Rapid prediction method for bonding strength value of dental restoration material based on machine learning

Abstract

The invention relates to the technical field of testing of the bonding strength of a dental restoration depending on resin cement bonding, in particular to a rapid prediction method for the bonding strength value of a dental restoration material based on machine learning, and the method comprises the following steps: S1, constructing an optimal XGB prediction model; s2, performing primary processing on to-be-predicted information; s3, inputting all related features; s4, obtaining a predicted bonding strength value range according to a binary classification result output after the operation model; s5, a prediction result provides reference for repair material selection, bonding product selection and bonding strategy optimization and adjustment; the method can be tightly combined with optimization of a dental restoration material bonding process, reference is provided for selection of the restoration material and dynamic adjustment of bonding operation steps and operation parameters by analyzing physical and chemical property parameters of the material in real time and combining the physical and chemical property parameters with a closed-loop control system of a prediction model, the optimal bonding effect of the restoration material is achieved, and the service life of the restoration material is prolonged. The reliability and the performance of the prosthesis in clinical application are improved, and the clinical service life of the prosthesis is prolonged.

Description

Technical Field

[0001] The present invention relates to the technical field of testing the bonding strength of a dental restoration that is dependent on resin cement bonding, and in particular to a method for quickly predicting the bonding strength value of a dental restoration material based on machine learning. Background Art

[0002] Porcelain, all-ceramic, polymer ceramics and other materials are the most commonly used restorative materials for patients with tooth defects and edentulous teeth in clinical practice due to their excellent aesthetic properties, mechanical properties and biocompatibility.

[0003] Bonding is a key factor in determining the success of restoration treatment and the length of clinical service life. Good bonding can not only provide a stable retention for the restoration, but also improve the aesthetic effect and marginal closure of the restoration.

[0004] Clinically, ideal bonding effects can be achieved by improving mechanical retention and chemical bonding. Resin cement is a conventional bonding material for dental restorations such as all-ceramic restorations and polymer ceramics. The usual bonding process involves pretreatment of the restoration surface, pretreatment of the tooth surface, and final bonding of the resin cement. For example, the pretreatment of the surface of all-ceramic restorations and polymer ceramic restorations usually includes hydrofluoric acid etching or sandblasting, the purpose of which is to roughen the surface and form a micromechanical interlock with the infiltrated resin cement; chemical adjustment of the surface, such as coating with silane coupling agent or 10-methacryloyloxydecyl dihydrogen phosphate (MDP), can promote the chemical bonding between the restoration material and the resin cement.

[0005] The most direct evaluation method for the bond strength between the restorative material and the resin cement is the in vitro bonding mechanics test, such as (micro) tensile bond strength and (micro) shear bond strength. However, the in vitro bonding test is limited. Although there are relatively standardized test methods and instruments, this standardization has made provisions on some parameters, such as specimen size, loading speed, loading head shape, etc., and has the assistance of statistical analysis, but it is far from enough to avoid the influence of errors caused by uncontrollable factors such as experimental instruments, operating technology sensitivity and test environment between different experiments. The values ​​measured between experiments or the comparative conclusions drawn may deviate greatly, and even completely opposite conclusions may appear. It is very common that even if there are hundreds or even thousands of similar papers on the same adherend, and each paper has a clear conclusion, these papers may present different conclusions, and it is impossible to judge which conclusion is closest to reality. Even though each of these traditional bonding experiments consumes a lot of time and resources for intensive specimen preparation and repeated laboratory measurements, the authors of the experiments all claim that their research conclusions are correct, but they have to admit that it is impossible to explain where the different conclusions in other studies come from, so the deviation between the conclusions of the experiments can only be attributed to the inherent shortcomings of traditional in vitro experiments. The low efficiency, low sensitivity and uncontrollability of bonding experiments make it difficult to compare the bonding results between experiments.

