A system and method for predicting and optimizing temperature rise values for root canal post space preparation

Abstract

The application relates to the field of dental defect repair treatment and related medical devices, in particular to a root canal post channel preparation temperature rise prediction and optimization method and system based on machine learning, the system comprises: a user input module: used for receiving actual clinical operation parameters input by an operator; a temperature rise prediction module: used for predicting a temperature rise range under the input parameter combination; a risk assessment module: used for judging a risk level according to the temperature rise prediction result and combining a clinical safety threshold; a parameter optimization suggestion module: used for outputting a suggested adjusted parameter combination when the prediction result is high risk; an auxiliary decision module: used for outputting an optimal recommended parameter combination; the invention system is based on a mature model trained on a large amount of early experimental data, only needs to input basic operation information, can realize temperature rise prediction and risk early warning in the root canal preparation process, and effectively avoids complications such as high-temperature burn and cell damage, and provides intelligent support for accurate, safe and efficient root canal post channel preparation.

Description

Technical Field

[0001] This invention relates to the field of dental defect repair and treatment and the development of related medical devices, and in particular to a system and method for predicting and optimizing the temperature rise value of root canal post preparation. Background Technology

[0002] In clinical practice, abnormally elevated root surface temperature caused by frictional heat remains a key biomechanical risk factor affecting long-term tooth prognosis. When the local temperature exceeds the generally accepted critical value of 10°C, it may trigger periodontal ligament inflammation, alveolar bone resorption, or even irreversible damage to the hard tissues of the tooth, significantly increasing the risk of serious complications such as osteonecrosis, bone resorption, and root fracture. This microscopic thermal effect caused by mechanical friction is essentially the result of a dynamic interaction between the characteristics of the bur system, the operational technique parameters, and the tooth's own anatomical structure, and its complexity far exceeds the simple superposition of a single factor.

[0003] From a mechanical perspective, the design features of the bur system directly determine the efficiency of frictional heat generation. Burs with different cutting edge shapes (such as round and oval burs) and taper designs exhibit significant differences in root canal wall contact area, cutting path, and material removal method. Clinical observations show that Gates Glidden / GPX reamers, due to their unique geometry, often exhibit milder heat generation characteristics, while Parapost burs and PD reamers are more prone to exceeding the safe temperature rise threshold. Notably, the temperature control advantage exhibited by oval burs in the apical region further confirms the crucial influence of the matching between the cutting edge shape and the anatomical area on the thermal effect.

[0004] The choice of operating techniques constitutes a dynamic variable system for controlling thermal effects. The speed setting of the pneumatic motor not only determines the linear velocity of the sprue but also directly affects the amount of material removed per unit time and the frictional heat generation power. Experimental data shows that increasing the speed from 4000 rpm to 10000 rpm significantly alters the heat generation rate; this non-linear change highlights the importance of speed as a core variable. The choice of cooling strategy directly determines the heat dissipation efficiency: although continuous water cooling has been proven to have a significant cooling effect, recent research has found that under specific sprue combinations and speed conditions, even continuous water cooling may still result in a critical temperature rise. This suggests that we need to re-examine the matching relationship between cooling methods and sprue characteristics. Furthermore, the control of preparation time is equally crucial—extending the continuous contact time from 3 seconds to 6 seconds leads to an exponential increase in the heat accumulation effect; this non-linear characteristic of the time-temperature curve requires the operator to precisely control the preparation rhythm.

[0005] The anatomical features of teeth themselves constitute a fundamental barrier to heat conduction. Measurements of remaining dentin thickness show that the thinner the dentin layer remaining in the root canal wall, the more efficient the heat conduction from the pulp chamber to the periodontal tissues. This individualized anatomical difference explains why different teeth exhibit significant differences in temperature response under the same operating conditions. In particular, when the remaining dentin thickness is below a critical value, even mild operating parameters can cause the root surface temperature to exceed the safe range.

