Tumor therapy electric field adjusting system based on multiple modes

Abstract

The invention discloses a tumor treatment electric field adjusting system based on multiple modes, relates to the field of biomedical engineering, and aims to realize intelligent adjustment of a tumor treatment electric field. According to the technical scheme, a programmable electric field generator is designed, and dynamic adjustment is conducted according to the tumor type, the physiological status of a patient and the treatment progress; the method comprises the following steps: collecting tumor data and body data of historical patient tumors and received electric field therapy parameter data, extracting features, constructing a training feature matrix and a test feature matrix, and correspondingly selecting a training feature tag matrix and a test feature tag matrix; constructing an initial neural network model, and training and optimizing the neural network model to obtain a final neural network model; and monitoring and collecting electrophysiological parameters and metabolic states of a tumor area and peripheral tissues and reactions of the whole body of the patient in real time, extracting a feature set, transmitting the real-time body data feature set of the tumor patient to the final neural network model, and adjusting electric field parameters as required.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical engineering, and in particular, relates to a tumor treatment electric field regulation system based on multi-mode. Background Art

[0002] In the field of tumor treatment, electric field therapy, as a non-invasive treatment method with no side effects, has gradually received attention in recent years. In the prior art, there are already a variety of electric field-based tumor treatment systems. These systems mainly interfere with the mitosis process of tumor cells by applying an electric field of a specific frequency in the tumor area, thereby achieving the purpose of inhibiting tumor growth or promoting apoptosis. However, most of the current tumor treatment electric field systems use a single electric field mode, and their therapeutic effects are limited by the heterogeneity of tumor cells and the fixedness of electric field parameters. In addition, the lack of precise adaptive adjustment mechanisms for individual differences in patients leads to unstable treatment effects and poor responses in some patients. Summary of the invention

[0003] The present invention aims to provide a multi-mode electric field regulation system for tumor treatment, which integrates multiple electric field modes and intelligent regulation mechanisms to achieve personalized and precise control of tumor treatment. Specifically, the present invention includes the following steps:

[0004] S1. It is designed with a programmable electric field generator that can generate a variety of electric field waveforms including sine wave, pulse wave, square wave, etc. as needed, as well as adjustable frequency, intensity and duration; these parameters can be dynamically adjusted according to tumor type, patient physiological condition and treatment progress;

[0005] S2. Collect tumor data, body data and received electric field therapy parameter data of historical patients' tumors to obtain historical tumor patient data; preprocess the historical tumor patient data to obtain preprocessed historical tumor patient data; select a historical tumor patient training data set and a historical tumor patient test data set from the preprocessed historical tumor patient data, extract features, construct a historical tumor patient training feature matrix and a historical tumor patient test feature matrix, and correspondingly select a historical tumor patient training feature label matrix and a historical tumor patient test feature label matrix;

[0006] S3, establishing an initial neural network model, using the historical tumor patient training matrix and the historical tumor patient training label matrix to train the initial neural network model; after the training is completed, a trained neural network model is obtained, and then the historical tumor patient test matrix and the historical tumor patient test label matrix are input into the trained neural network model for testing, and the trained neural network model is optimized according to the test results to obtain a final neural network model;

[0007] S4. Integrate biosensors to monitor and collect electrophysiological parameters, metabolic status and systemic responses of the tumor area and surrounding tissues in real time to obtain real-time physical data of tumor patients. Preprocess the real-time physical data of tumor patients to obtain preprocessed real-time physical data of tumor patients. Extract features of real-time physical data of tumor patients and construct a feature set of real-time physical data of tumor patients. Transmit the feature set of real-time physical data of tumor patients to the final neural network model and adjust the electric field parameters as needed.