[0006] At the same time, research on changing the composition and structure of restorative materials to improve the mechanical strength of ceramics has been ongoing. In vitro bonding experiments to evaluate a new bonding product or bonding strategy usually select one or several types of restorative products as adherends, because the cost and workload of the experiment need to be considered. This brings a problem. The conclusions obtained from the bonding experiment may not be applicable to other restorative materials with different chemical compositions from the restorative products selected in the experiment, or at least inaccurate. Another problem is that when evaluating the effect of a certain active ingredient in a bonding product on the bonding of a restorative material alone, whether this active ingredient is affected or interacted with other chemical components in the bonding product or system, and thus affects the bonding results, it is also impossible to evaluate. The composition of various chemical components in different products is different. There are at least hundreds of bonding products on the market. There is by no means only one advocacy for each product. In addition, the change in the chemical composition of the target restorative material to be bonded makes it unrealistic to accurately select the best bonding product and strategy for a restorative material from this ever-changing combination through traditional in vitro bonding experiments, even though every clinician pursues the best combination to ensure clinical life.

[0007] The above situation urges more powerful bonding experimental methods, which can evaluate the effects of factors such as the chemical composition of the bonding product, the composition of the repair material, and the bonding strategy on the bonding performance individually and interactively without the help of in vitro experiments, thereby helping to develop new restorative materials, new bonding products and bonding strategies.

[0008] Therefore, in order to address the above-mentioned problems of insufficient accuracy in predicting the bonding strength of dental restorative materials, low data utilization efficiency, lack of comprehensive analysis tools, and lack of comprehensive analysis tools, a method for rapid prediction of the bonding strength values ​​of dental restorative materials and resin cement based on machine learning can be designed. Summary of the invention

[0009] In order to overcome the current problems of insufficient measurement accuracy of the bond strength values ​​between dental restorative materials and resin cements, low data utilization efficiency, lack of comprehensive analysis tools and lack of comprehensive analysis tools.

[0010] The technical solution of the present invention is: a method for quickly predicting the bonding strength value of dental restoration materials based on machine learning, and the steps are as follows:

[0011] The steps are as follows:

[0012] S1: The optimal XGB prediction model was constructed based on the in vitro bonding experimental data of dental restorative materials by using stratified cross-validation, nested cross-validation and grid search methods;

[0013] S2: Preliminary processing of information such as the dental restoration material to be predicted, or the bonding strategy to be predicted, or the bonding operation method to be predicted to ensure data consistency. The format and structure of the new input data are consistent with the training data, such as containing the same features (variables), the same feature order, and the same data type, such as numerical or categorical;

[0014] S3: Input all relevant features of the adherends and related bonding treatment factors involved in the bonding performance of the repair material into the constructed python algorithm, including the chemical composition and processing method of the repair material, the roughening pretreatment method of the repair material surface, the chemical composition of the treatment agent applied to the repair material surface, the chemical composition of the adhesive applied to the repair material surface, the chemical composition of the resin cement, and the bonding steps;

[0015] S4: Through XGB model analysis, the predicted bonding strength value range is obtained according to the binary classification results output after the operation model;

[0016] S5: Based on the predicted bonding strength value range combined with the clinically acceptable threshold, determine the bonding results that can be achieved by the predicted restorative material, bonding strategy, or bonding method, and provide a reference for the selection of restorative materials, bonding products, and optimization and adjustment of bonding strategies.

[0017] Preferably, the preprocessed data is numerical, categorical or textual data that can be recognized by computer code.

[0018] Preferably, a rapid detection technology is used to predict and evaluate the bonding strength between a dental restoration that relies on resin cement bonding and a resin cement, and this technology does not require in vitro experiments.

[0019] Beneficial effects of the present invention:

[0020] 1. It can be closely combined with the optimization of dental restorative materials, corresponding bonding products and bonding strategies. By real-time analysis of material physical and chemical performance parameters such as chemical composition, surface roughening treatment method, bonding product chemical composition, bonding steps, etc., these key characteristics and the model's prediction results are synchronously fed back to the selection of restorative materials and the optimization of bonding strategies. Combined with its closed-loop control system with the prediction model, it provides a reference for selecting restorative materials, dynamically adjusting bonding operation steps and operating parameters, so as to better optimize the bonding strength, achieve the best bonding effect of restorative materials, improve the reliability and performance of restorations in clinical applications, and extend the clinical service life of restorations.

