Lymphoma patient survival prediction system and method
Through the combination of particle swarm algorithm and survival prediction neural network model, the problems of incomplete data and insufficient accuracy in the existing survival prediction methods are solved, and higher-precision survival prediction is achieved, providing clinicians with valuable reference basis.
Patent Information
- Application Number
- CN202510134354.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-07
- Publication Date
- 2025-05-13
AI Technical Summary
The underlying data used by existing survival prediction methods is not comprehensive enough and the data accuracy is insufficient, resulting in serious insufficient survival prediction accuracy and cannot provide doctors with valuable reference.
By acquiring and pre-processing clinical examination data of lymphoma patients, a particle swarm algorithm is used to create a data feature extraction method, combining information gain to extract features from the data, and a survival prediction neural network model is constructed to improve the robustness and generalization ability of the model through semi-supervised learning.
It achieves more comprehensive and accurate data extraction, improves the accuracy of survival prediction, provides clinicians with more valuable reference basis, and helps to formulate more effective treatment plans.
Smart Images
Figure CN119993460A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of survival prediction, and in particular to a system and method for predicting survival of lymphoma patients. Background Art
[0002] Survival analysis has been considered an effective tool for studying the impact of prognostic therapeutic interventions. Survival prediction requires drawing corresponding patterns and conclusions from a large amount of clinical data, and then determining the corresponding patient's ability to survive lymphoma in that specific time period. Clinicians often use survival analysis to make screening decisions or prescribe treatment plans, and patients adjust their lifestyles to reduce such risks based on the risk-related information obtained.
[0003] However, even for experts with extensive experience, it is difficult to directly judge the patient's survival probability based on some of his reactions or treatment conditions, so the reliability of the diagnosis cannot be guaranteed; and the existing survival prediction methods are difficult to automate the survival prediction process to help medical staff better formulate subsequent treatment plans for patients. Therefore, there is an urgent need to provide an automated survival prediction method. Summary of the invention
[0004] The present invention provides a lymphoma patient survival prediction system and method, which solves the problem that the underlying data used in the existing survival prediction methods are not comprehensive enough and the data accuracy is insufficient, which seriously affects the calculation complexity and the accuracy of subsequent calculations, and ultimately leads to seriously insufficient survival prediction accuracy and inability to provide valuable reference for doctors.
[0005] The lymphoma patient survival prediction system and method of the present invention specifically include the following technical solutions:
[0006] A method for predicting survival of lymphoma patients comprises the following steps:
[0007] S1. Acquire and preprocess the clinical examination data of lymphoma patients to obtain preprocessed clinical examination data; create a data feature extraction method to extract features from the preprocessed clinical examination data to obtain an optimal data feature set;
[0008] S2. Calculate the information gain of the data features in the optimal data feature set, and by setting a screening threshold, screen out the data features whose information gain is greater than the screening threshold as the sample data for survival prediction; construct a survival prediction neural network model, use the sample data for survival prediction as the input of the survival prediction neural network model, and output the survival prediction value of lymphoma patients.
[0009] Preferably, the S1 specifically includes:
[0010] A data feature extraction method is created based on the particle swarm algorithm. During the implementation of the data feature extraction method, each particle represents a data feature, and the data feature is extracted by continuously updating the position of individual particles.
[0011] Preferably, the S1 specifically includes:
[0012] Based on the probability p ε , select the particle position update method and calculate the next position of the particle:
[0013]
[0014] in, Represents the particle individual X r The next position in the k-dimensional space under the influence of the best individual particle, k∈[1,d]; d represents the spatial dimension; represents the best individual particle X * The position in the k-th dimension space; Represents the particle individual X r The position in the k-th dimensional space; a is the iterative loss factor; α1 and α2 are both random numbers between 0 and 1; b is the logarithmic factor; l is a random number between [-1, 1]; p is the probability of selecting the actual particle position update method.
[0015] Preferably, the S1 specifically includes:
[0016] During the particle movement, a particle is randomly selected from the current particle group to approach. At this time, the next position of the current particle is:
[0017]
[0018] in, Represents the position of a particle randomly selected from the current particle swarm in the k-th dimensional space.
