Orthognathic surgery risk assessment system based on machine learning and data driving

Abstract

The invention provides an orthognathic surgery risk assessment system based on machine learning and data driving, and relates to the technical field of medical and artificial intelligence systems.The orthognathic surgery risk assessment system comprises a data acquisition module used for acquiring orthognathic surgery data of a to-be-assessed patient, and a risk assessment system, the data expansion module is used for inputting the preprocessed orthognathic operation data of the patient into a risk classification model, and in the risk classification model, a generative adversarial network based on circulation consistency is adopted to carry out data expansion on the orthognathic operation data; performing feature extraction on the expanded data by adopting a neural network based on dynamic adaptive oscillation; inputting the extracted high-dimensional features into an auto-encoder based on potential space mapping for feature dimension reduction; and inputting the dimensionality-reduced feature representation into a support vector machine based on fuzzy constraint, processing uncertainty and noise in the orthognathic surgery data feature representation through a fuzzy membership function and an adaptive quantization perturbation mechanism, and outputting a final orthognathic surgery risk assessment auxiliary classification result.

Description

Technical Field

[0001] The present disclosure relates to the technical field of medicine and artificial intelligence systems, and in particular to an orthognathic surgery risk assessment system based on machine learning and data-driven. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Orthognathic surgery is a complex surgical procedure to correct facial skeletal deformities. Its goal is to improve the patient's facial function and appearance while improving their quality of life. However, orthognathic surgery involves deep adjustments to the facial anatomy, which is high-risk and highly individualized. Therefore, a comprehensive and scientific assessment of the patient's risk level is required before surgery. Effective risk assessment can not only help doctors develop personalized surgical plans, but also reduce surgical risks, improve postoperative recovery efficiency and patient satisfaction.

[0004] In recent years, machine learning methods have been gradually applied to the medical field, providing a powerful tool for solving complex problems. In the risk assessment of orthognathic surgery, by establishing a data-driven prediction model, it is expected to extract key information from multidimensional data and perform accurate classification.

[0005] The following problems still exist in existing machine learning methods: 1) Generative adversarial networks are used to expand orthognathic surgery data. However, relying solely on adversarial loss, the generated orthognathic surgery data often deviates from the real orthognathic surgery data distribution, especially when complex high-order features are involved, and the authenticity of the generated samples cannot be guaranteed; 2) Traditional fully connected neural networks are prone to gradient vanishing or gradient exploding problems when extracting features, and the optimization process may fall into local optimality, resulting in unstable or incomplete extracted features; 3) Existing dimensionality reduction methods are mostly based on fixed transformation or mapping, lacking in-depth mining of the internal structure of orthognathic surgery data. The expressive power of features after dimensionality reduction is limited, and it is difficult to fully retain the key information in the original orthognathic surgery data. 4) Traditional support vector machines cannot effectively handle fuzzy and noisy data during the classification process, and the classification results are easily affected by outliers and data uncertainty, resulting in a decrease in classification performance. Summary of the invention

[0006] In order to solve the above problems, the present invention proposes an orthognathic surgery risk assessment system based on machine learning and data-driven, which expands the orthognathic surgery data through a cycle-consistent generative adversarial network, uses a dynamic adaptive oscillation algorithm to extract the features of the intrinsic structure and pattern of the orthognathic surgery data, and dynamically adjusts the feature weights of the intrinsic structure and pattern of the orthognathic surgery data through an autoencoder based on latent space mapping to achieve feature dimensionality reduction; then, a support vector machine based on fuzzy constraints is used to process the uncertainty and noise in the orthognathic surgery data to achieve risk classification, enhance the robustness of the classifier and improve the classification accuracy.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions: Orthognathic surgery risk assessment system based on machine learning and data-driven, including: A data acquisition module, configured to acquire and pre-process orthognathic surgery data of a patient to be evaluated; A risk assessment system is configured to input the pre-processed orthognathic surgery data of the patient into a risk classification model to obtain an auxiliary classification result of orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0008] According to some embodiments, the present disclosure adopts the following technical solutions: A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method described below is implemented: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0009] According to some embodiments, the present disclosure adopts the following technical solutions: A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the following method is implemented: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0010] According to some embodiments, the present disclosure adopts the following technical solutions: An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device performs the following method: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0011] Compared with the prior art, the present invention has the following beneficial effects: The orthognathic surgery risk assessment system based on machine learning and data-driven disclosed in the present invention adopts a generative adversarial network based on cycle consistency to expand the orthognathic surgery data, and utilizes cycle consistency loss and low-rank tensor decomposition optimization to solve the problem that the orthognathic surgery data generated by the traditional generative adversarial network cannot fully match the real orthognathic surgery data distribution and the high-order feature information is lost; the generative adversarial network based on cycle consistency can maintain the consistency of the generated orthognathic surgery data with the real orthognathic surgery data distribution when expanding the orthognathic surgery data, while effectively retaining the high-order features, thereby improving the authenticity and diversity of the orthognathic surgery data expansion.

