Method for constructing capsule contracture prediction model after radiotherapy of fat transplantation prevention implant

Abstract

The invention discloses a method for constructing a prediction model for preventing capsule contracture after implant radiotherapy through fat transplantation, a machine learning model is trained by obtaining rat radiotherapy data of hemispherical prosthesis transplantation and surface fat injection treatment, a basic model is obtained, a machine learning algorithm can process massive multi-dimensional data, potential information can be mined, and the prediction model can be used for predicting capsule contracture. Constructing a prediction model basis of the scientific system; meanwhile, radiotherapy rat data and patient clinical data after the silica gel prosthesis is implanted are integrated, and a cross-species association rule is calculated based on a nuclear canonical correlation analysis model to adjust the basic model. According to the mode, multi-source data information is integrated, the scientificity and systematicness of the model are enhanced, and capsule contracture can be predicted more comprehensively and accurately; kernel canonical correlation analysis can process a non-linear relation, mine internal relation of cross-species data, and reasonably integrate animal experiment mandatory data into a model optimized based on patient data, so that the model gives consideration to mechanism understanding and clinical practice, and the accuracy and reliability of clinical application are improved.

Description

Technical Field

[0001] This invention relates to the field of predictive model construction technology, and in particular to a method for constructing a predictive model for preventing capsular contracture after fat grafting radiotherapy. Background Technology

[0002] Capsular contracture following radiotherapy is a medical complication that is common in patients who have received radiotherapy (such as post-operative radiotherapy for breast cancer) and have medical implants (such as breast implants, orthopedic implants, etc.). Its core mechanism is that the inflammatory and fibrotic reaction caused by radiotherapy leads to the formation of abnormally thickened scar tissue (capsule) around the implant, which gradually contracts, resulting in pain, implant deformation, or functional abnormalities.

[0003] Currently, the prediction and prevention of capsular contracture mainly rely on clinical experience and small-scale clinical observation studies, lacking a scientific and systematic predictive model. In addition, traditional experimental research methods are often limited to single data sources such as animal experiments or clinical cases, making it difficult to effectively combine the mechanistic data obtained from animal experiments with actual data from clinical patients, resulting in a low efficiency in translating research results into clinical applications. Summary of the Invention

[0004] In view of this, the present invention proposes a method for constructing a predictive model for preventing capsular contracture after fat grafting radiotherapy, which can effectively solve the shortcomings of existing technologies, such as the lack of a scientific and systematic predictive model and the difficulty in effectively combining mechanistic data obtained from animal experiments with actual data from clinical patients.

[0005] The technical solution of this invention is implemented as follows:

[0006] A method for constructing a predictive model for capsular contracture prevention after fat graft radiotherapy includes:

[0007] To obtain radiotherapy data from rats treated with hemispherical prosthesis transplantation and fat injection onto the surface of the hemispherical prosthesis;

[0008] A basic model was obtained by training a machine learning model based on data from rat radiotherapy.

[0009] To obtain data on rats that underwent radiotherapy after silicone prosthesis implantation and clinical data on patients who underwent radiotherapy after silicone prosthesis implantation;

[0010] Based on the nuclear canonical correlation analysis model, combined with rat data and patient clinical data, cross-species association rules were calculated;

[0011] By adjusting the basic model using cross-species association rules, a predictive model for capsular contracture prevention after implant radiotherapy following fat grafting was obtained.

[0012] As a further optional scheme for the method of constructing a predictive model for capsular contracture prevention after radiotherapy following fat grafting, the step of training a machine learning model based on rat radiotherapy data to obtain a basic model specifically includes:

[0013] The data from rat radiotherapy were preprocessed, and the preprocessed data were divided into training and testing sets.

[0014] The decision tree and random forest models are trained based on the training set to obtain the initial model;

[0015] The initial model is optimized based on the test set to obtain the base model.