[0008] Preferably, the S1 comprises the following steps:

[0009] S11. Select the desired electric field waveform according to the tumor type and the patient's physiological condition, including sine wave, pulse wave, and square wave;

[0010] S12, pre-setting the frequency, intensity and duration of the electric field, and making real-time adjustments based on feedback during subsequent treatment;

[0011] S13, the electric field generator generates a corresponding electric field waveform according to the set parameters, and transmits it to the patient's tumor area through the electrodes; the system records the electric field parameters used in each treatment for subsequent analysis and adjustment;

[0012] Selecting the electric field waveform based on tumor type and patient physiological condition can ensure that the treatment plan is more in line with individual differences and improve the treatment effect; the electric field parameters can be optimized by pre-setting the frequency, intensity and duration of the electric field and making real-time adjustments based on feedback during treatment.

[0013] Preferably, S2 comprises the following steps:

[0014] S21. Collect data on electrophysiological parameters, metabolic status, and systemic responses of historical patients' tumor areas and surrounding tissues before and after treatment with different electric field parameters to obtain historical tumor patient data, identify and process missing data points in the historical data, and fill them with mean, median, and mode; delete duplicate data records, unify data formats, and obtain pre-processed historical tumor patient data;

[0015] S22, select a training data set A from the pre-processed historical tumor patient data = {A1, A2, ..., A i ,...,A N}、Test data set B={B1,B2,...,B i ,...,B M}; where Ai represents the electric field therapy parameter data and tumor data received by the i-th patient in the training data set, and N represents a total of N patients in the training data set; B irepresents the electric field therapy parameter data and tumor data received by the i-th patient in the test data set, and M represents a total of M patients in the test data set;

[0016] S23, from the historical tumor patient training data set A = {A1, A2, ..., A i ,...,A N}、Historical cancer patient test data set B={B1,B2,...,B i ,...,B N} to extract features and construct a historical tumor patient training feature matrix Constructing a historical tumor patient test feature matrix Construct a historical tumor patient training feature label matrix based on treatment outcomes Historical tumor patient test feature label matrix

[0017] Among them C in represents the nth feature of the electric field therapy parameter data and tumor data received by the i-th patient in the historical tumor patient training feature matrix, D in represents the nth feature of the electric field therapy parameter data and tumor data received by the i-th patient in the historical tumor patient test feature matrix; E im represents the mth label corresponding to the i-th patient in the historical tumor patient training matrix, F im Represents the mth label corresponding to the i-th patient in the historical tumor patient test matrix;

[0018] The above steps can improve the quality and consistency of data, ensure the accuracy and reliability of subsequent analysis, reduce the impact of missing data on analysis results, and maintain data integrity; extract features from historical data and construct a feature matrix, which can provide data for machine learning models and help model learning and prediction; select training data sets and test data sets, which can be used to train machine learning models, and evaluate the performance and generalization ability of the models through test data sets; construct a feature label matrix based on treatment results, which can provide target variables for the model to predict patients' treatment responses and outcomes.

[0019] Preferably, S3 comprises the following steps:

[0020] S31, construct an initial neural network model, set the number of training iterations to a, the batch size to b, the optimizer to adam optimizer, and the initial learning rate to c;

[0021] S32, setting the current number of training iterations to d and the training error threshold to e; according to

[0022] The batch size b, the Adam optimizer, and the initial learning rate c are used to input the historical tumor patient training feature matrix A and the historical tumor patient training feature label matrix B into the initial neural network model for training. When the training error of the initial neural network model is <e or when d≥a, stop training to obtain the trained neural network model;

[0023] S33. Set the accuracy threshold q. Input the historical tumor patient test feature matrix C and the historical tumor patient test feature label matrix D into the trained neural network model for testing to obtain the test accuracy h. When h≥q, use the trained neural network model as the final neural network model. When h<q, return to S32 to continue training the trained neural network model until h≥q to obtain the final neural network model;

[0024] The above steps provide a basic framework for model training, which helps to control the subsequent training process; can control the training process to ensure that the model stops training after reaching a certain performance level to avoid overfitting; can make the model converge to the optimal solution more quickly to improve the training efficiency; can dynamically adjust the training process according to the training error and the number of iterations, and stop training when the performance requirements are met to save computing resources; by inputting the test data set into the trained neural network model for testing, the generalization ability and the actual application effect of the model can be evaluated; setting the accuracy threshold: setting the accuracy threshold helps to ensure that the model is accepted as the final model after reaching a certain accuracy level to guarantee the prediction performance of the model; improve the performance of the model through iterative optimization until the accuracy requirement is met.

