Anesthesia depth dynamic identification auxiliary analysis method and device, server and medium

Abstract

The invention discloses an anesthesia depth dynamic identification auxiliary analysis method and device, a server and a medium, and the method comprises the steps: collecting a plurality of signals in a period, carrying out the preprocessing of the signals, and forming a signal sequence; inputting the signal sequence into a random forest model, a vector machine model, a gradient elevator model and a convolutional neural network model which are trained to obtain four output anesthesia auxiliary analysis results, and fusing the four anesthesia auxiliary analysis results according to the weight value corresponding to each model to obtain an anesthesia auxiliary analysis result; before the signals of the next period are input into the four models, evaluation values are calculated according to the current anesthesia scene, the evaluated anesthesia physiological signals and the model evaluation mode, each weight value is adjusted according to the evaluation values, and a new-period anesthesia auxiliary analysis result is formed. A more accurate auxiliary analysis result can be obtained, and an accurate auxiliary judgment result is provided for medical workers to analyze the anesthesia state.

Description

Technical Field

[0001] The present invention relates to the field of auxiliary analysis technology, and in particular to an auxiliary analysis method, device, server and medium for dynamic identification of anesthesia depth. Background Art

[0002] Monitoring of anesthesia depth plays a vital role in anesthesiology. Accurate assessment of anesthesia depth can effectively avoid excessive or insufficient anesthesia and ensure that patients are in a safe and comfortable state during surgery. In recent years, some anesthesia depth monitoring methods based on conventional physiological signals have been gradually proposed and related research has been carried out. These methods infer the depth of anesthesia by monitoring conventional physiological signals such as heart rate, blood oxygen saturation, respiratory rate, and blood pressure. These signals are relatively easy to obtain and have high clinical universality, and can provide an indirect reference for the depth of anesthesia. However, changes in a single physiological signal often cannot fully reflect the depth of anesthesia, especially when the depth of anesthesia is shallow or changes rapidly, traditional methods are difficult to provide accurate judgments.

[0003] Most existing anesthesia depth monitoring systems are based on the analysis of a single or a few physiological signals, and their monitoring accuracy and real-time performance still face great challenges. During surgery, the depth of anesthesia is affected by many factors, such as individual differences in patients, the metabolism of anesthetic drugs, and the type of surgery. These factors make it difficult for changes in a single physiological signal to fully reflect the depth of anesthesia, and thus it is impossible to provide an accurate reference. Summary of the invention

[0004] The embodiments of the present invention provide a method, device, server and medium for auxiliary analysis of dynamic identification of anesthesia depth to solve the technical problem of low accuracy of the estimation result of anesthesia depth in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a method for assisting in the analysis of dynamic identification of anesthesia depth, comprising: Collecting a plurality of anesthesia physiological signals by using a wearable device within a cycle, and preprocessing the plurality of anesthesia physiological signals to form a signal sequence; The signal sequence is respectively input into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model to obtain a first anesthesia-assisted analysis result, a second anesthesia-assisted analysis result, a third anesthesia-assisted analysis result and a fourth anesthesia-assisted analysis result; The first anesthesia-assisted analysis result, the second anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result are fused according to the weight value corresponding to each model to obtain the anesthesia-assisted analysis result; Before inputting the signal of the next cycle into the four models, the anesthesia physiological signal for evaluation is selected according to the current anesthesia scenario, and the model evaluation method is determined based on the evaluated anesthesia physiological signal, and the evaluation value is calculated according to the model evaluation method, and each weight value is adjusted according to the evaluation value to form the anesthesia auxiliary analysis result of the new cycle.