[0006] Although existing research has revealed key influencing factors of temperature rise from different dimensions, clinical translation still faces multiple challenges. Current research paradigms generally suffer from three core limitations: First, the fragmented nature of experimental design makes it difficult to systematically integrate conclusions. Different studies yield seemingly contradictory results due to differences in sample selection and variable control. For example, the controversy surrounding the effectiveness of continuous water cooling actually reflects the complex interaction between bur type, rotation speed, and cooling mode. Second, traditional experimental methods are limited by single-factor or low-dimensional variable control, making it difficult to capture the true effects of multi-parameter nonlinear superposition. In particular, the dynamic coupling mechanism between bur parameters, operating conditions, and anatomical features remains unclear. Third, the inherent variability and limited number of extracted tooth samples severely restrict the extrapolation of research conclusions. The natural differences between different teeth in root canal morphology, dentin characteristics, etc., make conclusions based on small samples difficult to apply to complex clinical scenarios. These bottlenecks essentially stem from the inherent limitations of traditional research methods when dealing with high-dimensional, highly interactive biomechanical systems.

[0007] Existing technologies have several shortcomings. First, current root canal preparation temperature rise mainly relies on single-parameter linear models from extracted tooth experiments (such as those only related to rotation speed), which cannot quantify the multi-factor coupling effect of instrument design, operating parameters, and anatomical structures. Prediction errors often exceed ±2℃, making it difficult to meet the precise control requirements of clinical safety thresholds. Second, massive amounts of historical experimental data (such as temperature rise records under different bur brands, cooling schemes, and root canal morphology combinations) have not been systematically integrated, and traditional statistical methods cannot analyze the nonlinear relationships between high-dimensional features, resulting in low data utilization. Third, existing tools... It can only analyze the main effect of a single variable (such as rotational speed) on temperature rise, lacking the ability to model key interaction terms (such as "high torque × low coolant flow rate"), resulting in blind spots in risk prediction; fourth, doctors need to adjust parameters based on experience during surgery, without a real-time early warning system; fifth, in order to find a safe combination of parameters (such as the upper limit of rotational speed to avoid thermal damage), it is necessary to repeat extracted tooth experiments, with a single group validation taking more than 48 hours, and the conclusions are difficult to generalize to complex clinical scenarios; sixth, traditional models are only compatible with specific instruments, and when new products are introduced, experiments need to be repeated, resulting in low clinical translation efficiency.

[0008] Therefore, to address the aforementioned problems, this invention proposes a system and method for predicting and optimizing the temperature rise value during root canal preparation. By integrating multi-source heterogeneous data such as bur design parameters, operational technique variables, and tooth anatomical features, it deeply explores the nonlinear correlation and interaction mechanisms among various factors. This data-driven research paradigm can not only systematically analyze the complex causes of temperature rise phenomena but also provide personalized predictive support for clinical decision-making. By establishing a mapping relationship between operational parameters and temperature response, it helps doctors predict the thermal effect risks of different combination schemes before surgery, thereby formulating precise post preparation strategies. This innovative attempt not only provides an intelligent tool for solving the problem of thermal damage in root canal treatment but also provides new ideas for the transformation of biomechanical research paradigms. Summary of the Invention

[0009] To overcome the problem of insufficient accuracy in temperature rise prediction in existing technologies, this invention proposes a system and method for predicting and optimizing the temperature rise value of root canal piling preparation.

[0010] The technical solution of this invention is: a system for predicting and optimizing the temperature rise value during root canal preparation, comprising:

[0011] User input module: Used to receive actual clinical operation parameters input by the operator, including instrument parameters, rotation speed and cooling conditions;

[0012] Temperature rise prediction module: used to call the optimal extreme random tree model trained based on large sample experimental data to automatically predict the temperature rise range under the combination of input parameters;

[0013] Risk assessment module: Based on the temperature rise prediction results and combined with clinical safety thresholds, the risk level is determined and presented in the form of color and graded labels for easy and quick identification;

[0014] Parameter optimization suggestion module: used to automatically output suggested parameter combinations for adjustment when the prediction result is high risk;

[0015] Decision Support Module: This module combines clinical goals with inverse analysis based on established models to output recommended parameter combinations within the optimal or acceptable range for personalized reference.