[0019] Preferably, the S1 specifically includes:
[0020] The fitness of the particle swarm is calculated based on the fitness function, and whether to end the iteration is determined according to the fitness, so as to obtain the optimal data feature set.
[0021] Preferably, the S2 specifically includes:
[0022] The survival prediction neural network model includes input layer, mapping layer, pattern layer, loop layer, regression layer and output layer.
[0023] Preferably, the S2 specifically includes:
[0024] The sample data of survival prediction is input as input variables to the input layer. The number of neurons in the input layer is equal to the dimension of the sample data of survival prediction, and the input variable is passed to the mapping layer. The sample data of survival prediction is mapped to the calculation space through the mapping layer to generate the data after dimensionality reduction, and the data after dimensionality reduction is passed to the pattern layer. The number of neurons in the pattern layer is equal to the number of data after dimensionality reduction, and each neuron corresponds to different data after dimensionality reduction. Then the transfer function of each neuron is:
[0025]
[0026] Among them, tf u represents the transfer function of the u-th neuron; D is the input variable; D u is the sample data corresponding to the u-th neuron; σ is the smoothing factor; T represents transposition; the output of the pattern layer is passed to the recurrent layer.
[0027] Preferably, the S2 specifically includes:
[0028] The circulation layer is set with a threshold value. When the output of the circulation layer reaches the threshold value, the output of the circulation layer is passed to the regression layer; the regression layer calculates the regression probability based on the output of the circulation layer, and passes the regression probability to the output layer, and outputs the survival prediction value of lymphoma patients through the output layer.
[0029] Preferably, the S2 specifically includes:
[0030] The loss function is calculated based on the predicted survival value and expected output of lymphoma patients:
[0031] Loss = q(t) × mse(y, Y),
[0032] Among them, Loss represents the loss function; mse represents the mean square error function; q(t) represents the error factor; t represents the number of cycles of the loop layer; y represents the predicted survival value of lymphoma patients; Y represents the expected output; according to the loss function, the parameters in the survival prediction neural network model are modified to predict the survival status of lymphoma patients.
[0033] A lymphoma patient survival prediction system comprises the following parts:
[0034] Information acquisition module, preprocessing module, feature extraction module, feature enhancement module, survival prediction module, difference calculation module and correction module;
[0035] An information acquisition module is used to acquire clinical examination data of lymphoma patients and send the clinical examination data to the preprocessing module through data transmission;
[0036] A preprocessing module, used for preprocessing clinical examination data, and sending the preprocessed clinical examination data to the feature extraction module through data transmission;
[0037] A feature extraction module is used to create a data feature extraction method to extract features from the preprocessed clinical examination data; the extracted data features are sent to the feature enhancement module by means of data transmission;
[0038] The feature enhancement module is used to calculate the information gain of each data feature in the data features, and select the data features whose information gain is greater than the screening threshold as the sample data for survival prediction; and send the sample data for survival prediction to the survival prediction module by means of data transmission;
[0039] The survival prediction module is used to construct a survival prediction neural network model, input the sample data of survival prediction into the survival prediction neural network model, and output the survival prediction value of the lymphoma patient; the survival prediction value of the lymphoma patient is sent to the difference calculation module by data transmission;
[0040] A difference calculation module is used to calculate a loss function based on the predicted survival value of lymphoma patients and the expected output; and send the loss function to the correction module by means of data transmission;
[0041] The correction module is used to correct the parameters in the survival prediction neural network model according to the loss function, and send the corrected parameters in the survival prediction neural network model to the survival prediction module through data transmission.
[0042] The beneficial effects of the technical solution of the present invention are:
[0043] 1. Create a data feature extraction method based on the particle swarm algorithm, and combine information gain to select features from clinical examination data. This not only extracts more important features, but also solves the problems of premature convergence and local optimality of the particle swarm algorithm, thereby screening out more effective data features and achieving data enhancement, providing high-precision underlying data for subsequent calculations.