[0006] In a second aspect, an embodiment of the present invention further provides an auxiliary analysis device for dynamic identification of anesthesia depth, comprising: A preprocessing module, used to collect a variety of anesthesia physiological signals using a wearable device within a cycle, and preprocess the multiple anesthesia physiological signals to form a signal sequence; An input module, used to input the signal sequence into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model, respectively, to obtain the output first anesthesia-assisted analysis result, second anesthesia-assisted analysis result, third anesthesia-assisted analysis result and fourth anesthesia-assisted analysis result; A fusion module, used for fusing the first anesthesia-assisted analysis result, the second output anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result according to the weight value corresponding to each model to obtain the anesthesia-assisted analysis result; The adjustment module is used to select and evaluate the anesthesia physiological signal according to the current anesthesia scenario before inputting the signal of the next cycle into the four models, and determine the model evaluation method according to the evaluated anesthesia physiological signal, and calculate the evaluation value according to the model evaluation method, and adjust each weight value according to the evaluation value to form the anesthesia auxiliary analysis result of the new cycle.

[0007] In a third aspect, an embodiment of the present invention further provides a server, including: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the anesthesia depth dynamic identification auxiliary analysis method provided in the above-mentioned embodiment.

[0008] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to perform the method for auxiliary analysis of dynamic identification of anesthesia depth as provided in the above embodiment.

[0009] The embodiments of the present invention provide a method, device and storage medium for dynamic identification and auxiliary analysis of anesthesia depth. A wearable device is used to collect a variety of anesthesia physiological signals within a cycle, and the multiple anesthesia physiological signals are preprocessed to form a signal sequence; the signal sequence is respectively input into a trained random forest model, a vector machine model, a gradient boosting machine model and a convolutional neural network model to obtain a first anesthesia auxiliary analysis result, a second output anesthesia auxiliary analysis result, a third anesthesia auxiliary analysis result and a fourth anesthesia auxiliary analysis result; the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result are fused according to the weight value corresponding to each model to obtain an anesthesia auxiliary analysis result; before the signal of the next cycle is input into the four models, an anesthesia physiological signal for evaluation is selected according to the current anesthesia scenario, and a model evaluation method is determined according to the evaluated anesthesia physiological signal, and an evaluation value is calculated according to the model evaluation method, and each weight value is adjusted according to the evaluation value to form an anesthesia auxiliary analysis result of a new cycle. The four models can be used to output corresponding auxiliary analysis results according to their respective characteristics. They can make full use of a variety of anesthesia physiological signals, analyze the correlation between them separately, and analyze the prediction results of the four models according to actual scenarios to determine the prediction accuracy of each model. The prediction result weight of each model can be further adjusted dynamically according to the prediction accuracy, so as to obtain more accurate auxiliary analysis results and provide medical workers with accurate auxiliary judgment results for anesthesia status analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 1 is a flow chart of a method for dynamic identification and auxiliary analysis of anesthesia depth provided in Embodiment 1 of the present invention; Figure 2 is a flow chart of a method for dynamic identification and auxiliary analysis of anesthesia depth provided in Embodiment 2 of the present invention; Figure 3 is a schematic diagram of the structure of the device for dynamic identification and auxiliary analysis of anesthesia depth provided in the third embodiment of the present invention; Figure 4 It is a schematic diagram of the structure of the server provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0011] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0012] Embodiment 1 Figure 1 : is a flow chart of a method for dynamic identification and auxiliary analysis of anesthesia depth provided in Embodiment 1 of the present invention. This embodiment is applicable to the case of dynamic identification and auxiliary analysis of anesthesia depth during surgical anesthesia. The method can be performed by a device for dynamic identification and auxiliary analysis of anesthesia depth, and specifically includes the following steps: Step 110: collect multiple anesthesia physiological signals using a wearable device within a cycle, pre-process the multiple anesthesia physiological signals, and form a signal sequence.

[0013] In this embodiment, an acquisition cycle can be pre-set, and the cycle can be pre-set, and illustratively, 30s can be used, taking into account the integrity and flexibility of the collected data. Exemplarily, a wearable physiological signal collector can be used to collect various physiological signals during anesthesia. The overall design of the wearable multi-sensor module is based on flexible materials, combined with flexible circuit boards and sensor technology to achieve a bendable fit in the biological signal acquisition area and reduce signal artifacts caused by body surface movement. Using a multi-channel physiological signal sensing unit, a flexible integrated sensor structure, a multi-modal signal acquisition and a dynamic gain control circuit, the signals corresponding to conventional physiological parameters such as the patient's heart rate, respiratory rate, blood oxygen saturation and skin temperature can be collected in real time.