[0012] The orthognathic surgery risk assessment system based on machine learning and data-driven disclosed in the present invention adopts a neural network based on dynamic adaptive oscillation for feature extraction, and solves the problems of gradient vanishing, gradient explosion and local optimal solution in the training of fully connected neural networks by dynamically adjusting the oscillation frequency, phase difference and amplitude, thereby improving the stability of feature extraction; the dynamic adaptive oscillation algorithm enables the feature extraction process to be more fully optimized in the high-dimensional parameter space, avoids training instability, and improves the expression ability of the extracted features.

[0013] The orthognathic surgery risk assessment system based on machine learning and data-driven disclosed in the present invention adopts an autoencoder based on latent space mapping to perform feature dimensionality reduction. Through feature adaptive optimization and entropy regularization strategies, it solves the problems of difficulty in capturing the intrinsic structure of orthognathic surgery data and low feature quality after dimensionality reduction in traditional dimensionality reduction algorithms. The autoencoder based on latent space mapping significantly reduces the feature dimension by capturing the intrinsic pattern of orthognathic surgery data and dynamically adjusting feature weights, while retaining key information and improving the feature utilization rate after dimensionality reduction.

[0014] The orthognathic surgery risk assessment system based on machine learning and data-driven disclosed in the present invention adopts a support vector machine based on fuzzy constraints in the classification stage, and solves the problem of reduced classification performance of traditional classifiers when facing noisy and ambiguous orthognathic surgery data through fuzzy membership functions and adaptive quantization perturbation mechanisms; the support vector machine based on fuzzy constraints can accurately handle the uncertainty and noise in orthognathic surgery data during the classification process, thereby enhancing the robustness of the classifier and improving the classification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0016] Figure 1 The training process of the neural network based on dynamic adaptive oscillation according to the embodiment of the present disclosure; Figure 2 The present invention is a flowchart of the training process of the risk classification model of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0017] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0018] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0019] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0020] Example 1 In one embodiment of the present disclosure, a risk assessment system for orthognathic surgery based on machine learning and data-driven is provided, comprising: A data acquisition module, configured to acquire and pre-process orthognathic surgery data of a patient to be evaluated; A risk assessment system is configured to input the pre-processed orthognathic surgery data of the patient into a risk classification model to obtain an auxiliary classification result of orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0021] As an embodiment, the training process of the risk classification model in the orthognathic surgery risk assessment system based on machine learning and data-driven disclosed includes data collection and annotation, data expansion, training feature extraction model, training feature dimension reduction model and training classifier model, and then using the trained risk classification model to perform orthognathic surgery risk assessment. Specifically, the risk classification model training process is as follows: S1: Collection and annotation of orthognathic surgery training data In this embodiment, the orthognathic surgery training data is the patient's diagnosis and treatment process data before and after the operation, the patient's medical history and postoperative recovery status. All orthognathic surgery data are stored in a structured JSON format. The attributes of the orthognathic surgery data include: Ra is the patient's age, represented by an integer; Da is the gender, represented by a character, M represents male and F represents female; Pa is the preoperative diagnosis result, given in the form of a string list; Sa is the type of surgery, identified by multiple possible surgery categories; Ta is the duration of the surgery, an integer in minutes; Ua is the pain score before and after the surgery, using a scoring system of 0 to 10; Va is the postoperative recovery speed, represented by the number of days; Wa is the type of postoperative complications, represented by a string list; Xa is the patient satisfaction score, using a star rating system of 1 to 5; Ya is the postoperative facial symmetry evaluation, using specific quantitative standards and scoring methods.

[0022] Preferably, in practical applications, the attributes of orthognathic surgery data are usually more than 10 attributes, and the number of attributes of the data may reach dozens or even hundreds.

[0023] Furthermore, the collected orthognathic surgery data are labeled, and the labeling method disclosed in the present invention is manual labeling, wherein the labeling categories include "low risk", "medium risk" and "high risk".

[0024] S2: Orthognathic surgery training data expansion The acquisition, labeling and preprocessing of orthognathic surgery training data are time-consuming and labor-intensive, and insufficient training samples can easily lead to poor model generalization ability, while affecting the accuracy of the model. The present invention adopts a generative adversarial network based on cycle consistency to generate samples, thereby achieving orthognathic surgery data expansion. On the basis of the traditional generative adversarial network, the present invention adopts a cycle consistency loss function to ensure the consistency of the generated orthognathic surgery data with the real orthognathic surgery data in overall distribution. The discriminator not only needs to judge the authenticity of the orthognathic surgery data, but also needs to calculate the transformation consistency of the orthognathic surgery data before and after the transformation, thereby guiding the generator to generate more realistic orthognathic surgery data, including: S201: Initialize the parameters of the generator and discriminator of the generative adversarial network. Suppose the parameters of the generator and discriminator are and ,in are the parameters of the generator, is the parameter of the discriminator, and the generator is responsible for extracting the orthognathic surgery source domain data Generate orthognathic surgery target domain data. The discriminator function is used to distinguish the real orthognathic surgery target domain data from the generated orthognathic surgery target domain data, which is expressed as: (1) (2) In the formula, Orthognathic surgery data generated for the generator; is the probability output by the discriminator, indicating whether the orthognathic surgery data generated by the generator is real; is the generator function; is the discriminator function.