[0044] 2. Construct a survival prediction neural network model, and improve the robustness and generalization ability of the survival prediction neural network model through semi-supervised learning; solve the imbalance of data categories by calculating the loss function, and realize survival prediction based on deep learning, providing valuable reference for clinicians to formulate treatment plans in order to prolong the patient's survival period. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 This is a structural diagram of the lymphoma patient survival prediction system of the present invention;
[0046] Figure 2The present invention is a flowchart of the method for predicting survival of lymphoma patients. DETAILED DESCRIPTION
[0047] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0048] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0049] The specific scheme of the lymphoma patient survival prediction system and method provided by the present invention is described in detail below with reference to the accompanying drawings.
[0050] See attached Figure 1 , which shows a structural diagram of a lymphoma patient survival prediction system provided by an embodiment of the present invention, the system includes the following parts:
[0051] An information acquisition module 10 , a preprocessing module 20 , a feature extraction module 30 , a feature enhancement module 40 , a survival prediction module 50 , a difference calculation module 60 and a correction module 70 .
[0052] The information acquisition module 10 is used to acquire clinical examination data of lymphoma patients. The information acquisition module 10 sends the clinical examination data to the preprocessing module 20 by means of data transmission;
[0053] The preprocessing module 20 is used to perform preprocessing including data cleaning, integration and reduction on the clinical examination data to obtain preprocessed clinical examination data; the preprocessing module 20 sends the preprocessed clinical examination data to the feature extraction module 30 by means of data transmission;
[0054] The feature extraction module 30 is used to create a data feature extraction method to extract features from the preprocessed clinical examination data; the data feature extraction method can be regarded as N particles continuously updating individual positions until a satisfactory solution is found, that is, the data features are extracted; and a fitness function is set up to calculate the fitness of the particle group, and whether to end the iteration is determined according to the fitness; the feature extraction module 30 sends the extracted data features to the feature enhancement module 40 by means of data transmission;
[0055] The feature enhancement module 40 is used to calculate the information gain of each data feature in the data features, and select the data features whose information gain is greater than the screening threshold set according to the expert experience method as the sample data for survival prediction; the feature enhancement module 40 sends the sample data for survival prediction to the survival prediction module 50 by means of data transmission;
[0056] The survival prediction module 50 is used to construct a survival prediction neural network model, input the sample data of survival prediction into the survival prediction neural network model, and output the survival prediction value of the lymphoma patient; the survival prediction module 50 sends the output of the survival prediction neural network model, i.e., the survival prediction value of the lymphoma patient, to the difference calculation module 60 by means of data transmission;
[0057] The difference calculation module 60 is used to calculate the loss function according to the output of the survival prediction neural network model and the expected output; the difference calculation module 60 sends the loss function to the correction module 70 by means of data transmission;
[0058] The correction module 70 is used to correct the parameters in the survival prediction neural network model according to the loss function. The correction module 70 sends the corrected parameters in the survival prediction neural network model to the survival prediction module 50 through data transmission.
[0059] Refer to the attached Figure 2 , which shows a flow chart of a method for predicting survival of lymphoma patients provided by an embodiment of the present invention, the method comprising the following steps:
[0060] S1. Acquire and preprocess the clinical examination data of lymphoma patients to obtain preprocessed clinical examination data; create a data feature extraction method to extract features from the preprocessed clinical examination data to obtain an optimal data feature set;
[0061] The clinical examination data of lymphoma patients are obtained through the information acquisition module 10, and the clinical examination data include routine blood examination data, patient symptom pathological section examination data, etc. The clinical examination data are preprocessed by the preprocessing module 20 to obtain preprocessed clinical examination data; the data preprocessing method adopts existing technologies, including data cleaning, integration, reduction and other methods.
[0062] The feature extraction module 30 creates a data feature extraction method based on the particle swarm algorithm to extract features from the preprocessed clinical examination data. The specific implementation process of the data feature extraction method is as follows:
[0063] Set up N particles, each particle represents a data feature. The process of data feature extraction method can be regarded as N particles constantly updating their individual positions until a satisfactory solution is found, that is, the data feature is extracted.