[0014] After the acquisition is completed, the various acquired signals can be preprocessed to achieve denoising. For example, by weighted calculation of the signal quality index (SQI): QRS wave power spectrum distribution pSQI, kurtosis kSQI and baseline relative power basSQI. Use the global time reference to synchronize all acquired signals to achieve timestamp alignment. Form a variety of physiological signal sequences.

[0015] Step 120, respectively input the signal sequence into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model, and respectively obtain the output of the first anesthesia assisted analysis result, the second output anesthesia assisted analysis result, the third anesthesia assisted analysis result and the fourth anesthesia assisted analysis result.

[0016] The signal sequence is , each sample , four different models (RF, SVM, GBM, CNN) can be used to predict the depth of anesthesia, and four different prediction results can be obtained.

[0017] in, is the prediction result of the random forest model; is the prediction result of the support vector machine model; is the prediction result of the gradient boosting machine model; is the prediction result of the convolutional neural network model.

[0018] Step 130 , fusing the first anesthesia-assisted analysis result, the second anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result according to the weight value corresponding to each model to obtain the anesthesia-assisted analysis result.

[0019] In this embodiment, the fusion model is used to output the predicted value It is the weighted average of the prediction results of each model. The fusion model combines the outputs of the four models by weighted average. The final fusion model outputs the prediction value It is the weighted average of the prediction results of each model, and the calculation formula is: , Among them, the weight The following constraints are met: , is the weight of the random forest model; is the weight of the support vector machine model; is the weight of the gradient boosting machine model; is the weight of the convolutional neural network model.

[0020] The prediction results are mapped to a dimensionless anesthesia depth index of 0-100. Initially, the same weight can be set uniformly for each model.

[0021] Step 140, before inputting the signal of the next cycle into the four models, select the evaluation anesthesia physiological signal according to the current anesthesia scenario, and determine the model evaluation method according to the evaluation anesthesia physiological signal, and calculate the evaluation value according to the model evaluation method, and adjust each weight value according to the evaluation value to form a new cycle anesthesia auxiliary analysis result.

[0022] The anesthesia physiological process is a complex process, therefore, it is necessary to dynamically monitor the changes in physiological signals and dynamically return the anesthesia analysis results. Therefore, in this embodiment, before the data collected in the next cycle is input into the four models, it is necessary to readjust the weight value of each model according to the current scenario and the scenario requirements, so as to obtain the anesthesia auxiliary analysis results that meet the current state. Exemplarily, the core evaluation anesthesia physiological signals can be determined according to the anesthesia scenario, and the model evaluation method can be determined using the core evaluation anesthesia physiological signals, and the weight values ​​can be adjusted.

[0023] Exemplarily, it may include: calculating the sensitivity of each model to the heart rate signal in anesthesia scenarios for vascular surgery; calculating the predicted stability of each model when the heart rate signal changes; determining the model whose sensitivity is greater than a preset sensitivity threshold and whose predicted stability is less than a preset stability threshold, and increasing the weight of the model.

[0024] Heart rate is an important evaluation indicator during vascular surgery anesthesia. Therefore, whether the model can respond sensitively to changes in heart rate can determine whether the model can accurately grasp these changes and give relatively accurate evaluation results.

[0025] Therefore, in this embodiment, the sensitivity of each model to the heart rate is calculated first, and the applicability and accuracy of each model in the scenario are evaluated based on the sensitivity.

[0026] Optionally, the sensitivity of each model to heart rate can be calculated as follows: ;in, is a sensitivity index used to measure the model's response to changes in heart rate; is the rate of change calculation formula, which can be calculated by differential method. is the output of the model, which refers to the anesthesia depth index predicted by the model; It is the heart rate signal.

[0027] In the above manner, by calculating the ratio of the model output change rate and the heart rate change rate, the sensitivity of each model to the heart rate can be calculated.

[0028] Although the sensitivity of the model to heart rate can be used to determine whether the model can sense changes in heart rate and adjust quickly, these models may also be overly sensitive, resulting in erroneous evaluation results.