[0021] 2. In order to construct the optimal XGB model, a number of improvements were made in model construction and optimization. First, in view of the data complexity of the prediction of the bonding strength of repair materials, feature engineering methods and feature importance analysis were combined to deeply analyze the multiple factors affecting the bonding strength (such as the chemical composition of the repair material, the type of adhesive, the surface treatment method, etc.), screen key features, reduce the data dimension, and improve the prediction ability of the model. Secondly, the data set was divided into a training set and a test set, and stratified cross-validation, nested cross-validation, and grid search were performed on the test set to improve data utilization; the model prediction accuracy and generalization ability were further evaluated on the test set to improve the credibility of the model. Through these improvements, the XGB model not only shows stronger stability when processing high-dimensional data, but also enables the model to have better generalization ability in practical applications and can adapt to changes in different materials and processes;

[0022] 3. The model can predict the range of bonding strength, improve the accuracy and reliability of the prediction; efficiently process and utilize experimental data to improve data utilization; comprehensively analyze multiple influencing factors to provide comprehensive and systematic analysis tools; select and optimize restorative materials and appropriate bonding processes to achieve the best bonding strength and stability of the restoration. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 What is shown is a flow chart of a method for rapid prediction of bonding strength value of dental restorative materials based on machine learning of the present invention;

[0024] Figure 2What is presented is the prediction model screening of the bonding strength of lithium disilicate glass ceramics implemented in the present invention;

[0025] Figure 3 The results of analyzing the importance of features of lithium disilicate glass ceramics using 6 machine learning models implemented in the present invention are presented;

[0026] Figure 4 The average importance score of lithium disilicate glass-ceramics based on machine learning implemented in the present invention is shown;

[0027] Figure 5 What is shown is the ROC curve visualization on the training set and test set of the XGB model implemented in the present invention. DETAILED DESCRIPTION

[0028] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0029] Example 1

[0030] See also Figure 1 The present invention provides an embodiment: a method for quickly predicting the bonding strength value of dental restoration materials based on machine learning, the steps of which are as follows:

[0031] S1: The optimal XGB prediction model was constructed based on the in vitro bonding experimental data of dental restorative materials by using stratified cross-validation, nested cross-validation and grid search methods;

[0032] S2: Preliminary processing of the dental restoration material to be predicted, or the bonding strategy to be predicted, or the bonding method information to be predicted: the preprocessed data is numerical, categorical or text data that can be recognized by computer code; ensuring data consistency, the format and structure of the new input data are consistent with the training data, such as containing the same features (variables), the same feature order, and the same data type, such as numerical or categorical;

[0033] S3: Input all relevant features of the adherends and related bonding treatment factors involved in the bonding performance of dental restorative materials into the constructed python algorithm, including the chemical composition and processing method of the restorative materials, the roughening pretreatment method of the restorative material surface, the chemical composition of the treatment agent applied on the restorative material surface, the chemical composition of the adhesive applied on the restorative material surface, the chemical composition of the resin cement, and the bonding steps;

[0034] S4: Through XGB model analysis, the predicted bonding strength value range is obtained based on the binary classification results "1" or "0" output after the operation model is calculated;

[0035] S5: Based on the predicted bonding strength value range combined with the clinically acceptable threshold, determine the bonding performance that can be achieved by the bonding strategy or bonding method of the dental restorative material to be predicted, and provide a reference for the selection of restorative materials, bonding products, and optimization and adjustment of bonding strategies.

[0036] Example 2

[0037] See also Figure 2 The present invention provides an embodiment: establishment and screening of a prediction model for bonding strength of lithium disilicate glass ceramics based on machine learning, comprising the following steps:

[0038] 1. Data Collection

[0039] On the basis of Example 1, taking lithium disilicate reinforced glass ceramics as an example, considering that traditional bonding experiments require a lot of time and cost, and the tested bonding strategies and bonding methods are limited, this experiment collects data related to the in vitro bonding experiment of lithium disilicate reinforced glass ceramics from previous experimental data or articles to construct a data set;

[0040] 2. Feature Selection

[0041] Use feature engineering to extract useful information from raw data, including feature construction, feature encoding, and feature scaling;