[0016] Preferably, the risk level output by the temperature rise prediction module includes low risk, medium risk and high risk, where a high risk result will trigger the system to automatically generate a real-time warning and simultaneously display the recommended parameter combination for adjustment.

[0017] Preferably, after setting the temperature rise threshold, the auxiliary decision-making module calls the ET model and combines it with the Monte Carlo sampling method to select parameter combinations with low risk prediction results in the feature space, further statistically analyzes the distribution of their key features, and outputs personalized recommendation schemes.

[0018] As a preferred option, in clinical use, the operator only needs to input the operating parameters, and the system can complete the temperature rise prediction, risk assessment and parameter suggestion output in real time, so as to realize the rapid formulation and optimization of safe operation strategies.

[0019] The beneficial effects of this invention are:

[0020] 1. By training nine machine learning models in parallel and fusing feature importance, the prediction accuracy reached 87.22% (AUC = 0.9649), which can accurately identify high-risk operation combinations and provide reliable risk warnings for clinical practice.

[0021] 2. By collecting temperature rise data of 300 parameter combinations through a standardized high-throughput experimental platform, the data utilization rate is increased by 3-5 times, avoiding the problem of poor extrapolation of conclusions caused by insufficient parameter combinations in traditional experimental designs.

[0022] 3. By integrating tree-based models, SHAP values, and permutation importance analysis, this study systematically evaluates the interaction between bur parameters, operating parameters, cooling conditions, and tooth characteristics, providing a more comprehensive and systematic analysis of temperature rise risk than traditional univariate analysis.

[0023] 4. Based on the reverse reasoning module, it can recommend a safe parameter range (such as rotation speed ≤ 6000 rpm + continuous cooling + remaining dentin thickness ≥ 1.0 mm) for specific clinical scenarios within 10 seconds, covering 85% of clinical scenarios and achieving optimal temperature rise control.

[0024] 5. Clinical trials have shown that the combination of parameters recommended in this invention significantly improves treatment safety. At the same time, the automated feature engineering and model training process greatly shortens data processing time, with single-sample prediction time taking only a few seconds, thus significantly improving the efficiency of clinical decision-making. Attached Figure Description

[0025] Figure 1 The diagram shown illustrates the workflow of this invention.

[0026] Figure 2 The diagram shows the workflow of the root canal preparation temperature rise prediction method of the present invention.

[0027] Figure 3 The diagram shown is a bar chart illustrating the feature importance of the six models of this invention;

[0028] Figure 4 The diagram shown is a bar chart illustrating the average feature importance of the present invention.

[0029] Figure 5 The figures shown are ROC curves for the nine models of this invention.

[0030] Figure 6The diagram shows the Precision-Recall curves for the nine models of this invention.

[0031] Figure 7 The diagram shown is a confusion matrix representing the accuracy of the optimal model ET of this invention.

[0032] Figure 8 The diagram illustrates the workflow of the temperature rise risk prediction and classification method for real-time clinical decision support of the present invention.

[0033] Figure 9 The diagram shown is a bar chart illustrating the predicted probability distribution for different risk levels of this invention.

[0034] Figure 10 The present invention presents a low-risk parameter combination;

[0035] Figure 11 The image shown is a radar chart of the recommended parameter combinations of the present invention.

[0036] Figure 12 This demonstrates the contribution of each feature of the present invention to temperature rise prediction. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. 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.

[0038] This invention provides an embodiment: a system for predicting and optimizing the pre-treatment temperature rise of root canal post channels, including a user input module, a temperature rise prediction module, a risk assessment module, a parameter optimization suggestion module, and an auxiliary decision-making module.