[0025] S202: The generator generates fake orthognathic surgery target domain data by learning features from orthognathic surgery source domain data. During the training process, the generator continuously adjusts its weights to minimize the difference between the generated orthognathic surgery data and the orthognathic surgery target domain data. The difference between the generator output and the real target data is measured by the standard adversarial loss function, which is expressed as: (3) In the formula, represents the distribution of orthognathic surgery source domain data, It is the orthognathic surgery source domain data; Indicates compliance with a specific distribution; It is the target domain data of orthognathic surgery; express expectations; is the adversarial loss function of the generator.

[0026] Furthermore, the output of the discriminator is calculated through the Bayesian inference model, and the calculation method is expressed as: (4) In the formula, is the Sigmoid activation function.

[0027] Furthermore, in order to ensure that the conversion of the generated orthognathic surgery data between the target domain and the source domain is consistent, the generator needs to learn an inverse process, that is, to generate and restore the orthognathic surgery source domain data from the target domain. The present disclosure uses a cycle consistency loss to calculate the error of the generated orthognathic surgery data from the target domain back to the source domain, which is expressed as: (5) In the formula, is the cycle consistency loss function; represents the inverse process of the generator, i.e. mapping the generated orthognathic surgery data in the target domain back to the source domain; represents the inverse matrix of the generator parameter matrix; express Norm, used to measure the restoration error.

[0028] Furthermore, the inverse process of the generator can be assumed to be a combination of linear mapping and nonlinear adjustment, expressed as: (6) In the formula, and are the learning parameters of the inverse generator, representing the weight matrix and bias in the inverse mapping process respectively.

[0029] Furthermore, the calculation method of the total loss function of the generator is expressed as: (7) In the formula, is the weight hyperparameter of the cycle consistency loss; is the total loss function of the generator; is the low-rank tensor decomposition loss function; is the weight hyperparameter of the low-rank tensor decomposition loss. Preferably, Set to 0.2, Set to 0.3.

[0030] S203: In traditional generative adversarial networks, the learning process of the generator and the inverse generator often focuses on direct linear mapping. During the orthognathic surgery data distribution conversion in the present disclosure, simple linear mapping cannot effectively retain the high-order structural information of the orthognathic surgery data, resulting in a decrease in the accuracy of the inverse mapping result. The present disclosure utilizes a generator inverse mapping optimization strategy based on tensor low-rank approximation, so that the generator can better retain the high-order feature information of the orthognathic surgery data during the orthognathic surgery data conversion process, while improving the accuracy of the inverse mapping. Specifically, the key to the tensor low-rank approximation is to decompose the high-dimensional orthognathic surgery data into a low-rank tensor representation, that is, to reconstruct it by low-rank tensor decomposition to obtain a more effective mapping. For the orthognathic surgery data generated in the target domain, the optimization process of the inverse generator is expressed as: (8) In the formula, represents the low-rank tensor decomposition operation, is the rank of the tensor. Low-rank tensor decomposition projects the generated orthognathic surgery data into a lower dimensional space and retains the main structural information of the orthognathic surgery data. is the eigenvalue of the tensor decomposition, indicating the The contribution weight of the dimension; and are the left and right singular vectors of the tensor, respectively, representing the underlying structure of the orthognathic surgery data.

[0031] Furthermore, the goal of inverse generator optimization is to minimize the reconstruction error between the generated orthognathic surgery data and the inverse mapping orthognathic surgery data after low-rank approximation, and to measure the difference between the inverse mapping of the target data after low-rank tensor decomposition and the orthognathic surgery source domain data. The goal is to minimize this error. The calculation method of the low-rank tensor decomposition loss function is expressed as: (9) In the formula, represents the Frobenius norm, that is, the square root of the sum of the squares of the elements of the matrix, which represents the reconstruction error; Represents the distribution of orthognathic surgery target domain data.

[0032] S204: The task of the discriminator is to judge the authenticity of the orthognathic surgery data and to evaluate whether the orthognathic surgery data generated by the generator conforms to the characteristics of the orthognathic surgery target domain data. The goal of the discriminator is to minimize the discrimination error between the real orthognathic surgery data and the generated orthognathic surgery data. The adversarial loss function of the discriminator is expressed as: (10) In the formula, Represents the distribution of orthognathic surgery target domain data; is the adversarial loss function of the discriminator, that is, the total loss function of the discriminator.