[0016] As a further optional scheme for the method of constructing the prediction model for capsular contracture after radiotherapy in fat grafting, the calculation of cross-species association rules based on the nuclear canonical correlation analysis model, combined with rat data and patient clinical data, specifically includes:

[0017] Data from rats that underwent radiotherapy after silicone implantation and clinical data from patients who underwent radiotherapy after silicone implantation were input into a nuclear canonical correlation analysis model. The model was mapped to a high-dimensional feature space through a kernel function to extract nonlinear correlation features across species data.

[0018] Based on nonlinear correlation features, the KCCA projection vector is calculated, which is used to characterize the direction of maximum correlation and key feature weights between rat data and patient clinical data.

[0019] As a further optional scheme for the method of constructing the prediction model for capsular contracture after radiotherapy in fat grafting, the KCCA projection vector is calculated based on nonlinear correlation features. The KCCA projection vector is used to characterize the direction of maximum correlation and key feature weights between rat data and patient clinical data, specifically including:

[0020] From the nonlinear correlation features between rat data and patient clinical data extracted by the nuclear canonical correlation analysis model, a subset of features that meet the preset correlation threshold is selected.

[0021] The feature subset is calculated by the objective function of the kernel canonical correlation analysis model to obtain the orthogonalized KCCA projection vector. The direction of the KCCA projection vector represents the direction of maximum correlation of cross-species data in the high-dimensional feature space.

[0022] Based on the magnitude and component coefficients of the KCCA projection vector, the key feature weights between rat data and patient clinical data are determined, where the weight values ​​reflect the contribution of each feature to cross-species association.

[0023] The output includes a set of KCCA projection vectors containing the maximum relevance direction and key feature weights. These KCCA projection vector sets are used for cross-species data transformation and cross-domain adaptation of prediction models.

[0024] As a further optional scheme of the method for constructing the prediction model for capsular contracture prevention after radiotherapy with fat grafts, the method of adjusting the basic model using cross-species association rules to obtain the prediction model for capsular contracture prevention after radiotherapy with fat grafts specifically includes:

[0025] Based on KCCA projection vectors, a cross-species association rule base between rat data and patient clinical data is constructed. The rule base includes feature mapping relationships and weight parameters.

[0026] By utilizing the feature mapping relationships in the rule base, the input space of the basic model is transformed, and the feature distribution of rat data is mapped to the feature space of patient clinical data, thereby achieving cross-species data adaptation.

[0027] By using weight parameters from the rule base to adjust the input features of the basic model, a predictive model for preventing capsular contracture after fat grafting radiotherapy is obtained.

[0028] As a further optional scheme for constructing the predictive model of capsular contracture after radiotherapy for fat grafting, the rat radiotherapy data includes rat physiological data, hemispherical prosthesis data, radiotherapy conditions, rat skin change data, and fat injection volume.

[0029] A system for constructing a predictive model for capsular contracture prevention after fat graft radiotherapy includes:

[0030] The data acquisition module is used to acquire radiotherapy data of rats treated with hemispherical prosthesis transplantation and hemispherical prosthesis surface fat injection, data of rats treated with radiotherapy after silicone prosthesis implantation, and clinical data of patients treated with radiotherapy after silicone prosthesis implantation.

[0031] The basic model building module is used to receive rat radiotherapy data and train the rat radiotherapy data based on machine learning algorithms to generate a basic prediction model.

[0032] The cross-species association rule calculation module is used to calculate cross-species association rules based on the nuclear canonical correlation analysis model, combined with the acquired rat data and patient clinical data.

[0033] The model adjustment module is used to adjust the basic model using cross-species association rules to obtain a predictive model for preventing capsular contracture after fat grafting radiotherapy.

[0034] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described methods for constructing a predictive model for preventing capsular contracture after radiotherapy with fat graft implants.

[0035] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described methods for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting.