[0025] Preferably, S32 includes the following steps:

[0026] S321. The training error of the initial neural network model is calculated using the loss function, and the function expression is as follows:

[0027]

[0028] where x is the number of samples; y i is the true label of the i-th sample, taking values 0 or 1; is the predicted probability of the i-th sample, indicating the probability that the model predicts that the sample belongs to class 1;

[0029] By calculating the training error of the initial neural network model using the loss function, the obtained result is used as an important indicator in the model training process to more accurately evaluate the performance of the model and guide the optimization and adjustment of the model.

[0030] Preferably, S33 includes the following steps:

[0031] S331, construct a particle swarm; set the number of particles in the particle swarm to r, and the particle swarm is represented by g = {g1, g2, ..., g i ,...,g r}, g i represents the i-th particle in the particle swarm; each particle is randomly assigned an initial position and initial velocity, and the position of the particle swarm is initialized to h = {h1, h2, ..., h i ,...,h r}, initialize the velocity of the particle swarm v={v1,v2,...,v i ,...,v r}, initialize the optimal position of each particle

[0032] f={f1,f2,...,f i ,...,f r}; where h i represents the initial position of the i-th particle in the particle swarm, v i represents the initial velocity of the i-th particle in the particle swarm, f i represents the best position reached by the ith particle; initialize the global best position l to any position in f; set the maximum number of iterations to f′, and the performance of the global best position reaches the target value f″;

[0033] S332. For each particle, update its velocity Where t represents the moment when the particle is at the current position, t+1 represents the moment when the particle is at the next position, j is the inertia weight, c1 and c2 are learning factors, and r1 and r2 are random numbers in the range of 0 to 1;

[0034] Update Location Use the updated position as the new learning rate to train the neural network model and evaluate its performance. The performance is better than f i t , then update f i If the new position The performance is not better than f i t , keep the original f i ; At t+1, traverse the historical best positions f of all particles i , and select the position with the best performance as the new global optimal position l t+1 ;

[0035] S333, repeat S332, when the performance of the global best position reaches the target value f″ or the number of iterations reaches f′, stop the iteration process, output the learning rate value corresponding to the global best position, and obtain the optimal learning rate;

[0036] S334, using the optimal learning rate as the learning rate of the trained neural network model to obtain an optimized neural network model;

[0037] The particle swarm optimization algorithm can search in a large range, avoid local optimality, and help find better neural network model parameters; by updating the speed and position of the particles, the algorithm can adaptively adjust the learning rate so that the neural network model can converge better during the training process; in each iteration, each particle in the particle swarm will update its position according to the current global optimal position, and this dynamic adjustment helps to find the optimal learning rate; by optimizing the learning rate, the performance of the neural network model can be improved so that its accuracy on the test set reaches or exceeds the set threshold; the particle swarm optimization algorithm can automatically find the optimal learning rate, and by optimizing the parameters of the neural network model, the generalization ability of the model can be enhanced so that it can maintain good performance on new and unseen data.