[0033] S205: By alternately training the generator and the discriminator, the generator can continuously improve the quality of generated samples, while the discriminator improves its ability to judge the authenticity of orthognathic surgery data. After each round of training, the generator will update its weights based on the feedback from the discriminator, and the discriminator will update its judgment criteria based on new samples.

[0034] Repeat the above steps until the preset stop iteration condition is met, which means that the model training is completed. In one embodiment, the preset stop iteration condition is reaching a preset maximum number of iterations. Preferably, the preset maximum number of iterations is set to 1000 times.

[0035] After the orthognathic surgery data expansion model is trained, the number of samples is increased using the trained orthognathic surgery data expansion model. In one embodiment, assuming that the original collected samples are 800, and the orthognathic surgery data expansion model generates 200 samples by expansion, the expanded orthognathic surgery data set contains 1000 samples.

[0036] S3: Training feature extraction model The feature extraction model is trained using the expanded orthognathic surgery data. The present disclosure uses a 6-layer fully connected neural network for feature extraction. The number of nodes in each layer of the fully connected layer is 128, that is, the layer contains 128 neurons. In the prior art, some solutions use neural networks for feature extraction. In certain neural network structures, problems such as gradient vanishing, gradient explosion, or falling into local optimal solutions may occur, affecting the stability of training and the performance of the model. The present disclosure uses a neural network based on dynamic adaptive oscillation as a feature extraction model. By simulating the nonlinear oscillation behavior in physical phenomena, the algorithm can effectively explore and utilize local extreme values ​​in high-dimensional parameter space, thereby achieving the purpose of optimizing the neural network. Figure 1 As shown, including: S301: Initialize the model and adaptive oscillation hyperparameters, initialize the parameters of the neural network, and initialize the hyperparameters of the dynamic adaptive oscillation, including the initial phase of the oscillation and the initial amplitude of the oscillation. In one embodiment, the initialization method is expressed as:

[0037]

[0038]

[0039] (11) In the formula, Indicates compliance with a specific distribution; It means the mean is 0 and the variance is Normal distribution of represents normal distribution; Represents the initialization variance of the neural network; is the initial weight of the neural network, is the initial bias of the neural network; is the initial phase of oscillation; is the initial amplitude of oscillation. Preferably, Set to 0.01.

[0040] S302: Calculate the oscillation frequency of each parameter based on the current loss function. The adjustment of the oscillation frequency depends on the local curvature estimation of the loss surface. The adjustment is made using the historical gradient information to make the parameter more Smoothness and more efficient search in parameter space are possible. The adjustment method is expressed as: (12) In the formula, is the historical gradient weight factor, which controls the influence of the historical gradient on the current frequency adjustment; is the oscillation frequency, For the The oscillation frequency of the iteration; For the The weights of the neural network at the iteration; For the The weights of the neural network at the iteration; is the loss function of the neural network; is the symbol of partial derivative; is the fundamental oscillation frequency; is a hyperparameter that adjusts the oscillation responsiveness.

[0041] Preferably, Set to 5, Set to 0.3; the loss function of the neural network uses cross entropy loss, which is calculated by the preset Softmax on the output of the last layer of the neural network.

[0042] S303: The update of each parameter is also affected by the phase difference, which is determined by the success rate of the previous update and the interaction strength between the parameters. The phase update depends on the effect of the previous parameter update, which is used to simulate the delayed effect of causality. The update method is expressed as: (13) In the formula, is the phase of the oscillation, For the The phase of the oscillation of the iteration; For the The phase of the oscillation of the iteration; is the phase adjustment factor; It is a hyperbolic tangent function, which can limit the adjustment range of the phase and avoid excessive adjustment that may cause instability in the algorithm. is a hyperparameter that adjusts phase sensitivity. Preferably, Set to 0.1, Set to 0.3.

[0043] S304: Automatically adjust the amplitude according to the effect of parameter update in the past few iterations. If the update of a certain parameter continues to reduce the loss, increase its amplitude; otherwise, reduce the amplitude. The adaptive adjustment method of the amplitude is expressed as: (14) In the formula, is the amplitude of the oscillation, It is The amplitude of the oscillation of the iteration; It is The amplitude of the oscillation of the iteration; is the amplitude adjustment coefficient. Preferably, Set to 0.95.

[0044] S305: Calculate the updated value of the parameters of each neural network. Specifically, the parameters are updated in combination with the oscillation behavior. The sine function and cosine function are used to simulate the oscillation behavior. The parameters are allowed to be periodically explored in the gradient direction of the loss function. The update method is expressed as: (15) (16) In the formula, For the The weights of the neural network at the iteration; For the The bias of the neural network at this iteration; For the The weights of the neural network at the iteration; For the The bias of the neural network at this iteration; is the current iteration number of the neural network.