[0039] Please see Figure 1This invention rapidly outputs temperature rise risk results through an established predictive model, providing a basis for adjusting operational strategies and improving treatment safety and operational efficiency. First, it receives actual clinical operational parameters input by the operator, including instrument parameters, rotation speed, cooling conditions, etc. Then, it calls the optimal extreme random tree (ET) model trained based on large-sample experimental data to automatically predict the temperature rise range under the input parameter combination. Based on the temperature rise prediction results, combined with clinical safety thresholds, it determines the risk level and presents it in the form of color and grade labels. When the prediction result is high-risk, it automatically outputs suggested parameter combinations for adjustment to assist the operator in optimizing the operation plan. Combining clinical goals (such as setting an upper limit for temperature rise, root canal wall thickness, etc.), it performs inverse analysis based on the established model, outputting recommended parameter combinations within the optimal or acceptable range for personalized reference.

[0040] Furthermore, in Example 1, the construction and validation of a machine learning-based root canal preparation temperature rise prediction model are described:

[0041] Please see Figure 2 The present invention provides an embodiment: a root canal preparation temperature rise prediction method based on machine learning, which includes the following steps:

[0042] S1, Data Collection and Preprocessing:

[0043] Experimental data were collected using a high-precision in vitro experimental platform. A root canal filling simulation model was established according to the ISO 3630-1 standard. Under different operating parameters, the temperature rise of the root canal wall was recorded in real time at a frequency of 100Hz using an infrared thermal imager.

[0044] The parameters collected included: rotation speed (8000rpm / 12000rpm / 20000rpm), dentin thickness (1.4mm / 1.8mm / 2.3mm), cooling method (0=no cooling, 1=water cooling), taper (0% / 2% / 6% / 8%), working tip diameter (0.5mm / 0.7mm / 0.9mm), and single operation time, totaling 300 sets. After removing outliers, each set was repeated 5 times and the average was taken.

[0045] For continuous variables (such as rotational speed, thickness, time, etc.), min-max normalization is used, and for categorical variables (such as cooling, taper), one-hot encoding is performed to finally construct a standardized 9-dimensional feature vector;

[0046] The root canal temperature rise was classified into three levels: low risk (temperature rise < 8℃, classification label 0), medium risk (8 ≤ temperature rise ≤ 10℃, classification label 1), and high risk (temperature rise > 10℃, classification label 2).

[0047] S2, Multi-model feature importance analysis:

[0048] Nine features were extracted from the raw data, including rotational speed, thickness, cooling, taper, working tip diameter, and operating time.

[0049] The importance of features was evaluated using six mainstream machine learning models (XGBoost, Random Forest, Gradient Boosting, Logistic Regression, Decision Tree, and ExtraTrees).

[0050] The feature importance values ​​output by each model are normalized using the following maximum-minimum normalization formula:

[0051]

[0052] The combined weight of each feature is calculated by merging the normalized average scores of the six models:

[0053]

[0054] Please see Figure 3 and Figure 4 The results showed that the top 9 characteristics were: rotational speed, blade spacing, taper, blade width, root canal wall thickness, working tip diameter, water cooling, continuous operation time, and interval time.

[0055] S3, Model Training and Evaluation:

[0056] Construct a candidate model pool containing 9 models (LR, DT, RF, ET, GB, XGB, KNN, SVM, MLP);

[0057] Hierarchical K-fold cross-validation (5 folds) and nested cross-validation (inner 3-fold parameter tuning) are used to improve the stability of model evaluation; evaluation metrics include classification accuracy, F1 score, multi-class macro average AUC (Macro AUC), ROC curve, precision-recall curve, and confusion matrix.

[0058] Please see Figure 5-7 The evaluation results show that the ExtraTreesClassifier model performs best in overall performance. The optimal parameters are criterion='entropy', max_depth=None, min_samples_leaf=2, min_samples_split=5, n_estimators=100. On the test set, it achieves the following metrics: accuracy 86.67% and macro-average AUC=0.956, as detailed in Table 1.