[0029] Therefore, in this embodiment, the prediction stability of each model when the heart rate signal changes needs to be calculated. Exemplarily, the prediction stability of each model when the heart rate signal changes can be calculated in the following manner: ,in, is the predictive stability of the model; The standard deviation of the model output.

[0030] Specifically, the standardization can be calculated as follows: ; in, is the number of samples of model data; is the standard deviation of the model output, which reflects the stability of the model output; are the output values ​​of different models; is the average of all model outputs.

[0031] For each model, after calculating the heart rate sensitivity and predicted stability, determine the model whose sensitivity is greater than the preset sensitivity threshold and whose predicted stability is less than the preset stability threshold, and increase the weight of the above-determined model, which can improve the accuracy of the anesthesia-assisted analysis results.

[0032] The above process can be repeated in the next cycle to readjust the weight of each model and achieve dynamic anesthesia-assisted analysis results.

[0033] In this embodiment, the wearable device is used to collect Collect a variety of anesthesia physiological signals, pre-process the multiple anesthesia physiological signals to form a signal sequence; input the signal sequence into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model respectively, and obtain the output of the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result respectively; fuse the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result according to the weight value corresponding to each model to obtain the anesthesia auxiliary analysis result; before inputting the signal of the next cycle into the four models, according to the current anesthesia scenario, select the evaluation anesthesia physiological signal, and determine the model evaluation method according to the evaluation anesthesia physiological signal, and calculate the evaluation value according to the model evaluation method, and adjust each weight value according to the evaluation value to form the anesthesia auxiliary analysis result of the new cycle. The four models can be used to output corresponding auxiliary analysis results according to their respective characteristics. They can make full use of a variety of anesthesia physiological signals, analyze the correlation between them separately, and analyze the prediction results of the four models according to actual scenarios to determine the prediction accuracy of each model. The prediction result weight of each model can be further adjusted dynamically according to the prediction accuracy, so as to obtain more accurate auxiliary analysis results and provide medical workers with accurate auxiliary judgment results for anesthesia status analysis.

[0034] In a preferred implementation of this embodiment, the method may further add the following step: when the duration of vascular surgery exceeds a preset duration threshold, increase the weights of skin temperature and heart rate variability. In long-term surgery, skin temperature (ST) and heart rate variability (HRV) are key indicators because they can reflect the patient's metabolic state and the stability of the autonomic nervous system. Therefore, it is necessary to increase the weights of these two features. Exemplarily, the weights of skin temperature and heart rate variability can be increased by adjusting the model contribution values ​​of the above two signals and reversely adjusting the collected signal values ​​of the two signals. Exemplarily, this can be achieved in the following way: ; in, is the weight of the heart rate feature; 0.35 is the initial weight of the heart rate feature; is the SHAP value of the heart rate feature; is the average of the SHAP values ​​of all features.

[0035] .

[0036] in, is the weight of the blood pressure feature; 0.30 is the initial weight of the blood pressure feature; is the SHAP value of the blood pressure feature; The average of the SHAP values ​​of all features.

[0037] Utilizing the above method, the accuracy of the model anesthesia-assisted analysis can be further improved during long-term surgery.

[0038] In another preferred implementation manner obtained in this embodiment, the method may further add the following steps: calculating the contribution of each anesthetic physiological signal in each model, determining the anesthetic physiological signal whose contribution exceeds a preset contribution threshold in all models, and increasing the feature weight of the anesthetic physiological signal according to a preset ratio. SHAP is a feature importance analysis method based on game theory, which can explain the contribution of each feature to model prediction. By calculating the SHAP value of each anesthetic physiological signal in each model, determining the anesthetic physiological signal whose contribution exceeds a preset contribution threshold in all models, and then determining the importance of the anesthetic physiological signal at the current stage. Therefore, it is necessary to increase the weight value of the anesthetic physiological signal. Exemplarily, the contribution value can still be used to increase the anesthetic physiological signal by adjusting the contribution value according to a preset ratio.

