurological function damage, a three-dimensional grid mapping relationship is constructed, and the three-dimensional grid mapping relationship is interpolated and smoothed to generate a probability density response surface, wherein the first dimension is the contradiction index, the second dimension is the diagnosis result classification, and the third dimension is the probability density value; On the probability density response surface, locating the coordinate point corresponding to the neurological function damage contradiction index, and extracting the local curvature parameter of the contour line to which the coordinate point belongs; According to the direction of the partial derivative of the local curvature parameter in the tangent plane corresponding to the probability density response surface, the direction sign is obtained by analyzing the changing trend of the partial derivative direction, and the curvature change rate of the contour line is calculated. Based on the absolute value of the curvature change rate and the direction sign, the bandwidth parameter of the probability density function is dynamically adjusted.

6. The method according to claim 2, characterized in that The method extracts the microcirculation waveform attenuation rate from the microcirculation dynamic waveform, and parses the spatial diffusion direction parameter from the trauma diffusion trend vector of the composite characteristic flow, dynamically divides the hemostatic pressure level according to the microcirculation waveform attenuation rate and the spatial diffusion direction parameter, and generates a hemostatic intervention intensity spectrum in combination with the hemostatic pressure level division result and the patient's current body position stability, including: Performing time-frequency decomposition on the microcirculation dynamic waveform, and extracting the energy attenuation slope of the microcirculation dynamic waveform within a preset frequency band as the microcirculation waveform attenuation rate; The spatial gradient distribution of the blood flow diffusion path is parsed from the trauma diffusion trend vector of the composite characteristic flow, the main change direction and the secondary change direction of the spatial gradient distribution are extracted by performing eigenvalue decomposition on the spatial gradient distribution, a spatial diffusion direction parameter is generated based on the cosine value of the angle between the main change direction and the secondary change direction, and the hemostasis pressure level is divided according to the joint distribution characteristics of the microcirculation waveform attenuation rate and the spatial diffusion direction parameter; Combined with the hemostatic pressure level classification result, the inertial sensor data of the patient's posture is collected in real time, the trunk inclination variance and limb displacement covariance matrix are extracted from the inertial sensor data, and the sum of the elements on the main diagonal between the trunk inclination variance and the limb displacement covariance matrix is ​​calculated to generate the patient's current posture stability. Based on the product coupling relationship between the hemostatic pressure level classification result and the posture stability, a hemostatic intervention intensity spectrum is constructed.

7. The method according to claim 1, characterized in that When the consciousness state classification index of the composite characteristic flow is lower than a preset index threshold, the tissue perfusion elasticity coefficient is calculated by the phase mutation characteristics of the microcirculation dynamic waveform, including: Monitoring the consciousness state classification index, and when the consciousness state classification index is lower than a preset index threshold, triggering a graded pressure control protocol of the variable stiffness actuator of the medical robot, so as to drive the variable stiffness actuator of the medical robot to apply a stepped pressure stimulus through the graded pressure control protocol, wherein the graded pressure control protocol includes a pressure amplitude of an increasing step sequence and an action time window of each step, and constructing a pressure stimulus energy spectrum according to a multiplication relationship between the pressure amplitude of the increasing step sequence and the action time window; The phase angle variation and waveform distortion within a preset frequency band are extracted from the microcirculation dynamic waveform to generate a phase mutation feature, and the tissue perfusion elasticity coefficient is calculated based on the coupling response between the phase mutation feature and the pressure stimulation energy spectrum.

8. The method according to claim 1, characterized in that Based on the hemostatic intervention intensity spectrum, the harmonic component change characteristics of the microcirculation waveform are monitored, and based on the probability distribution of nerve function damage, the robustness parameter is fed back to the motion planner of the robot arm to generate an anti-interference control loop, and based on the harmonic component change characteristics and the anti-interference control loop, a closed-loop emergency treatment plan is generated, including: According to the time-varying mapping table of the spatial coordinate sequence of the pressurization position in the hemostasis intervention intensity spectrum and the pressure amplitude, the medical robot mechanical arm is driven to perform an adaptive pressurization action to collect the energy distribution ratio and phase synchronization parameters of the microcirculation dynamic waveform within a preset frequency band, and generate harmonic component change characteristics; When the proportion of high-risk sub-segments of the probability distribution of neurological function damage exceeds a preset proportion, the collected multimodal sensor data stream of the medical robot is hierarchically compressed and encoded to generate a compressed data packet, triggering a compression transmission protocol for a remote consultation data packet, so as to send the compressed data packet to the mechanical arm motion planner of the medical robot based on the compression transmission protocol, and output an anti-interference control loop in combination with a preset gain adaptive adjustment rule, wherein the robustness parameter is carried in the compressed data packet; The harmonic component change characteristics, the transmission status in the compressed data packet and the robot arm motion path correction vector in the anti-interference control loop are timestamped and aligned to construct a multimodal feedback matrix, and a closed-loop emergency treatment plan is generated based on the priority scheduling strategy of each feature weight in the multimodal feedback matrix.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a multi-mode based emergency treatment process automation processing method as described in any one of claims 1 to 8.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, an automated processing method for emergency treatment process based on multiple modes as described in any one of claims 1 to 8 is implemented.

Images