[0038] Preferably, S4 comprises the following steps:

[0039] S41, placing the biosensor in the patient's tumor area and surrounding tissues, monitoring and collecting the patient's physical data in real time, and obtaining the real-time physical data of the tumor patient;

[0040] S42, cleaning, denoising, and normalizing the real-time physical data of the tumor patient to obtain pre-processed real-time physical data of the tumor patient;

[0041] S43, extracting features from the pre-processed real-time body data of tumor patients to obtain real-time body data features of tumor patients, and constructing a real-time body data feature set of tumor patients G = {G1, G2, ..., G i ,...,G n};

[0042] S44, transmitting the real-time physical data feature set of the tumor patient to the final neural network model, and obtaining the electric field parameters required by the current tumor patient based on the real-time physical data feature set of the tumor patient. When the electric field parameters need to be adjusted, the system will automatically adjust the parameters of the electric field generator to optimize the treatment effect, ensuring that the electric field treatment always meets the patient's real-time response and needs;

[0043] By placing biosensors in the patient's tumor area and surrounding tissues, the patient's physical data can be monitored in real time. By transmitting the feature set to the final neural network model, the electric field parameters required by the patient can be obtained in real time and the parameters of the electric field generator can be automatically adjusted to optimize the treatment effect. Through real-time monitoring and feedback mechanisms, the treatment plan can be continuously optimized and the electric field parameters can be adjusted in time to respond to changes in the patient's physical condition and improve the safety of treatment.

[0044] A multi-mode based tumor treatment electric field regulation system includes a multi-mode electric field generator module, a data collection and feature extraction module, a model training and optimization module, and a model application module.

[0045] The multi-mode electric field generator module generates electric field related parameters as needed that can be dynamically adjusted according to tumor type, patient physiological condition and treatment progress;

[0046] The data collection and feature extraction module is used to collect historical tumor patient data, extract features after preprocessing, and construct a historical tumor patient feature matrix;

[0047] The model training and optimization module is used to train, test and optimize the initial neural network model to obtain the final neural network model;

[0048] The model application module is used to extract features from tumor patient data collected in real time and substitute the feature set into the final neural network model, and adjust the electric field parameters as needed.

[0049] The present invention has the following beneficial effects:

[0050] The multi-mode electric field generator can selectively interfere with the division of cancer cells and prevent the division of cancer cells. The frequency, intensity and duration of the electric field can be dynamically adjusted for patients with different types of tumors and different physiological conditions to meet different treatment needs. The equipment can monitor the effects of the electric field in real time and adjust parameters according to the progress of treatment to ensure the continued effectiveness of the treatment.

[0051] By preprocessing data and extracting features, noise and outliers can be eliminated and the accuracy of the training model can be improved; training models based on historical patient tumor data can provide cancer patients with electric field therapy plans that better meet their needs.

[0052] By training and optimizing the neural network model, the model can learn what parameters of electric field therapy should be used for patients with tumors with different characteristics, and verify its performance on the test set, thereby improving the accuracy and reliability of the neural network model.

[0053] By real-time monitoring and collecting the electrophysiological parameters, metabolic status and systemic response of the tumor area and surrounding tissues, the real-time physical data features of tumor patients are extracted and a real-time physical data feature set of tumor patients is constructed; the real-time physical data feature set of tumor patients is transmitted to the final neural network model, which can update the parameters of the electric field therapy required by tumor patients in real time, adjust the treatment plan more timely and accurately, and help patients recover.

[0054] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0056] Figure 1 The present invention is a schematic diagram of the process flow of a multi-mode tumor treatment electric field regulation system. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0058] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0059] S1. It is designed with a programmable electric field generator that can generate a variety of electric field waveforms including sine wave, pulse wave, square wave, etc. as needed, as well as adjustable frequency, intensity and duration; these parameters can be dynamically adjusted according to tumor type, patient physiological condition and treatment progress;

[0060] The S1 comprises the following steps:

[0061] S11. Select the desired electric field waveform according to the tumor type and the patient's physiological condition, including sine wave, pulse wave, and square wave;

[0062] S12, pre-setting the frequency, intensity and duration of the electric field, and making real-time adjustments based on feedback during subsequent treatment;

[0063] S13, the electric field generator generates a corresponding electric field waveform according to the set parameters, and transmits it to the patient's tumor area through the electrodes; the system records the electric field parameters used in each treatment for subsequent analysis and adjustment;