[0045] S306: Repeat the above steps until the preset stop iteration condition is met, and output the high-dimensional feature vector after feature extraction , indicating that the model training is completed.

[0046] In one embodiment, the preset condition for stopping iteration is reaching a preset maximum number of iterations. Preferably, the preset maximum number of iterations is set to 1000 times.

[0047] S4: Training feature dimensionality reduction model The high-dimensional features of the orthognathic surgery data after feature extraction are input into the feature dimensionality reduction model to train the feature dimensionality reduction model. The present disclosure adopts an autoencoder based on latent space mapping as a feature dimensionality reduction model. The latent space mapping refers to the encoder mapping the input high-dimensional features to a low-dimensional latent space representation through column nonlinear transformation. Using the feature adaptive optimization mechanism, the encoder parameters will be dynamically adjusted according to the statistical characteristics and correlation of the input features to better capture the inherent structure and pattern of the orthognathic surgery data after feature extraction. Specifically, the encoder converts the input high-dimensional features into a low-dimensional latent representation, and the decoder reconstructs the original features from the latent representation. On the basis of the traditional autoencoder algorithm, the present disclosure adopts a feature weight adaptive adjustment strategy so that the model can focus on the features that have a greater impact on the dimensionality reduction results, thereby improving the quality of the orthognathic surgery data after dimensionality reduction. Including: S401: Initialize the parameters of the autoencoder network. is the weight matrix of the encoder, is the encoder bias vector, is the weight matrix of the decoder, is the bias vector of the decoder. In one embodiment, the parameters of the autoencoder network are initialized randomly.

[0048] S402: Input the high-dimensional feature vector after feature extraction into the encoder. Suppose the high-dimensional feature vector input into the encoder is expressed as: (17) In the formula, is the high-dimensional feature vector input to the encoder; For the The feature vector of the samples, is the total number of samples.

[0049] Furthermore, the encoder maps the input high-dimensional features to a low-dimensional latent space representation through nonlinear transformation. The present disclosure adopts a feature adaptive optimization mechanism, and the parameters of the encoder are dynamically adjusted according to the statistical characteristics and correlation of the input features to better capture the intrinsic structure and pattern of the orthognathic surgery data after feature extraction, which is expressed as: (18) In the formula, is a low-dimensional representation of the latent space; is the mapping function of the encoder; is an adaptive activation function.

[0050] Furthermore, , the calculation method of the adaptive activation function is expressed as: (19) In the formula, is the input of the adaptive activation function, is the feature adaptive parameter; is a dynamic weight coefficient with a value range of [0, 1], which is used to balance the influence of the two activation functions; is the hyperbolic tangent function; is the ReLU activation function; is the adaptive parameter of the ReLU activation function.

[0051] Furthermore, the calculation method of the feature adaptation parameters is expressed as: (20) In the formula, is the adjustment coefficient of the parameter; is the mean of the input vector, calculated as , is the dimension of the input vector; To avoid a small constant that divides by zero. Preferably, Set to 0.2, Set to 0.001.

[0052] Furthermore, the calculation method of the dynamic weight coefficient is expressed as: (twenty one) In the formula, is the L2 norm.

[0053] Furthermore, the calculation method of the adaptive parameters of the ReLU activation function is expressed as: (twenty two) In the formula, is the learning rate of the adaptive parameters; is the control parameter of the adaptive parameter; is the skewness index of the feature. Preferably, Set to 0.2.

[0054] Furthermore, the skewness index of the feature is calculated based on the feature variance statistics and is expressed as: (twenty three) (twenty four) In the formula, is the characteristic variance, is the total number of samples.

[0055] S403: The decoder receives the low-dimensional potential representation from the encoder and attempts to reconstruct the original high-dimensional feature vector through an inverse nonlinear transformation, which is expressed as: (25) In the formula, High-dimensional feature vector reconstructed for the decoder; is the mapping function of the decoder; is an adaptive activation function.

[0056] S404: Compare the reconstructed features output by the decoder with the original input features, and calculate the reconstruction error between the two. The reconstruction error reflects the degree of information loss in the process of dimensionality reduction and reconstruction. The calculation method is expressed as: (26) In the formula, is the reconstruction loss of the autoencoder; For the decoder The reconstructed feature vector of samples; is the regularization parameter; is the regularization term of the autoencoder weight; is the entropy regularization coefficient, which controls the impact of the entropy term on the total loss; is the entropy term of the low-dimensional representation of the latent space.