[0039] Embodiment 2 Figure 2It is a flow chart of the auxiliary analysis method for dynamic identification of anesthesia depth provided in Example 2 of the present invention. This embodiment is optimized on the basis of the above-mentioned embodiment, and the evaluation of anesthesia physiological signals is selected according to the current anesthesia scenario, and the model evaluation method is determined according to the evaluated anesthesia physiological signals. The specific optimization is: in the local anesthesia scenario of non-vascular surgery, the contribution values ​​of all anesthesia physiological signals in each model are used to calculate the diversity of all anesthesia physiological signals used by each model; the correlation between the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result is calculated to obtain a correlation score; the model whose diversity exceeds the diversity threshold and the correlation is within the preset correlation range is determined, and the weight value of the model is increased.

[0040] See also Figure 2 The method for dynamic identification and auxiliary analysis of anesthesia depth comprises: Step 210: collect multiple anesthesia physiological signals using a wearable device within a cycle, and pre-process the multiple anesthesia physiological signals to form a signal sequence.

[0041] In step 220, the signal sequence is input into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model respectively to obtain the output of the first anesthesia assisted analysis result, the second anesthesia assisted analysis result, the third anesthesia assisted analysis result and the fourth anesthesia assisted analysis result respectively.

[0042] Step 230 , according to the weight value corresponding to each model, the first anesthesia-assisted analysis result, the second output anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result are fused to obtain the anesthesia-assisted analysis result.

[0043] Step 240, before inputting the next cycle signal into the four models, in the scenario of local anesthesia for non-vascular surgery, the contribution values ​​of all anesthesia physiological signals in each model are used to calculate the diversity of all anesthesia physiological signals adopted by each model.

[0044] In local anesthesia or non-cardiovascular surgery, heart rate is not a key indicator, so it is necessary to keep the model weights relatively balanced to avoid over-reliance on heart rate. The overall prediction performance is optimized by dynamically adjusting the model weights and feature weights.

[0045] Therefore, in this embodiment, the comprehensive processing ability of the model for multiple physiological signals is evaluated through feature diversity. Feature diversity is measured by calculating the SHAP value distribution of different features of the model. For example, it can be calculated as follows: The standard deviation can also be calculated using the method provided in the above embodiment.

[0046] in, is the feature diversity score of the model; is the SHAP value of all features.

[0047] Step 250, calculating the correlation among the first anesthesia-assisted analysis result, the second output anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result to obtain a correlation score.

[0048] In this embodiment, the model prediction consistency (ModelConsistencyScore, MCS) is used to evaluate the consistency of the model prediction results. MCS measures the consistency of the model output by calculating the correlation between the model output and other model outputs: ; in, Predict a consistency score for the model, is the output of the current model; is the output of other models.

[0049] Step 260, determine the model whose diversity exceeds the diversity threshold and whose correlation is within the preset correlation range, increase the weight value of the model, and adjust each weight value according to the evaluation value to form a new cycle anesthesia auxiliary analysis result.

[0050] In this embodiment, a model whose feature diversity is greater than a preset diversity threshold and whose correlation is within a preset correlation range is selected, and the weight value of the model is increased during fusion.

[0051] This embodiment is optimized by selecting and evaluating anesthesia physiological signals according to the current anesthesia scenario, and determining the model evaluation method according to the evaluated anesthesia physiological signals: in the case of local anesthesia scenarios for non-vascular surgery, the contribution values ​​of all anesthesia physiological signals in each model are used to calculate the diversity of all anesthesia physiological signals used by each model; the correlation between the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result is calculated to obtain a correlation score; the model whose diversity exceeds the diversity threshold is determined, and whether the correlation of the model is within the preset correlation range is determined, and the weight value of the model is increased. For other anesthesia scenarios, standardized and consistent evaluations can be used to determine which model is more accurate in this scenario, and then its weight can be dynamically adjusted to further improve the accuracy of anesthesia auxiliary analysis in different anesthesia scenarios.