[0064] Assume that in the d-dimensional space, the current best particle individual X * The location is Particle Individual X r The location is Then the particle individual X r The next position under the influence of the best particle individual The calculation formula of the first particle position update method is:
[0065]
[0066] in, Represents the particle individual X r The next position in the k-th dimension space under the influence of the best individual particle, k∈[1,d]; represents the best individual particle X * The position in the k-th dimensional space; Represents the particle individual X r The position in the k-th dimensional space; a is the iterative loss factor, which decreases with the increase of the number of iterations; α1 and α2 are both random numbers between 0 and 1.
[0067] By optimizing the position update strategy, i.e. the second particle position update method, the particle is moved from the current position to the optimal particle position. By reducing the value of the iterative loss factor a, a(2α1-1) is made to fluctuate within the interval [-1, 1]; then the next position of the particle is updated, and the current position of the particle is set to X r , then the calculation of the next position is:
[0068]
[0069] Where b is a logarithmic factor used to limit the path of particle movement; l is a random number between [-1, 1].
[0070] Based on the probability p ε , select the particle position update method, so the next position of the particle is:
[0071]
[0072] Among them, p is the probability of selecting the two actual particle position update methods.
[0073] During the particle movement, it may not approach the best individual particle, but randomly select a particle from the current particle group to approach. Although this operation may cause the current individual particle to deviate from the best particle position, it will also enhance the global search ability of the particle group. At this time, the next position of the current particle is:
[0074]
[0075] in, Represents the position of a particle randomly selected from the current particle swarm in the k-th dimensional space.
[0076] A fitness function is established to calculate the fitness of the particle swarm, and whether to end the iteration is determined based on the fitness. The specific formula of the fitness function is:
[0077]
[0078] Among them, fit represents fitness and dis represents the distance between positions. If the fitness is greater than the threshold set according to the expert experience method, the iteration is stopped to obtain the optimal data feature set.
[0079] S2. Calculate the information gain of the data features in the optimal data feature set, and by setting a screening threshold, screen out the data features whose information gain is greater than the screening threshold as the sample data for survival prediction; construct a survival prediction neural network model, use the sample data for survival prediction as the input of the survival prediction neural network model, and output the survival prediction value of lymphoma patients.
[0080] The optimal data feature set is represented as D = {D1, D2, ..., D n}, where n represents the type of data feature, and any data feature in the optimal data feature set is represented as D i , i∈[1,n], Wherein, m indicates that the ith data feature has m possible values. The information gain of each data feature in the optimal data feature set is calculated by the feature enhancement module 40:
[0081]
[0082] Among them, IG i represents the information gain of the i-th data feature; p i represents the proportion of the i-th data feature in the optimal data feature set; |D i | represents the number of the i-th data feature; N is the total number of data features; p j It represents the proportion of the j-th value of the i-th data feature in all the values of the i-th data feature, j∈[1,m].
[0083] A screening threshold is set in advance according to the expert experience method, and data features with information gain greater than the screening threshold are selected from the optimal data feature set as sample data for survival prediction. A survival prediction neural network model is constructed through the survival prediction module 50, and the sample data is input into the survival prediction neural network model to output the survival prediction value of the lymphoma patient; through deep learning and multiple trainings, a survival prediction neural network model with the desired accuracy is finally obtained.
[0084] The survival prediction neural network model includes input layer, mapping layer, pattern layer, loop layer, regression layer, and output layer. The specific implementation steps are as follows:
[0085] Input layer: Sample data D = {D1, D2, ..., D M} is input as an input variable to the input layer. The number of neurons in the input layer is equal to the dimension M of the input sample data. Each neuron is a simple distribution unit. The input variable is directly passed to the mapping layer.