[0042] Perform feature engineering and feature importance analysis on relevant features in the data set to ensure that the input data is highly relevant and reduce the data dimension to improve the prediction performance of the model;

[0043] Six machine learning models (LR, DT, RF, ET, GB, and XGB) were selected for feature importance analysis ( Figure 3 ); Each model provides a feature importance score for each input feature, denoted as I i,j , i=1,2,...,6 represents the model index, j=1,2,...,n represents the feature index;

[0044] Since the magnitude of the feature importance scores of each model is different, the feature importance scores of each model are normalized using min-max normalization to ensure comparability. The normalization formula is as follows:

[0045]

[0046] in, Represents model M i Medium Feature X i The normalized importance score of min,i and I max,i Model M ithe minimum and maximum importance scores of ;

[0047] After normalization, the average importance score of each feature is obtained by calculating the average of the normalized scores of the six models, as follows:

[0048]

[0049] in, Represents feature X i The average importance score of

[0050] Finally, the features were ranked according to their average importance scores to determine the relative importance of each feature in predicting the bond strength classification ( Figure 4 ), features with higher average scores are considered to be more influential in model prediction and have greater improvement significance in bonding process optimization. The features with higher average feature importance scores are restructured into datasets to reduce data complexity and improve prediction accuracy;

[0051] 3. Model training and evaluation

[0052] The dataset is divided into a training set (80%) and a test set (20%). The training set is used to build the machine learning model, and the test set is used to verify the performance of the model to ensure that the model still performs well on unseen data.

[0053] (1) Model training

[0054] The bonding strength of lithium pyrosilicate glass-ceramics was predicted using a binary classification model, an XGB model was constructed, and grid search was used for parameter optimization to find the best hyperparameter combination on the training set. Specifically, stratified k-fold cross validation (with a random parameter of 42) was used on the training set to evaluate the generalization performance of the model through different data partitions to ensure that the distribution of the bonding strength categories in each fold was balanced. Grid search cross validation was used to adjust the model hyperparameters to improve the prediction ability of the model.

[0055] (2) Model evaluation

[0056] The AUC-ROC curve (area under the receiver operating characteristic curve) is used to evaluate the classification effect of the model. This indicator can fully reflect the performance of the model. The classification performance of the model is further verified by the accuracy rate to ensure the accuracy of the prediction results. In order to reduce the impact of random parameters on the AUC score, the AUC scores of stratified cross-validation with random parameters of 0-29 are plotted as box plots. At the same time, in order to better evaluate the performance of the model, nested cross-validation is also used;

[0057] ROC (Receiver Operating Characteristic) curve generation method: For a binary classification model, the model performs For classification, where x i is the sample feature, y i ∈{0, 1} is the label, and the model outputs the predicted value For sample x i is classified as the positive class (i.e. ) probability or score;

[0058] The steps of constructing ROC curve: By setting the classification threshold t, the predicted value Convert to discrete labels That is: when hour, otherwise Among them, t∈[0,1] is an adjustable threshold. In most cases, the classification model sets the threshold to 0.5 by default, which means that if the model predicts that the probability of a sample is greater than or equal to 0.5, it is classified as a positive class;

[0059] Calculation of true positive rate (TPR) and false positive rate (FPR): For each threshold t, the true positive rate (TPR) and false positive rate (FPR) are calculated by the following formula:

[0060]

[0061]

[0062] Where I(·) is an indicator function, which takes the value of 1 when the condition in the brackets is met, otherwise it takes the value of 0. According to different thresholds t, the ROC curve is constructed, where each point is (FPR(t), TPR(t));

[0063] Calculation of the area under the ROC curve (AUC): Let t1, t2, …, t k is a threshold sequence arranged in descending order, and the corresponding TPR and FPR values ​​are (TPR1, FPR1), (TPR2, FPR2), ..., (TPR k ,FPR k ), then the area under the ROC curve (AUC) can be calculated by the trapezoidal method, the formula is:

[0064]

[0065] The above method constructs a framework that can generate ROC curves and calculate AUC values, where the ROC curve reflects the performance changes of the classification model under different thresholds, and the AUC value represents a comprehensive measure of the model's ability to distinguish between positive and negative samples;

[0066] 4. Model Validation

[0067] Obtain the AUC-ROC curve and accuracy performance indicators on the test set;

[0068] The optimal parameters and evaluation results of the model are shown in Table 1. Comprehensive analysis shows that XGB is the optimal model, which has achieved good generalization ability and accuracy in both the training set and the test set. Figure 5 Visualize the ROC curve on the XGB model training set and test set.