[0052] In a preferred implementation of this embodiment, the method may further add the following steps: calculate the contribution value of each anesthesia physiological signal of each cycle, and when the contribution value of the respiratory rate or blood oxygen saturation is higher than the contribution value of other anesthesia victory signals by more than a preset multiple, increase the feature weight of the respiratory rate or blood oxygen saturation. Children usually metabolize anesthetic drugs faster and have a more fragile respiratory system, so respiratory rate (RR) and blood oxygen saturation (SpO2) are key indicators. Increasing the weights of these two features can better reflect the tolerance of children to anesthesia and changes in physiological state. According to the real-time SHAP contribution value analysis, if the SHAP value of the respiratory rate or blood oxygen saturation is significantly higher than other features, its weight is further increased.

[0053] Exemplarily, this can be achieved in the following ways: , is the weight of the respiratory frequency feature; 0.35 is the initial weight of the respiratory frequency feature; is the SHAP value of the respiratory frequency feature; is the average of the SHAP values ​​of all features.

[0054] , is the weight of the blood oxygen saturation feature; 0.30 is the initial weight of the blood oxygen saturation feature; is the SHAP value of the blood oxygen saturation feature; is the average of the SHAP values ​​of all features.

[0055] Through the above method, more accurate anesthesia assistance analysis results can be given for children's anesthesia scenarios.

[0056] In another preferred implementation of this embodiment, the method may further include the following steps: determining a model whose correlation score is less than a preset correlation score threshold, and reducing the weight of the model. If the correlation score is too low, it may indicate that its prediction result is too different from other models, and its prediction accuracy is not high, therefore, it is necessary to reduce the weight of the model.

[0057] Embodiment 3 Figure 3 is a schematic diagram of the structure of the anesthesia depth dynamic identification auxiliary analysis device provided in the third embodiment of the present invention, see Figure 3 The anesthesia depth dynamic identification auxiliary analysis device comprises: A preprocessing module 310 is used to collect a variety of anesthesia physiological signals using a wearable device within a cycle, and preprocess the multiple anesthesia physiological signals to form a signal sequence; An input module 320, used to input the signal sequence into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model, respectively, to obtain a first anesthesia-assisted analysis result, a second anesthesia-assisted analysis result, a third anesthesia-assisted analysis result and a fourth anesthesia-assisted analysis result; A fusion module 330, configured to fuse the first anesthesia-assisted analysis result, the second anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result according to the weight value corresponding to each model to obtain an anesthesia-assisted analysis result; The adjustment module 340 is used to select and evaluate the anesthesia physiological signal according to the current anesthesia scenario before inputting the signal of the next cycle into the four models, and determine the model evaluation method according to the evaluated anesthesia physiological signal, and calculate the evaluation value according to the model evaluation method, and adjust each weight value according to the evaluation value to form a new cycle anesthesia auxiliary analysis result.

[0058] The anesthesia depth dynamic identification auxiliary analysis device provided in the present embodiment collects multiple anesthesia physiological signals by using a wearable device within one cycle, pre-processes the multiple anesthesia physiological signals, and forms a signal sequence; respectively inputs the signal sequence into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model, and respectively obtains the output of the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result; according to the weight value corresponding to each model, the first anesthesia auxiliary analysis result, the second output anesthesia auxiliary analysis result, the third anesthesia auxiliary analysis result and the fourth anesthesia auxiliary analysis result are fused to obtain the anesthesia auxiliary analysis result; before inputting the signal of the next cycle into the four models, according to the current anesthesia scenario, select the evaluation anesthesia physiological signal, and according to the evaluation anesthesia physiological signal, determine the model evaluation method, and calculate the evaluation value according to the model evaluation method, and adjust each weight value according to the evaluation value to form the anesthesia auxiliary analysis result of the new cycle. The four models can be used to output corresponding auxiliary analysis results according to their respective characteristics. They can make full use of a variety of anesthesia physiological signals, analyze the correlation between them separately, and analyze the prediction results of the four models according to actual scenarios to determine the prediction accuracy of each model. The prediction result weight of each model can be further adjusted dynamically according to the prediction accuracy, so as to obtain more accurate auxiliary analysis results and provide medical workers with accurate auxiliary judgment results for anesthesia status analysis.