[0086] Mapping layer: maps sample data to the computational space to generate reduced-dimensional data; reduces data dimension and computational complexity through mapping processing; the mapping layer passes the reduced-dimensional data to the pattern layer;
[0087] Pattern layer: The number of neurons in the pattern layer is equal to the number of data after dimensionality reduction. Each neuron corresponds to different data after dimensionality reduction. The transfer function of the u-th neuron is:
[0088]
[0089] Among them, tf u represents the transfer function of the u-th neuron; D is the input variable; D u is the sample data corresponding to the u-th neuron; σ is the smoothing factor; T represents transposition; the pattern layer passes the calculated data to the recurrent layer;
[0090] Circulation layer: The circulation layer has a threshold value. When the calculation result reaches the threshold value, it can continue to be passed down, otherwise it will continue to circulate; the calculation formula of the circulation layer is:
[0091] h v =f(U×tf u +Wh v-1 +ζ),
[0092] cl v =f(V×h v +c),
[0093] Among them, h v is the hidden state of the vth neuron in the recurrent layer; the three matrices U, W, and V are the linear relationship parameters of the recurrent layer; h v-1 is the hidden state of the v-1th neuron in the recurrent layer; f is the activation function; ζ is the bias of the linear relationship; cl v is the output of the loop layer; c is a constant coefficient. v When the threshold is reached, the output is passed to the regression layer.
[0094] Regression layer: The regression probability expression is:
[0095]
[0096] Where P represents the regression probability; cl 1：R represents the set {cl1,…,cl v ,…,cl R}, v∈[1,R]; R represents the number of neurons in the recurrent layer; ω v Represents the weight of the vth neuron. The regression layer passes the calculation results to the output layer;
[0097] Output layer: The output result of the output layer is:
[0098] y=f(ωP+B),
[0099] Among them, y represents the output of the survival prediction neural network model, that is, the predicted survival value of lymphoma patients; ω represents the connection weight; and B represents the bias of the output layer.
[0100] The difference calculation module 60 calculates the loss function according to the output y and the expected output Y of the survival prediction neural network model:
[0101] Loss = q(t) × mse(y, Y),
[0102]
[0103] Wherein, Loss represents the loss function; mse represents the mean square error function; q(t) represents the error factor; t represents the number of cycles of the loop layer; T1, T2 and δ are all hyperparameters of the survival prediction neural network model. If the loss function value does not reach the accuracy preset according to the expert experience method, the correction module 70 corrects the parameters in the survival prediction neural network model according to the loss function, and finally completes the training of the survival prediction neural network model, thereby predicting the survival status of lymphoma patients.
[0104] In summary, the lymphoma patient survival prediction system and method are completed.
[0105] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0106] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
[0107] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.
Claims
1. A method for predicting survival of lymphoma patients, characterized in that: The following steps are involved: S1. Acquire and preprocess the clinical examination data of lymphoma patients to obtain preprocessed clinical examination data; create a data feature extraction method to extract features from the preprocessed clinical examination data to obtain an optimal data feature set; S2. Calculate the information gain of the data features in the optimal data feature set, and by setting a screening threshold, screen out the data features whose information gain is greater than the screening threshold as the sample data for survival prediction; construct a survival prediction neural network model, use the sample data for survival prediction as the input of the survival prediction neural network model, and output the survival prediction value of lymphoma patients.
2. The method for predicting survival of lymphoma patients according to claim 1, characterized in that: The S1 specifically includes: A data feature extraction method is created based on the particle swarm algorithm. During the implementation of the data feature extraction method, each particle represents a data feature, and the data feature is extracted by continuously updating the position of individual particles.
3. The method for predicting survival of lymphoma patients according to claim 2, characterized in that: The S1 specifically includes: Based on the probability p ε , select the particle position update method and calculate the next position of the particle: in, Represents the particle individual X r The next position in the k-dimensional space under the influence of the best individual particle, k∈[1,d]; d represents the spatial dimension; represents the best individual particle X * The position in the k-th dimension space; Represents the particle individual X r The position in the k-th dimensional space; a is the iterative loss factor; α1 and α2 are both random numbers between 0 and 1; b is the logarithmic factor; l is a random number between [-1, 1]; p is the probability of selecting the actual particle position update method.