[0064] S2. Collect tumor data, body data and received electric field therapy parameter data of historical patients' tumors to obtain historical tumor patient data; preprocess the historical tumor patient data to obtain preprocessed historical tumor patient data; select a historical tumor patient training data set and a historical tumor patient test data set from the preprocessed historical tumor patient data, extract features, construct a historical tumor patient training feature matrix and a historical tumor patient test feature matrix, and correspondingly select a historical tumor patient training feature label matrix and a historical tumor patient test feature label matrix;

[0065] The S2 comprises the following steps:

[0066] S21. Collect data on electrophysiological parameters, metabolic status, and systemic responses of historical patients' tumor areas and surrounding tissues before and after treatment with different electric field parameters to obtain historical tumor patient data, identify and process missing data points in the historical data, and fill them with mean, median, and mode; delete duplicate data records, unify data formats, and obtain pre-processed historical tumor patient data;

[0067] S22, select a training data set A from the pre-processed historical tumor patient data = {A1, A2, ..., A i ,...,A N}、Test data set B={B1,B2,...,B i ,...,B M}; where Ai represents the electric field therapy parameter data and tumor data received by the i-th patient in the training data set, and N represents a total of N patients in the training data set; B i represents the electric field therapy parameter data and tumor data received by the i-th patient in the test data set, and M represents a total of M patients in the test data set;

[0068] S23, from the historical tumor patient training data set A = {A1, A2, ..., A i ,...,A N}、Historical cancer patient test data set B={B1,B2,...,B i ,...,B N} to extract features and construct a historical tumor patient training feature matrix Constructing a historical tumor patient test feature matrix Construct a historical tumor patient training feature label matrix based on treatment outcomes Historical tumor patient test feature label matrix

[0069] Among them C in represents the nth feature of the electric field therapy parameter data and tumor data received by the i-th patient in the historical tumor patient training feature matrix, D inRepresents the nth feature of the electric field therapy parameter data and tumor data received by the ith patient in the historical tumor patient test feature matrix; E im Represents the mth label corresponding to the ith patient in the historical tumor patient training matrix, F im Represents the mth label corresponding to the ith patient in the historical tumor patient test matrix;

[0070] S3. Establish an initial neural network model, and use the historical tumor patient training matrix and the historical tumor patient training label matrix to train the initial neural network model; after the training is completed, obtain the trained neural network model, and then input the historical tumor patient test matrix and the historical tumor patient test label matrix into the trained neural network model for testing, and optimize the trained neural network model according to the test results to obtain the final neural network model;

[0071] The S3 includes the following steps:

[0072] S31. Construct an initial neural network model, and set the number of training iterations to a, the batch size to b, the optimizer to the adam optimizer, and the initial learning rate to c;

[0073] S32. Set the current training iteration number to d and the training error threshold to e; according to

[0074] the batch size b, the adam optimizer, and the initial learning rate c, input the historical tumor patient training feature matrix A and the historical tumor patient training feature label matrix B into the initial neural network model for training. When the training error of the initial neural network model < e or when d ≥ a, stop the training to obtain the trained neural network model;

[0075] The S32 includes the following steps:

[0076] S321. The training error of the initial neural network model is calculated using the loss function, and the function expression is as follows:

[0077]

[0078] where x is the number of samples; y i is the true label of the ith sample, taking values 0 or 1; is the predicted probability of the ith sample, indicating the probability that the model predicts that the sample belongs to class 1;

[0079] S33. Set the accuracy threshold q, input the historical tumor patient test feature matrix C and the historical tumor patient test feature label matrix D into the trained neural network model for testing to obtain the test accuracy h. When h ≥ q, use the trained neural network model as the final neural network model. When h < q, return to S32 to continue training the trained neural network model until h ≥ q to obtain the final neural network model;

[0080] Preferably, S33 includes the following steps:

[0081] S331. Construct a particle swarm. Set the number of particles in the particle swarm to r, and the particle swarm is represented as g = {g1, g2,..., g i ,..., g r}, where g i represents the i-th particle in the particle swarm. Each particle randomly has an initial position and an initial velocity. Initialize the position of the particle swarm h = {h1, h2,..., h i ,..., h r}, initialize the velocity of the particle swarm v = {v1, v2,..., v i ,..., v r}, and initialize the best position of each particle

[0082] f = {f1, f2,..., f i ,..., f r}; where h i represents the initial position of the i-th particle in the particle swarm, v i represents the initial velocity of the i-th particle in the particle swarm, and f i represents the best position reached by the i-th particle. Initialize the global best position l as any position in f; set the maximum number of iterations to f′, and the performance of the global best position reaches the target value f″;

[0083] S332. For each particle, update its velocity where t represents the moment of the particle at the current position, t + 1 represents the moment of the particle at the next position, j is the inertia weight, c1 and c2 are learning factors, and r1 and r2 are random numbers in the range from 0 to 1;

[0084] Update the position Use the updated position as the new learning rate to train the neural network model and evaluate its performance. If the performance of the new position is better than f i t , then update f i . If the performance of the new position is not better than fi t , keep the original f i ; At t+1, traverse the historical best positions f of all particles i , and select the position with the best performance as the new global optimal position l t+1 ;

[0085] S333, repeat S332, when the performance of the global best position reaches the target value f″ or the number of iterations reaches f′, stop the iteration process, output the learning rate value corresponding to the global best position, and obtain the optimal learning rate;

[0086] S334, using the optimal learning rate as the learning rate of the trained neural network model to obtain an optimized neural network model;

[0087] S4, integrating biosensors, monitoring and collecting electrophysiological parameters, metabolic status and systemic responses of the tumor area and surrounding tissues in real time, obtaining real-time body data of tumor patients, preprocessing the real-time body data of tumor patients, obtaining preprocessed real-time body data of tumor patients; extracting features of real-time body data of tumor patients and constructing feature sets of real-time body data of tumor patients; transmitting the feature sets of real-time body data of tumor patients to the final neural network model, and adjusting electric field parameters as needed;

[0088] Preferably, S4 comprises the following steps:

[0089] S41, placing the biosensor in the patient's tumor area and surrounding tissues, monitoring and collecting the patient's physical data in real time, and obtaining the real-time physical data of the tumor patient;

[0090] S42, cleaning, denoising, and normalizing the real-time physical data of the tumor patient to obtain pre-processed real-time physical data of the tumor patient;

[0091] S43, extracting features from the pre-processed real-time body data of tumor patients to obtain real-time body data features of tumor patients, and constructing a real-time body data feature set of tumor patients G = {G1, G2, ..., G i ,...,G n};

[0092] S44, transmitting the real-time physical data feature set of the tumor patient to the final neural network model, and obtaining the electric field parameters required by the current tumor patient based on the real-time physical data feature set of the tumor patient. When the electric field parameters need to be adjusted, the system will automatically adjust the parameters of the electric field generator to optimize the treatment effect, ensuring that the electric field treatment always meets the patient's real-time response and needs;

[0093] A multi-mode based tumor treatment electric field regulation system includes a multi-mode electric field generator module, a data collection and feature extraction module, a model training and optimization module, and a model application module.

[0094] The multi-mode electric field generator module generates electric field related parameters as needed that can be dynamically adjusted according to tumor type, patient physiological condition and treatment progress;

[0095] The data collection and feature extraction module is used to collect historical tumor patient data, extract features after preprocessing, and construct a historical tumor patient feature matrix;

[0096] The model training and optimization module is used to train, test and optimize the initial neural network model to obtain the final neural network model;

[0097] The model application module is used to extract features from tumor patient data collected in real time and substitute the feature set into the final neural network model, and adjust the electric field parameters as needed.