[0036] The beneficial effects of this invention are as follows: By acquiring radiotherapy data from rats treated with hemispherical prosthesis implantation and fat injection onto the surface of the hemispherical prosthesis, and training a machine learning model based on this data to obtain a basic model, the machine learning algorithm can automatically learn patterns and rules from large amounts of complex data. Compared to traditional methods based on limited clinical experience or small-scale observational studies, it can handle multi-dimensional and massive amounts of data, uncovering potential information hidden within the data, thereby constructing a data-driven, scientific, and systematic predictive model foundation. Furthermore, in addition to utilizing rat experimental data, it also acquires data from rats that have undergone radiotherapy after silicone prosthesis implantation and patient clinical data. Based on a nuclear canonical correlation analysis model, it combines this data to calculate cross-species association rules, thereby adjusting the basic model. This method of integrating multi-source data fully considers… The characteristics and value of different data sources enable the model to integrate information from multiple aspects, further enhancing its scientific rigor and systematicity, and enabling it to predict the occurrence of capsular contracture more comprehensively and accurately. Furthermore, by employing a nuclear canonical correlation analysis model, combined with rat data and patient clinical data to calculate cross-species association rules, nuclear canonical correlation analysis can handle nonlinear relationships and uncover the intrinsic connections between data from different species. This means that the information contained in the mechanistic data obtained from animal experiments can be rationally integrated into the model optimized based on patient clinical data through cross-species association rules. After this adjustment, the model retains the understanding of the capsular contracture mechanism from animal experimental data while fully considering the actual situation of clinical patients, achieving an effective combination of animal experimental mechanistic data and actual clinical patient data, thus improving the accuracy and reliability of the model in clinical applications. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1This is a flowchart of a method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, according to the present invention.

[0039] Figure 2 This is a schematic diagram of the components of a fat grafting prevention capsular contracture prediction model construction system according to the present invention.

[0040] Figure 3 This is a schematic diagram of the composition of a computing device according to the present invention. Detailed Implementation

[0041] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] refer to Figures 1 to 3 A method for constructing a predictive model for capsular contracture prevention after fat grafting radiotherapy, comprising:

[0043] Data on radiotherapy of rats treated with hemispherical prosthesis transplantation and fat injection on the surface of the hemispherical prosthesis were obtained; in some embodiments, the radiotherapy data of rats included rat physiological data, hemispherical prosthesis data, radiotherapy conditions, rat skin changes data, and fat injection volume.

[0044] Specifically, the acquired rat radiotherapy data covers multiple dimensions of information, including rat physiological data, hemispherical prosthesis data, radiotherapy conditions, rat skin changes, and fat injection volume. This comprehensive data collection can reflect the overall condition of rats after receiving hemispherical prosthesis transplantation, fat treatment, and radiotherapy from multiple perspectives, providing rich and detailed basic data for subsequent analysis and model construction, and helping to understand the related factors of capsular contracture more deeply.

[0045] Based on existing data on rat radiotherapy, there is no need to conduct new experiments to obtain data, avoiding duplication of work and waste of resources, significantly improving the efficiency of data utilization, and enabling research to move more quickly to the analysis and modeling stage.

[0046] A basic model is obtained by training a machine learning model based on rat radiotherapy data; in some embodiments, the training of the machine learning model based on rat radiotherapy data to obtain the basic model specifically includes:

[0047] The data from rat radiotherapy were preprocessed, and the preprocessed data were divided into training and testing sets.

[0048] The decision tree and random forest models are trained based on the training set to obtain the initial model;

[0049] The initial model is optimized based on the test set to obtain the base model.

[0050] Specifically, preprocessing the rat radiotherapy data can effectively remove unreasonable information such as noise, outliers, and missing values ​​from the data. Through data cleaning and organization, the quality and reliability of the data are improved, providing a more accurate data foundation for subsequent model training and helping the model learn the truly valuable patterns and rules in the data.

[0051] Decision tree and random forest models were trained separately on the training set. The decision tree model has the advantages of being intuitive, easy to understand and interpret, and can clearly show the decision rules between features and target variables. The random forest model, by integrating multiple decision trees, improves the stability and accuracy of the model and can better handle complex data relationships. By training these two different types of models, their respective advantages can be fully utilized to explore a model structure more suitable for the dataset. The initial models obtained after training have been able to learn certain knowledge and patterns from the training set data, and have made preliminary identification and modeling of the relationship between rat radiotherapy data and capsular contracture. This lays a good foundation for the subsequent optimization of the model and the construction of the final prediction model.