[0057] Furthermore, the regularization term of the autoencoder weights uses sparse regularization and feature decoration terms to prevent overfitting, promote the sparsity of the autoencoder weights, and reduce the correlation between the encoder weight vectors. The calculation method is expressed as: (27) In the formula, Represents the Frobenius norm, which is used to prevent overfitting and promote the generalization ability of the model; is the first regularization coefficient of the autoencoder; is the second regularization coefficient of the autoencoder; is the encoder weight matrix Line Elements of a column; is the encoder weight matrix Elements of the row; is the encoder weight matrix Elements of a row.

[0058] S405: Calculate the entropy term of the low-dimensional representation of the latent space using a latent space information compression strategy based on entropy constraints. The calculation method is expressed as: (28) In the formula, For the The samples in the latent space The value of the dimension; is the total number of samples; is the dimension of the low-dimensional representation of the latent space, that is, the dimension of the feature vector after dimensionality reduction; For the The samples in the latent space The probability density function of the value of dimension.

[0059] Furthermore, since the potential representation is the continuous output of the encoder, its probability density function needs to be estimated, and the kernel density estimation method is used for calculation, which is expressed as: (29) In the formula, is the bandwidth parameter, which controls the smoothness of the kernel function; For the The samples in the latent space The value of the dimension; is the kernel function. Preferably, Set to 2.

[0060] Furthermore, the kernel function is calculated using a Gaussian kernel, expressed as: (30) In the formula, Represents the input of the function.

[0061] S406: Update the parameters of the encoder and decoder, and dynamically adjust the corresponding weights and learning rates so that the model pays more attention to the accurate reconstruction of important features. The updating method of the autoencoder parameters is expressed as:

[0062]

[0063]

[0064] (31) In the formula, is the adaptive learning rate of the autoencoder; It is the parameter update operation; is the gradient of the reconstruction loss of the autoencoder with respect to the encoder weight parameters; is the gradient of the reconstruction loss of the autoencoder with respect to the decoder weight parameters; is the gradient of the autoencoder reconstruction loss with respect to the encoder bias parameter; is the gradient of the autoencoder reconstruction loss with respect to the decoder bias parameters.

[0065] Furthermore, the calculation method of the adaptive learning rate of the autoencoder is expressed as: (32) S407: Repeat the above steps until the preset stop iteration condition is met, which means that the model training is completed and the feature vector after dimensionality reduction is obtained. .

[0066] In one embodiment, the preset condition for stopping iteration is reaching a preset maximum number of iterations. Preferably, the preset maximum number of iterations is set to 1000 times.

[0067] S5: Training the classifier model The feature vector of the orthognathic surgery data after dimensionality reduction is input into the classifier for classification. The present invention adopts a support vector machine based on fuzzy constraints as a classifier model. On the basis of the traditional support vector machine, fuzzy logic is used to solve the limitations of the traditional support vector machine algorithm in dealing with uncertainty, noise and fuzzy data, so that it can be more accurately classified and has stronger robustness when facing the task of diagnosing temporomandibular joint diseases.

[0068] Structurally, the fuzzy constraint-based support vector machine consists of the following main parts: Fuzzification layer: Fuzzification is performed on the input orthognathic surgery data after feature dimension reduction, so that each data point is no longer a definite value, but is represented by multiple fuzzy values. Through this layer, the uncertainty and ambiguity in the orthognathic surgery data can be handled; Optimization layer: Using the framework of fuzzy-constrained support vector machine, an optimization model is established. This model obtains a classification hyperplane by maximizing the inter-class interval while minimizing the intra-class difference. Fuzzifying the orthognathic surgery data enables the classification hyperplane to maintain high accuracy in the presence of noise and uncertainty. Decision layer: Based on the hyperplane obtained by the optimization layer, decision reasoning is performed to determine which category the new sample belongs to. Through the fuzzy reasoning mechanism, the uncertainty of the classification result is processed to give the final classification result.

[0069] Specifically, the training process of the support vector machine based on fuzzy constraints is as follows: S501: Fuzzify the orthognathic surgery data after feature dimension reduction. The feature value of each sample will be converted into a fuzzy value. The degree of fuzzification is controlled by the fuzzy membership function. The purpose of fuzzification is to introduce the fuzziness and uncertainty of the input orthognathic surgery data into the model, so as to provide richer information for the subsequent training process. Specifically, for each input data point , is the first orthognathic surgery data input to the support vector machine The features are converted into fuzzy values ​​using the fuzzy membership function to represent the fuzziness of each data point, expressed as: (33) In the formula, is the fuzzy membership function, Indicates that the sample is Fuzzy membership on features; For the sample Features For the sample The mean of the features; is the fuzzy control parameter; is an exponential function.

[0070] Furthermore, the fuzzy control parameter determines the steepness of the fuzzy membership curve, which is dynamically adjusted according to the characteristic variance of the sample. The calculation method is expressed as: (34) In the formula, is the standard deviation of all characteristics of the sample; is a small positive number to prevent division by zero errors. Preferably, Set to 0.01.