[0098] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0099] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A multi-mode tumor treatment electric field adjustment method, characterized in that: The following steps are involved: S1. A programmable electric field generator is designed to generate a variety of electric field waveforms including sine wave, pulse wave, square wave, as well as adjustable frequency, intensity and duration; These parameters can be dynamically adjusted according to tumor type, patient physiology, and treatment progress; S2. collecting tumor data, body data, and received electric field therapy parameter data of historical patients' tumors to obtain historical tumor patient data; preprocessing the historical tumor patient data to obtain preprocessed historical tumor patient data; A historical tumor patient training data set and a historical tumor patient test data set are selected from the preprocessed historical tumor patient data, and features are extracted to construct a historical tumor patient training feature matrix and a historical tumor patient test feature matrix, and correspondingly a historical tumor patient training feature label matrix and a historical tumor patient test feature label matrix are selected; S3, establishing an initial neural network model, and training the initial neural network model using a historical tumor patient training matrix and a historical tumor patient training label matrix; After the training is completed, a trained neural network model is obtained, and then the historical tumor patient test matrix and the historical tumor patient test label matrix are input into the trained neural network model for testing, and the trained neural network model is optimized according to the test results to obtain the final neural network model; S4. Integrate biosensors to monitor and collect electrophysiological parameters, metabolic status and systemic responses of the tumor area and surrounding tissues in real time to obtain real-time physical data of tumor patients. Preprocess the real-time physical data of tumor patients to obtain preprocessed real-time physical data of tumor patients. Extract features of real-time physical data of tumor patients and construct a feature set of real-time physical data of tumor patients. Transmit the feature set of real-time physical data of tumor patients to the final neural network model and adjust the electric field parameters as needed.

2. The method for adjusting electric field for tumor treatment based on multi-mode according to claim 1, characterized in that: The S1 comprises the following steps: S11. Select the desired electric field waveform according to the tumor type and the patient's physiological condition, including sine wave, pulse wave, and square wave; S12, pre-setting the frequency, intensity and duration of the electric field, and making real-time adjustments based on feedback during subsequent treatment; S13. The electric field generator generates a corresponding electric field waveform according to the set parameters and transmits it to the patient's tumor area through electrodes; the system records the electric field parameters used in each treatment for subsequent analysis and adjustment.

3. The method for adjusting electric field for tumor treatment based on multi-mode according to claim 1, characterized in that: The S2 comprises the following steps: S21. Collect tumor data, body data and received electric field therapy parameter data of historical patients' tumors to obtain historical tumor patient data, identify and process missing data points in the historical data, and fill them with mean, median and mode; delete duplicate data records, unify data formats, and obtain pre-processed historical tumor patient data; S22, selecting a historical tumor patient training data set and a historical tumor patient test data set from the pre-processed historical tumor patient data; S23. Extract features from the historical tumor patient training dataset and the historical tumor patient test dataset to construct a historical tumor patient training feature matrix; construct a historical tumor patient test feature matrix; construct a historical tumor patient training feature label matrix and a historical tumor patient test feature label matrix according to the treatment results.

4. The method for adjusting electric field for tumor treatment based on multi-mode according to claim 1, characterized in that: The S3 includes the following steps: S31. Construct an initial neural network model, set the number of training iterations as a, the batch size as b, the optimizer as the adam optimizer, and the initial learning rate as c; S32. Set the current training iteration number as d and the training error threshold as e; According to the batch size b, the adam optimizer, and the initial learning rate c, input the historical tumor patient training feature matrix and the historical tumor patient training feature label matrix into the initial neural network model for training. When the training error of the initial neural network model < e or when d ≥ a, stop training to obtain the trained neural network model; S33. Set the accuracy threshold q, input the historical tumor patient test feature matrix and the historical tumor patient test feature label matrix into the trained neural network model for testing to obtain the test accuracy h; When h ≥ q, use the trained neural network model as the final neural network model; When h < q, return to S32 to continue training the trained neural network model until h ≥ q to obtain the final neural network model.