[0052] The initial model is optimized based on the test set. By adjusting the model parameters and feature selection, the model performance is further improved. The optimized basic model can fit the data more accurately, improve the prediction accuracy of capsular contracture, and reduce prediction error.

[0053] We obtained data on rats that underwent radiotherapy after silicone implantation and clinical data on patients who underwent radiotherapy after silicone implantation.

[0054] Specifically, the rat data and patient clinical data after silicone prosthesis implantation followed by radiotherapy are based on existing data. The existing rat data includes a series of responses and changes produced by radiotherapy after silicone prosthesis implantation under different experimental conditions, such as changes in physiological indicators and tissue responses. The patient clinical data covers various aspects such as individual characteristics, treatment process, and treatment effects of patients in real clinical scenarios. This rich data provides a comprehensive and detailed foundation for in-depth research on capsular contracture.

[0055] By combining rat data and patient clinical data, cross-species research was achieved. As a model organism, rats share certain physiological mechanisms with humans. By comparing and analyzing the data from both species, we can uncover conserved biological laws and mechanisms among species, while also discovering human-specific responses and characteristics. This cross-species research perspective can provide a more comprehensive and in-depth understanding of the mechanism of capsular contracture induced by radiotherapy after silicone implantation.

[0056] Existing data can be analyzed from multiple dimensions. For example, rat data can be used to study the influence mechanism of factors such as radiation dose and prosthesis characteristics on capsular contracture; patient clinical data can be used to analyze the relationship between individual differences such as age, gender, and underlying diseases and capsular contracture. Multidimensional analysis helps to comprehensively reveal the various factors affecting capsular contracture and provides a basis for developing targeted prevention and treatment strategies.

[0057] Based on the nuclear canonical correlation analysis model, and combining rat data and patient clinical data, cross-species association rules are calculated; in some embodiments, the calculation of cross-species association rules based on the nuclear canonical correlation analysis model, and combining rat data and patient clinical data, specifically includes:

[0058] Data from rats that underwent radiotherapy after silicone implantation and clinical data from patients who underwent radiotherapy after silicone implantation were input into a nuclear canonical correlation analysis model. The model was mapped to a high-dimensional feature space through a kernel function to extract nonlinear correlation features across species data.

[0059] Based on nonlinear correlation features, the KCCA projection vector is calculated, which is used to characterize the direction of maximum correlation and key feature weights between rat data and patient clinical data.

[0060] Specifically, rat data and patient clinical data are input into a kernel canonical correlation analysis model and mapped to a high-dimensional feature space through a kernel function. The kernel function can handle nonlinear relationships, mapping complex and difficult-to-observe nonlinear patterns in the original data to a high-dimensional space, making them present in a linearly separable or more easily analyzable form. This step uncovers the hidden nonlinear correlation features in cross-species data, providing richer and more representative data features for subsequent analysis. Extracting nonlinear correlation features of cross-species data in the high-dimensional feature space can more comprehensively capture the intrinsic connections between rat data and patient clinical data. These correlation features not only include simple linear correlations but also cover complex nonlinear relationships, which helps to deepen the understanding of the interaction and influence mechanisms between data from different species.

[0061] The KCCA projection vector is calculated based on nonlinear correlation features. This projection vector is used to characterize the direction of maximum correlation between rat data and patient clinical data. By determining the direction of maximum correlation, the strongest association between the two datasets can be identified, providing crucial directional guidance for subsequent research and applications. For example, when studying the influencing factors of capsular contracture, we can further analyze which factors have consistent associations across species along this direction of maximum correlation. The KCCA projection vector also characterizes the weights of key features, reasonably quantifying the importance of each feature in cross-species association. This helps to screen features that have a significant impact on capsular contracture prediction and related research, avoiding interference from irrelevant or secondary features. When building a predictive model, resources can be allocated reasonably according to feature weights to improve the efficiency and accuracy of the model.