4. The method for predicting survival of lymphoma patients according to claim 3, characterized in that: The S1 specifically includes: During the particle movement, a particle is randomly selected from the current particle group to approach. At this time, the next position of the current particle is: in, Represents the position of a particle randomly selected from the current particle swarm in the k-th dimensional space.
5. The method for predicting survival of lymphoma patients according to claim 4, characterized in that: The S1 specifically includes: The fitness of the particle swarm is calculated based on the fitness function, and whether to end the iteration is determined according to the fitness, so as to obtain the optimal data feature set.
6. The method for predicting survival of lymphoma patients according to claim 1, characterized in that: The S2 specifically includes: The survival prediction neural network model includes input layer, mapping layer, pattern layer, loop layer, regression layer and output layer.
7. The method for predicting survival of lymphoma patients according to claim 6, characterized in that: The S2 specifically includes: The sample data of survival prediction is input as input variables to the input layer. The number of neurons in the input layer is equal to the dimension of the sample data of survival prediction, and the input variable is passed to the mapping layer. The sample data of survival prediction is mapped to the calculation space through the mapping layer to generate the data after dimensionality reduction, and the data after dimensionality reduction is passed to the pattern layer. The number of neurons in the pattern layer is equal to the number of data after dimensionality reduction, and each neuron corresponds to different data after dimensionality reduction. Then the transfer function of each neuron is: Among them, tf u represents the transfer function of the u-th neuron; D is the input variable; D u is the sample data corresponding to the u-th neuron; σ is the smoothing factor; T represents transposition; the output of the pattern layer is passed to the recurrent layer.
8. The method for predicting survival of lymphoma patients according to claim 7, characterized in that: The S2 specifically includes: The circulation layer is set with a threshold value. When the output of the circulation layer reaches the threshold value, the output of the circulation layer is passed to the regression layer; the regression layer calculates the regression probability based on the output of the circulation layer, and passes the regression probability to the output layer, and outputs the survival prediction value of lymphoma patients through the output layer.
9. The method for predicting survival of lymphoma patients according to claim 8, characterized in that: The S2 specifically includes: The loss function is calculated based on the predicted survival value and expected output of lymphoma patients: Loss = q(t) × mse(y, Y), Among them, Loss represents the loss function; mse represents the mean square error function; q(t) represents the error factor; t represents the number of cycles of the loop layer; y represents the predicted survival value of lymphoma patients; Y represents the expected output; according to the loss function, the parameters in the survival prediction neural network model are modified to predict the survival status of lymphoma patients.
10. A lymphoma patient survival prediction system, applied to the lymphoma patient survival prediction method according to any one of claims 1 to 9, characterized in that: Includes the following sections: Information acquisition module, preprocessing module, feature extraction module, feature enhancement module, survival prediction module, difference calculation module and correction module; An information acquisition module is used to acquire clinical examination data of lymphoma patients and send the clinical examination data to the preprocessing module through data transmission; A preprocessing module, used for preprocessing clinical examination data, and sending the preprocessed clinical examination data to the feature extraction module through data transmission; A feature extraction module is used to create a data feature extraction method to extract features from the preprocessed clinical examination data; the extracted data features are sent to the feature enhancement module by means of data transmission; The feature enhancement module is used to calculate the information gain of each data feature in the data features, and select the data features whose information gain is greater than the screening threshold as the sample data for survival prediction; and send the sample data for survival prediction to the survival prediction module by means of data transmission; A survival prediction module is used to construct a survival prediction neural network model, input sample data of survival prediction into the survival prediction neural network model, and output the survival prediction value of lymphoma patients; and send the survival prediction value of lymphoma patients to the difference calculation module by means of data transmission; A difference calculation module is used to calculate a loss function based on the predicted survival value of lymphoma patients and the expected output; and send the loss function to the correction module by means of data transmission; The correction module is used to correct the parameters in the survival prediction neural network model according to the loss function, and send the corrected parameters in the survival prediction neural network model to the survival prediction module through data transmission.