5. The method for adjusting electric field for tumor treatment based on multi-mode according to claim 4, characterized in that: The S32 includes the following steps: S321. The training error of the initial neural network model is calculated using a loss function, and the function expression is as follows: Where x is the sample size; y is the sample size. i is the true label of the i-th sample, which takes the value 0 or 1; is the predicted probability of the i-th sample, indicating the probability that the model predicts that the sample belongs to category 1.

6. The method for adjusting electric field for tumor treatment based on multi-mode according to claim 4, characterized in that: The S33 includes the following steps: S331, construct a particle swarm; set the number of particles in the particle swarm to r, and the particle swarm is represented by g = {g1, g2, ..., g i ,...,g r }, g i represents the i-th particle in the particle swarm; each particle is randomly assigned an initial position and initial velocity, and the position of the particle swarm is initialized to h = {h1, h2, ..., h i ,...,h r }, initialize the velocity of the particle swarm v={v1,v2,...,v i ,...,v r }, initialize the optimal position of each particle f={f1,f2,...,f i ,...,f r }; where h i represents the initial position of the i-th particle in the particle swarm, v i represents the initial velocity of the i-th particle in the particle swarm, f i represents the best position reached by the ith particle; initialize the global best position l to any position in f; set the maximum number of iterations to f′, and the performance of the global best position reaches the target value f″; S332. For each particle, update its velocity Where t represents the moment when the particle is at the current position, t+1 represents the moment when the particle is at the next position, j is the inertia weight, c1 and c2 are learning factors, and r1 and r2 are random numbers in the range of 0 to 1; Update Location Use the updated position as the new learning rate to train the neural network model and evaluate its performance. The performance is better than f i t , then update f i If the new position The performance is not better than f i t , keep the original f i ; At t+1, traverse the historical best positions f of all particles i , and select the position with the best performance as the new global optimal position l t+1 ; S333. Repeat S332. When the performance at the global best position reaches the target value f″ or the number of iterations reaches f′, stop the iteration process, output the learning rate value corresponding to the global best position to obtain the optimal learning rate; S334. Use the optimal learning rate as the learning rate of the trained neural network model to obtain an optimized neural network model.

7. The method for adjusting electric field for tumor treatment based on multi-mode according to claim 1, characterized in that: The S4 includes the following steps: S41. Place the biosensor on the patient's tumor area and surrounding tissues, and monitor and collect the patient's body data in real time to obtain the real-time body data of the tumor patient; S42. Clean, denoise, and normalize the real-time body data of the tumor patient; obtain the preprocessed real-time body data of the tumor patient; S43. Extract features from the preprocessed real-time body data of the tumor patient to obtain the real-time body data features of the tumor patient, and construct a real-time body data feature set of the tumor patient; S44. Transmit the real-time body data feature set of the tumor patient to the final neural network model, and obtain the electric field parameters required by the current tumor patient according to the real-time body data feature set of the tumor patient. When the electric field parameters need to be adjusted, the system will automatically adjust the parameters of the electric field generator to optimize the treatment effect; ensure that the electric field treatment always conforms to the real-time reaction and needs of the patient.

8. A multi-mode tumor treatment electric field adjustment system according to any one of claims 1 to 7, characterized in that: Multi-mode electric field generator module, data collection and feature extraction module, model training and optimization module, model application module; The multi-mode electric field generator module generates electric field related parameters as needed that can be dynamically adjusted according to tumor type, patient physiological condition and treatment progress; The data collection and feature extraction module is used to collect historical tumor patient data, extract features after preprocessing, and construct a historical tumor patient feature matrix; The model training and optimization module is used to train, test and optimize the initial neural network model to obtain the final neural network model; The model application module is used to extract features from tumor patient data collected in real time and substitute the feature set into the final neural network model, and adjust the electric field parameters as needed.

Images