[0071] S502: In the training process of the support vector machine based on fuzzy constraints, the fuzzified orthognathic surgery data is input into the support vector machine model based on fuzzy constraints. The goal of the support vector machine algorithm based on fuzzy constraints is to construct a classifier by maximizing the inter-class interval (i.e., the distance between the decision hyperplane and each class of samples) and minimizing the intra-class difference (i.e., the distance between the sample and the hyperplane). In this process, the optimization goal is to improve the robustness of classification by maximizing the inter-class interval and to enhance the classification accuracy by minimizing the intra-class difference. Specifically, suppose the training set input to the support vector machine is , is the fuzzy feature of the sample, For the corresponding labels, the optimization objective of the support vector machine uses fuzzy constraints and disturbance terms to enhance the robustness of the model, which is expressed as: (35) The constraints are: (36) In the formula, is the weight vector of the sample; is the L2 norm; is the bias term of the support vector machine; is a slack variable used to allow misclassification of some samples; is the regularization parameter of the support vector machine, which is used to balance the weight between maximizing the interval and minimizing the misclassification; The number of samples input into the support vector machine for the current batch; is the adjustment coefficient of the fuzzy constraint optimization item; is the fuzzy constraint optimization term, which represents the uncertainty of fuzzy membership; is the adjustment coefficient of disturbance intensity; is a random disturbance term.

[0072] S503: Adopting an adaptive quantitative perturbation mechanism, dynamically adjusting the perturbation intensity according to the distribution of sample features and the degree of noise, helps to avoid overfitting and improve the model's tolerance to outliers. The calculation method of the random perturbation term is expressed as: (37) In the formula, is the adjustment coefficient of disturbance intensity; is the growth rate parameter of the disturbance; is a preset threshold used to control the triggering condition of disturbance; is the disturbance intensity adjustment function. Preferably, Set to 2, Set to 0.5.

[0073] Furthermore, the fuzzy constraint optimization strategy is adopted to make the fuzzy membership adaptively adjusted instead of being fixed. The classifier can automatically adjust the fuzzy boundary according to different sample characteristics, enhancing the adaptability of the model in the variable orthognathic surgery data environment. The calculation method of the fuzzy constraint optimization term is expressed as: (38) In the formula, is the feature dimension of the sample input to the support vector machine; Indicates that the sample is Fuzzy membership on features; For the sample Features.

[0074] Furthermore, the perturbation intensity adjustment function is a dynamic function related to the model weight and input features. The perturbation intensity is dynamically adjusted according to the local perturbation and fuzzification of the current sample. The calculation method is expressed as: (39) In the formula, It is a coefficient automatically learned during the training process, indicating the adjustment speed of the disturbance intensity.

[0075] S504: After all the training processes are completed, the model will obtain an optimal hyperplane that can process fuzzy data. Classification is performed on this hyperplane. The model will calculate the membership degree of each category based on the input new sample and output the final classification result. Specifically, the final decision hyperplane is , and the classification decision function obtained is: (40) In the formula, is the classification result of the new sample; is the fuzzy feature of the new sample; is a symbolic function that determines the classification result based on the position of the hyperplane; is the random disturbance term; for each input data point; T is the transpose.

[0076] S6: Application of risk assessment in orthognathic surgery In one embodiment, the risk assessment of orthognathic surgery is performed using the trained risk classification model. First, the original orthognathic surgery data is collected and input into the trained risk classification model. The trained feature extraction and feature dimension reduction model is used to perform feature processing. Further, the processed features are input into the classifier model for classification, thereby obtaining the classification results. In this embodiment, the classification categories include "low risk", "medium risk" and "high risk".

[0077] Example 2 In one embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the following method is implemented: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0078] Example 3 In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the following method is implemented: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0079] Example 4 In one embodiment of the present disclosure, an electronic device is provided, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the following method: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

[0080] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0081] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0082] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. Orthognathic surgery risk assessment system based on machine learning and data-driven, characterized by: include: A data acquisition module, configured to acquire and pre-process orthognathic surgery data of a patient to be evaluated; A risk assessment system is configured to input the pre-processed orthognathic surgery data of the patient into a risk classification model to obtain an auxiliary classification result of orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

2. The orthognathic surgery risk assessment system based on machine learning and data-driven according to claim 1, characterized in that: The orthognathic surgery data is augmented by a generative adversarial network based on cycle consistency, and the cycle consistency loss function is used to ensure the consistency of the generated orthognathic surgery data with the real orthognathic surgery data in overall distribution, including: 1) Initialize the parameters of the generator and discriminator of the generative adversarial network. The generator is responsible for generating orthognathic surgery target domain data from orthognathic surgery source domain data, and the discriminator is used to distinguish between the real orthognathic surgery target domain data and the generated orthognathic surgery target domain data; 2) The generator generates fake orthognathic surgery target domain data by learning features from orthognathic surgery source domain data; 3) Using the generator inverse mapping optimization strategy based on tensor low-rank approximation, the generator can retain the high-order feature information of orthognathic surgery data during the orthognathic surgery data conversion process; 4) The discriminator judges the authenticity of the orthognathic surgery data and evaluates whether the orthognathic surgery data generated by the generator conforms to the characteristics of the orthognathic surgery target domain data, so as to minimize the discrimination error between the real orthognathic surgery data and the generated orthognathic surgery data and construct an adversarial loss function.