[0059] Based on the above embodiments, the adjustment module includes: a sensitivity calculation unit, used to calculate the sensitivity of each model to the heart rate signal in anesthesia scenarios for vascular surgery; a stability calculation unit, used to calculate the prediction stability of each model when the heart rate signal changes; The first weight increasing unit is used to determine a model whose sensitivity is greater than a preset sensitivity threshold and whose predicted stability is less than a preset stability threshold, and to increase the weight of the model.

[0060] Based on the above embodiments, the adjustment module further includes: A diversity calculation unit, used for selecting all anesthesia physiological signals as evaluation anesthesia physiological signals in a local anesthesia scenario for non-vascular surgery, and calculating the diversity of all anesthesia physiological signals used by each model using contribution values ​​of all anesthesia physiological signals in each model; a correlation calculation unit, used to calculate the correlation among the first anesthesia-assisted analysis result, the second output anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result, and obtain a correlation score; The second weight increasing unit is used to determine a model whose diversity exceeds a diversity threshold and whose correlation is within a preset correlation range, and to increase a weight value of the model.

[0061] Based on the above embodiments, the device further includes: The weight reduction module is used to determine a model whose relevance score is less than a preset relevance score threshold and reduce the weight of the model.

[0062] Based on the above embodiments, the device further includes: The first growth module is used to calculate the contribution of each anesthetic physiological signal in each model, determine the anesthetic physiological signals whose contributions in all models exceed a preset contribution threshold, and increase the characteristic weight of the anesthetic physiological signal according to a preset ratio.

[0063] Based on the above embodiments, the device further includes: The second growth module is used to calculate the contribution value of each anesthesia physiological signal in each cycle, and when the contribution value of the respiratory rate or blood oxygen saturation is higher than the contribution value of other anesthesia victory signals by more than a preset multiple, increase the characteristic weight of the respiratory rate or blood oxygen saturation.

[0064] Based on the above embodiments, the device further includes: The third growth module is used to increase the weights of skin temperature and heart rate variability when the duration of the vascular surgery exceeds a preset duration threshold.

[0065] The device for dynamic identification and auxiliary analysis of anesthesia depth provided in the embodiment of the present invention can execute the method for dynamic identification and auxiliary analysis of anesthesia depth provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0066] Embodiment 4 Figure 4 A schematic diagram of the structure of a server provided in Embodiment 6 of the present invention. Figure 4 A block diagram of an exemplary server 12 suitable for use in implementing embodiments of the present invention is shown. Figure 4 The server 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0067] like Figure 4 As shown, the server 12 is in the form of a general purpose computing device. The components of the server 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).

[0068] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0069] The server 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the server 12, including volatile and non-volatile media, removable and non-removable media.

[0070] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The server 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 4 not shown, usually called a "hard drive"). Although Figure 4Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.

[0071] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0072] The server 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable users to interact with the server 12, and / or may communicate with any device that enables the server 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. In addition, the server 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the server 12 via a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the server 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0073] The processing unit 16 executes various functional applications and data processing by running the programs stored in the system memory 28, such as implementing the anesthesia depth dynamic identification auxiliary analysis method provided in the embodiment of the present invention.

[0074] Embodiment 5 Embodiment 5 of the present invention further provides a storage medium comprising computer executable instructions, wherein the computer executable instructions, when executed by a computer processor, are used to execute any of the anesthesia depth dynamic identification auxiliary analysis methods provided in the above embodiments.

[0075] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0076] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0077] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0078] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or device. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0079] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for dynamic identification and auxiliary analysis of anesthesia depth, characterized in that: include: Collecting a plurality of anesthesia physiological signals by using a wearable device within a cycle, and preprocessing the plurality of anesthesia physiological signals to form a signal sequence; The signal sequence is respectively input into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model to obtain a first anesthesia-assisted analysis result, a second anesthesia-assisted analysis result, a third anesthesia-assisted analysis result and a fourth anesthesia-assisted analysis result; The first anesthesia-assisted analysis result, the second anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result are fused according to the weight value corresponding to each model to obtain the anesthesia-assisted analysis result; Before inputting the signal of the next cycle into the four models, the anesthesia physiological signal for evaluation is selected according to the current anesthesia scenario, and the model evaluation method is determined based on the evaluated anesthesia physiological signal, and the evaluation value is calculated according to the model evaluation method, and each weight value is adjusted according to the evaluation value to form the anesthesia auxiliary analysis result of the new cycle.