[0062] This approach effectively combines rat experimental data with patient clinical data, breaking down species barriers. In traditional research, animal experiments and clinical studies are often independent, making it difficult to apply mechanistic findings from animal experiments to clinical practice. However, by calculating cross-species association rules using a nuclear canonical correlation analysis model, commonalities and differences between the two can be identified, providing a scientific basis for extending animal experimental conclusions to human clinical practice. The calculated cross-species association rules also facilitate the knowledge transfer between basic research and clinical applications. Knowledge about the mechanism of capsular contracture obtained from rat data can be correlated with patient clinical data through association rules, thereby better guiding clinical diagnosis and treatment. For example, based on the relationship between certain biomarkers found in rat experiments and capsular contracture, combined with association rules from patient clinical data, these biomarkers can be monitored clinically to predict the risk of capsular contracture in advance.

[0063] In some embodiments, the calculation of the KCCA projection vector based on nonlinear correlation features, wherein the KCCA projection vector is used to characterize the direction of maximum correlation and key feature weights between rat data and patient clinical data, specifically including:

[0064] From the nonlinear correlation features between rat data and patient clinical data extracted by the nuclear canonical correlation analysis model, a subset of features that meet the preset correlation threshold is selected.

[0065] The feature subset is calculated by the objective function of the kernel canonical correlation analysis model to obtain the orthogonalized KCCA projection vector. The direction of the KCCA projection vector represents the direction of maximum correlation of cross-species data in the high-dimensional feature space.

[0066] Based on the magnitude and component coefficients of the KCCA projection vector, the key feature weights between rat data and patient clinical data are determined, where the weight values ​​reflect the contribution of each feature to cross-species association.

[0067] The output includes a set of KCCA projection vectors containing the maximum relevance direction and key feature weights. These KCCA projection vector sets are used for cross-species data transformation and cross-domain adaptation of prediction models.

[0068] Specifically, from the nonlinear correlation features extracted by the nuclear canonical correlation analysis model, a subset of features that meet the preset correlation threshold is selected. This step can effectively filter out features that are not closely related to cross-species relationships and reduce noise interference in the data.

[0069] The feature subset is calculated by the objective function of the kernel canonical correlation analysis model, and the orthogonalized KCCA projection vector is obtained. The design of the objective function makes the calculation process have a clear direction and purpose, which can ensure that the obtained projection vector is optimal in the direction of the maximum correlation of cross-species data. The orthogonalization further guarantees the independence between each projection vector, avoids redundancy between features, and enables the features to more accurately reflect the internal structure of the data.

[0070] The direction of the KCCA projection vector represents the direction of maximum correlation between cross-species data in a high-dimensional feature space. Clarifying this direction is crucial for understanding the intrinsic relationship between rat data and patient clinical data. It provides key directional guidance for subsequent research and applications. For example, when studying the influencing factors of capsular contracture, we can analyze which factors have consistent correlations across species along this direction of maximum correlation, which helps to reveal the core mechanism of capsular contracture. Based on the magnitude and component coefficients of the KCCA projection vector, the key feature weights between rat data and patient clinical data are determined. The weight values ​​reflect the contribution of each feature to the cross-species association. This quantification method allows researchers to clearly understand the importance of each feature in the cross-species association, providing a scientific basis for feature selection and model construction. When building predictive models, resources can be allocated reasonably according to feature weights, focusing on features that contribute greatly to the association, thereby improving model performance.

[0071] The KCCA projection vector set is used for cross-domain adaptation of prediction models, which can significantly improve the model's adaptability across different data domains (rat data domain and patient clinical data domain). The model can better integrate information from cross-species data and use mechanistic research results from rat data to optimize the prediction of patient clinical data, thereby improving the accuracy and reliability of the prediction model in clinical applications and providing a more effective tool for predicting capsular contracture after fat grafting for the prevention of implant radiotherapy.