3. The orthognathic surgery risk assessment system based on machine learning and data-driven according to claim 1, characterized in that: A dynamic adaptive oscillation-based neural network is used for feature extraction, including: The dynamic adaptive oscillation-based neural network is a multi-layer fully connected neural network, in which each layer of the fully connected layer has 128 nodes. By simulating nonlinear oscillation behaviors in physical phenomena, the local extreme value optimization neural network is effectively extracted and utilized in a high-dimensional parameter space.

4. The orthognathic surgery risk assessment system based on machine learning and data-driven according to claim 1, characterized in that: In the dynamic adaptive oscillation-based neural network, the oscillation frequency of each parameter is calculated based on the current loss function. The adjustment of the oscillation frequency depends on the local curvature estimation of the loss surface. The adjustment is made using historical gradient information, so that the parameter update is smoother and more efficient search can be performed in the parameter space. The parameter update is performed in combination with the oscillation behavior, and the sine function and cosine function are used to simulate the oscillation behavior, allowing the parameters to be periodically explored in the gradient direction of the loss function.

5. The orthognathic surgery risk assessment system based on machine learning and data-driven according to claim 1, characterized in that: An autoencoder based on latent space mapping is used as the feature dimensionality reduction model. The autoencoder converts the input high-dimensional features into low-dimensional latent representations. The feature adaptive optimization mechanism is adopted. The parameters of the encoder are dynamically adjusted according to the statistical characteristics and correlation of the input features to better capture the intrinsic structure and pattern of the orthognathic surgery data after feature extraction. The decoder reconstructs the original high-dimensional feature vector through inverse nonlinear transformation. The feature weight adaptive adjustment strategy is adopted to focus on the features that have a great impact on the dimensionality reduction results.

6. The orthognathic surgery risk assessment system based on machine learning and data-driven according to claim 1, characterized in that: A support vector machine based on fuzzy constraints is used as the classifier model. The classifier model includes a fuzzification layer, an optimization layer and a decision layer. The fuzzification layer fuzzifies the input orthognathic surgery data after feature dimension reduction, so that each data point is no longer a definite value, but is represented by multiple fuzzy values; the optimization layer uses the framework of a support vector machine based on fuzzy constraints to establish an optimization model, and obtains a classification hyperplane by maximizing the inter-class interval and minimizing the intra-class difference. The fuzzified orthognathic surgery data enables the classification hyperplane to maintain high accuracy in the presence of noise and uncertainty; the decision layer performs decision reasoning based on the hyperplane obtained by the optimization layer to determine which category the new sample belongs to, and processes the uncertainty of the classification result through the fuzzy reasoning mechanism to give the final classification result.

7. The orthognathic surgery risk assessment system based on machine learning and data-driven according to claim 6, characterized in that: The final decision hyperplane is , and the obtained classification decision function is: In the formula, is the classification result of the new sample; is the fuzzy feature of the new sample; is a symbolic function that determines the classification result based on the position of the hyperplane; is the random disturbance term; is the bias term of the support vector machine; is the weight vector of the sample; For each input data point, T is the transpose.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the following method steps are implemented: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by the processor, the following method steps are implemented: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

10. An electronic device, characterized in that: include: A processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the following method steps: Obtain orthognathic surgery data of patients to be evaluated and pre-process them; Inputting the pre-processed orthognathic surgery data of the patient into the risk classification model to obtain auxiliary classification results for orthognathic surgery risk assessment; Among them, in the risk classification model, first, a generative adversarial network based on cycle consistency is used to expand the orthognathic surgery data; for the expanded orthognathic surgery data, a neural network based on dynamic adaptive oscillation is used to extract features; then the extracted high-dimensional features are used for feature dimensionality reduction using an autoencoder based on latent space mapping, and the input high-dimensional features are mapped to a low-dimensional latent space representation. A feature adaptive optimization mechanism is used to dynamically adjust the encoder parameters according to the statistical characteristics and correlation of the input features to further capture the intrinsic structure and pattern of the orthognathic surgery data; the reduced feature representation is input into a support vector machine based on fuzzy constraints, and the uncertainty and noise in the feature representation of the orthognathic surgery data are processed through the fuzzy membership function and the adaptive quantization perturbation mechanism, and the final orthognathic surgery risk assessment auxiliary classification result is output.

Images