[0069] Example 3

[0070] Reliable prediction of bond strength values ​​of lithium pyrosilicate glass-ceramics using the XGB model

[0071] The known lithium disilicate glass-ceramic bonding data that were not included in the training or test set were used as separate validation data to determine the bonding product, glass-ceramic composition, bonding strategy and other information. By inputting key feature values ​​such as hydrofluoric acid concentration and treatment time, silane coupling agent composition, adhesive composition (such as Bis-GMA, UDMA, TEGDMA, filler), and cement composition (such as solvent type, Bis-GMA, UDMA), etc., and importing them into the constructed XGB optimal model (parameters are colsample_bytree:1.0, learning_rate:0.2, max_depth:5, n_estimators:200, subsample:0.9), the predicted range of bonding strength values ​​can be obtained (Table 1). The specific output results are 0 (low strength value) or 1 (high strength value). Compared with the actual results, the obtained results have higher accuracy, which is consistent with the prediction accuracy verified by the model.

[0072] Table 1 Performance evaluation of various models based on lithium disilicate glass-ceramics data.

[0073]

[0074]

[0075] Through the above steps, the following technical effects are achieved:

[0076] (1) Bond strength prediction: The main technical effect of the present invention is to use machine learning models to accurately predict the bond strength of dental restorative materials, thereby guiding the selection and use of materials.

[0077] (2) Optimized decision-making: By using a machine learning-based classification model, the bonding strength categories of dental restorative materials can be accurately distinguished, thereby improving the efficiency and accuracy of material selection in practical applications.

[0078] (3) Enhance the accuracy and robustness of the model: Stratified cross-validation (CV) and nested cross-validation and performance evaluation indicators (AUC-ROC curve, accuracy) are used to ensure the robustness of model evaluation and improve the generalization ability of the model.

[0079] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for rapidly predicting the bonding strength value of dental restoration materials based on machine learning, characterized in that: The steps are as follows: S1: The optimal XGB prediction model was constructed based on the in vitro bonding experimental data of dental restorative materials by using stratified cross-validation, nested cross-validation and grid search methods; S2: Preliminary processing of information such as the dental restoration material to be predicted, or the bonding strategy to be predicted, or the bonding operation method to be predicted to ensure data consistency. The format and structure of the new input data are consistent with the training data, such as containing the same features (variables), the same feature order, and the same data type, such as numerical or categorical; S3: Input all relevant features of the adherends and related bonding treatment factors involved in the bonding performance of the repair material into the constructed python algorithm, including the chemical composition and processing method of the repair material, the roughening pretreatment method of the repair material surface, the chemical composition of the treatment agent applied to the repair material surface, the chemical composition of the adhesive applied to the repair material surface, the chemical composition of the resin cement, and the bonding steps; S4: Through XGB model analysis, the predicted bonding strength value range is obtained according to the binary classification results output after the operation model; S5: Based on the predicted bonding strength value range combined with the clinically acceptable threshold, determine the bonding results that can be achieved by the predicted restorative material, bonding strategy, or bonding method, and provide a reference for the selection of restorative materials, bonding products, and optimization and adjustment of bonding strategies.

2. The method for rapid prediction of bonding strength value of dental restorative materials based on machine learning according to claim 1, characterized in that: The preprocessed data is numerical, categorical or textual data that can be recognized by computer code.

3. The method for rapid prediction of bonding strength value of dental restoration materials based on machine learning according to claim 1, characterized in that: A rapid testing technology used to predict and evaluate the bonding strength between dental restorations that rely on resin cement and resin cement. This technology does not require in vitro experiments.

Images