[0072] The basic model is adjusted using cross-species association rules to obtain a predictive model for capsular contracture prevention after radiotherapy with fat grafts; in some embodiments, the adjustment of the basic model using cross-species association rules to obtain a predictive model for capsular contracture prevention after radiotherapy with fat grafts specifically includes:

[0073] Based on KCCA projection vectors, a cross-species association rule base between rat data and patient clinical data is constructed. The rule base includes feature mapping relationships and weight parameters.

[0074] By utilizing the feature mapping relationships in the rule base, the input space of the basic model is transformed, and the feature distribution of rat data is mapped to the feature space of patient clinical data, thereby achieving cross-species data adaptation.

[0075] By using weight parameters from the rule base to adjust the input features of the basic model, a predictive model for preventing capsular contracture after fat grafting radiotherapy is obtained.

[0076] Specifically, a cross-species association rule base between rat data and patient clinical data is constructed based on KCCA projection vectors. The rule base includes feature mapping relationships and weight parameters. This step builds a bridge for communication between the data of two different species, enabling the mechanistic data obtained from rat experiments to be effectively connected with patient clinical data. Through feature mapping relationships, the correspondence between data features of different species is clarified, while the weight parameters quantify the importance of these features in cross-species association.

[0077] By utilizing the feature mapping relationships in the rule base to transform the input space of the basic model, the feature distribution of rat data is mapped to the feature space of patient clinical data, successfully achieving cross-species data adaptation. This process breaks down the data barriers between species, allowing data that originally belonged to different distributions and feature spaces to be analyzed and processed within the same framework, fully leveraging the advantages of rat data in mechanism research and the practical application value of patient clinical data.

[0078] The input features of the base model are weighted and adjusted using weight parameters from the rule base. These weight parameters reflect the contribution of each feature to cross-species associations. Through weighted adjustment, the model can focus more on features that play a key role in cross-species associations, thereby improving the model's specificity in predicting capsular contracture. This adjustment method allows the model to better integrate information from cross-species data, optimizing model performance. The fat grafting model for preventing capsular contracture after radiotherapy, adjusted using cross-species association rules, can more accurately capture the key factors affecting capsular contracture. Because it combines mechanistic research findings from rat experiments with actual clinical data from patients, the model is more accurate in predicting the risk of capsular contracture after radiotherapy, reducing prediction errors.

[0079] This predictive model can provide important reference for clinicians when developing treatment plans. Based on the model's prediction of the patient's risk of capsular contracture, and combined with the patient's individual circumstances, doctors can take corresponding preventive measures in advance, such as adjusting the fat grafting method, dosage, or radiotherapy parameters, thereby reducing the incidence of capsular contracture and improving treatment outcomes.

[0080] A system for constructing a predictive model for capsular contracture prevention after fat graft radiotherapy includes:

[0081] The data acquisition module is used to acquire radiotherapy data of rats treated with hemispherical prosthesis transplantation and hemispherical prosthesis surface fat injection, data of rats treated with radiotherapy after silicone prosthesis implantation, and clinical data of patients treated with radiotherapy after silicone prosthesis implantation.

[0082] The basic model building module is used to receive rat radiotherapy data and train the rat radiotherapy data based on machine learning algorithms to generate a basic prediction model.

[0083] The cross-species association rule calculation module is used to calculate cross-species association rules based on the nuclear canonical correlation analysis model, combined with the acquired rat data and patient clinical data.

[0084] The model adjustment module is used to adjust the basic model using cross-species association rules to obtain a predictive model for preventing capsular contracture after fat grafting radiotherapy.

[0085] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described methods for constructing a predictive model for preventing capsular contracture after radiotherapy with fat graft implants.

[0086] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described methods for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting.

[0087] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a predictive model for capsular contracture prevention after radiotherapy following fat grafting, characterized in that, include: To obtain radiotherapy data from rats treated with hemispherical prosthesis transplantation and fat injection onto the surface of the hemispherical prosthesis; A basic model was obtained by training a machine learning model based on data from rat radiotherapy. To obtain data on rats that underwent radiotherapy after silicone prosthesis implantation and clinical data on patients who underwent radiotherapy after silicone prosthesis implantation; Based on the nuclear canonical correlation analysis model, combined with rat data and patient clinical data, cross-species association rules were calculated; By adjusting the basic model using cross-species association rules, a predictive model for capsular contracture prevention after implant radiotherapy following fat grafting was obtained.