2. The method according to claim 1, characterized in that: The step of selecting an evaluation anesthesia physiological signal according to the current anesthesia scenario and determining a model evaluation method according to the evaluation anesthesia physiological signal includes: The sensitivity of each model to the heart rate signal was calculated when performing anesthesia for vascular surgery scenarios; Calculate the prediction stability of each model when the heart rate signal changes; Determine a model whose sensitivity is greater than a preset sensitivity threshold and whose predicted stability is less than a preset stability threshold, and increase the weight of the model.

3. The method according to claim 1, characterized in that The step of selecting an evaluation anesthesia physiological signal according to the current anesthesia scenario and determining a model evaluation method according to the evaluation anesthesia physiological signal includes: In the local anesthesia scenario for non-vascular surgery, all anesthesia physiological signals are selected as evaluation anesthesia physiological signals, and the contribution values ​​of all anesthesia physiological signals in each model are used to calculate the diversity of all anesthesia physiological signals used by each model; Calculating the correlation between the first anesthesia-assisted analysis result, the second anesthesia-assisted analysis result, the third anesthesia-assisted analysis result, and the fourth anesthesia-assisted analysis result to obtain a correlation score for each model; Models whose diversity exceeds a diversity threshold and whose relevance scores are within a preset score range are determined, and the weight value of the model is increased.

4. The method according to claim 3, characterized in that The method further comprises: Models with relevance scores less than a preset relevance score threshold are determined, and the weights of the models are reduced.

5. The method according to claim 1, characterized in that The method further comprises: The contribution of each anesthetic physiological signal in each model is calculated, and the anesthetic physiological signal whose contribution in all models exceeds a preset contribution threshold is determined, and the characteristic weight of the anesthetic physiological signal is increased according to a preset ratio.

6. The method according to claim 1, characterized in that The method further comprises: The contribution value of each anesthesia physiological signal of each cycle is calculated, and when the contribution value of the respiratory rate or blood oxygen saturation is higher than the contribution value of other anesthesia victory signals by more than a preset multiple, the characteristic weight of the respiratory rate or blood oxygen saturation is increased.

7. The method according to claim 2, characterized in that The method further comprises: When the duration of vascular surgery exceeds a preset duration threshold, the weights of skin temperature and heart rate variability are increased.

8. A dynamic identification and analysis device for anesthesia depth, characterized in that: include: A preprocessing module, used to collect a variety of anesthesia physiological signals using a wearable device within a cycle, and preprocess the multiple anesthesia physiological signals to form a signal sequence; An input module, used to input the signal sequence into the trained random forest model, vector machine model, gradient boosting machine model and convolutional neural network model, respectively, to obtain the output first anesthesia-assisted analysis result, second anesthesia-assisted analysis result, third anesthesia-assisted analysis result and fourth anesthesia-assisted analysis result; A fusion module, used for fusing the first anesthesia-assisted analysis result, the second output anesthesia-assisted analysis result, the third anesthesia-assisted analysis result and the fourth anesthesia-assisted analysis result according to the weight value corresponding to each model to obtain the anesthesia-assisted analysis result; The adjustment module is used to select and evaluate the anesthesia physiological signal according to the current anesthesia scenario before inputting the signal of the next cycle into the four models, and determine the model evaluation method according to the evaluated anesthesia physiological signal, and calculate the evaluation value according to the model evaluation method, and adjust each weight value according to the evaluation value to form the anesthesia auxiliary analysis result of the new cycle.

9. A server, characterized in that: The server comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the anesthesia depth dynamic identification auxiliary analysis method as described in any one of claims 1-7.

10. A storage medium containing computer executable instructions, characterized in that: When executed by a computer processor, the computer executable instructions are used to execute the anesthesia depth dynamic identification auxiliary analysis method as described in any one of claims 1-7.

Images