2. The method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in claim 1, is characterized in that... The basic model is obtained by training a machine learning model based on rat radiotherapy data, specifically including: The data from rat radiotherapy were preprocessed, and the preprocessed data were divided into training and testing sets. The decision tree and random forest models are trained based on the training set to obtain the initial model; The initial model is optimized based on the test set to obtain the base model.

3. The method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in claim 2, is characterized in that... The cross-species association rule calculated based on the nuclear canonical correlation analysis model, combined with rat data and patient clinical data, specifically includes: Data from rats that underwent radiotherapy after silicone implantation and clinical data from patients who underwent radiotherapy after silicone implantation were input into a nuclear canonical correlation analysis model. The model was mapped to a high-dimensional feature space through a kernel function to extract nonlinear correlation features across species data. Based on nonlinear correlation features, the KCCA projection vector is calculated, which is used to characterize the direction of maximum correlation and key feature weights between rat data and patient clinical data.

4. The method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in claim 3, is characterized in that... Based on nonlinear correlation features, a KCCA projection vector is calculated. This KCCA projection vector is used to characterize the direction of maximum correlation and key feature weights between rat data and patient clinical data, specifically including: From the nonlinear correlation features between rat data and patient clinical data extracted by the nuclear canonical correlation analysis model, a subset of features that meet the preset correlation threshold is selected. The feature subset is calculated by the objective function of the kernel canonical correlation analysis model to obtain the orthogonalized KCCA projection vector. The direction of the KCCA projection vector represents the direction of maximum correlation of cross-species data in the high-dimensional feature space. Based on the magnitude and component coefficients of the KCCA projection vector, the key feature weights between rat data and patient clinical data are determined, where the weight values ​​reflect the contribution of each feature to cross-species association. The output includes a set of KCCA projection vectors containing the maximum relevance direction and key feature weights. These KCCA projection vector sets are used for cross-species data transformation and cross-domain adaptation of prediction models.

5. The method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in claim 4, is characterized in that... The basic model is adjusted using cross-species association rules to obtain a predictive model for capsular contracture prevention after radiotherapy following fat grafting, specifically including: Based on KCCA projection vectors, a cross-species association rule base between rat data and patient clinical data is constructed. The rule base includes feature mapping relationships and weight parameters. By utilizing the feature mapping relationships in the rule base, the input space of the basic model is transformed, and the feature distribution of rat data is mapped to the feature space of patient clinical data, thereby achieving cross-species data adaptation. By using weight parameters from the rule base to adjust the input features of the basic model, a predictive model for preventing capsular contracture after fat grafting radiotherapy is obtained.

6. The method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in claim 5, is characterized in that... The rat radiotherapy data includes rat physiological data, hemispherical prosthesis data, radiotherapy conditions, rat skin changes data, and fat injection volume.

7. A system for constructing a predictive model for capsular contracture after fat grafting for prevention of implant radiotherapy, characterized in that, include: The data acquisition module is used to acquire radiotherapy data of rats treated with hemispherical prosthesis transplantation and hemispherical prosthesis surface fat injection, data of rats treated with radiotherapy after silicone prosthesis implantation, and clinical data of patients treated with radiotherapy after silicone prosthesis implantation. The basic model building module is used to receive rat radiotherapy data and train the rat radiotherapy data based on machine learning algorithms to generate a basic prediction model. The cross-species association rule calculation module is used to calculate cross-species association rules based on the nuclear canonical correlation analysis model, combined with the acquired rat data and patient clinical data. The model adjustment module is used to adjust the basic model using cross-species association rules to obtain a predictive model for preventing capsular contracture after fat grafting radiotherapy.

8. A computing device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for constructing a predictive model for preventing capsular contracture after radiotherapy following fat grafting, as described in any one of claims 1-6.

Images