[0059] Table 1. Accuracy and F1 score of the 9 models

[0060] ET 86.67％ 87.1％ 87.3％ 88.4％ LR 85.20％ 87.4％ 85.0％ 89.7％ DT 78.67％ 86.6％ 79.7％ 66.7％ RF 86.00％ 82.4％ 86.2％ 89.9％ GB 87.33％ 85.0％ 86.3％ 87.6％ XGB 86.00％ 84.5％ 86.0％ 87.6％ KNN 86.00％ 85.4％ 84.7％ 88.4％ SVM 86.67％ 87.1％ 86.0％ 87.1％ MLP 87.00％ 86.3％ 85.3％ 88.0％

[0061] S4, Model Output and Application:

[0062] After training the best-performing ExtraTrees model, it is saved as a Python inference module. Users can quickly predict the temperature rise level during the corresponding root canal preparation process by inputting parameter combinations (such as rotation speed, taper, cooling method, etc.), thus achieving preoperative risk assessment and parameter optimization.

[0063] Example 2: Machine Learning-Based Method for Predicting and Grading Temperature Rise Risk in Root Canal Preparation

[0064] Please see Figure 8 This embodiment proposes a method for predicting and classifying the risk of temperature rise that can be used for real-time clinical decision support. It uses a pre-trained ExtraTrees classification model, and the specific steps are as follows:

[0065] S1, Input operation parameters and perform standardization processing:

[0066] Taking clinical simulation as an example, the input parameters are:

[0067] The rotation speed is 12000rpm, the dentin thickness is 1.4mm, the cooling is on (1), the operation time is 3 seconds, the interval time is 4s, the bur taper is 2%, the blade spacing is 0.2mm, the blade width is 0.18mm, and the working tip diameter is 0.8mm.

[0068] After input, the system standardizes the data (using the same StandardScaler object as in the training phase) to ensure that the feature order and encoding method are consistent with the model.

[0069] S2, using the trained ExtraTrees classification model for risk prediction:

[0070] Load the saved ExtraTreesClassifier model and make predictions on the standardized new input data:

[0071] The model outputs a classification label: "low risk".

[0072] The corresponding predicted probability is:

[0073] Low risk (Category 0): 61.6%;

[0074] Medium risk (Category 1): 38.4%;

[0075] High risk (Category 2): 0.0%.

[0076] Please see Figure 9 It displays a bar chart showing the predicted probability distribution for different risk levels.

[0077] S3, Automatic System Warnings and Suggestions:

[0078] If the predicted risk level of temperature rise is "high risk" or the model outputs a high probability warning, the system will automatically generate a warning message and suggested parameter adjustments: Warning: The predicted risk of temperature rise is HIGH. Please consider enabling water cooling or reducing the rotation speed.

[0079] S4, Instantaneous Parameter Substitution Analysis (Non-Monte Carlo):

[0080] To improve the clinical interpretability of model predictions, the system supports one-to-one variable replacement for the current input parameters and quickly predicts changes in temperature rise risk based on the model. This function is used to quickly assess the impact of key variables before or during surgery: such as partially or completely adjusting parameters like rotation speed and water cooling application to obtain risk assessment prediction indicators under the new operating parameters.

[0081] This analysis does not rely on full parameter space sampling and is suitable for rapid clinical decision support rather than systematic recommendations.

[0082] Example 3: A strategy recommendation method for reversibly generating safe operating parameter combinations:

[0083] This embodiment proposes a machine learning-based inverse strategy recommendation method to generate a set of optimal operating parameter combinations while meeting the temperature rise safety threshold. The method includes the following steps:

[0084] S1, Target Setting:

[0085] To prevent excessive temperature rise during root canal preparation that could lead to pulp or dentin damage, a clinical safety target is first set: given a known root canal wall thickness of 1.4 mm, ensuring that the temperature rise during the procedure is below 8°C is considered a low-risk level (category 0).

[0086] S2, Monte Carlo simulation sampling and temperature rise risk prediction:

[0087] Based on the already trained ExtraTrees classification model (model parameters are as follows: criterion='entropy', :max_depth=None, min_samples_leaf=2, min_samples_split=5, n_estimators=100), the following operational parameter space is constructed:

[0088] Rotation speed: 8000, 12000, 20000 rpm;

[0089] Blade spacing: 0.2–0.5 mm, uniform sampling;

[0090] Taper: 0%, 2%, 6%, 8%;

[0091] Blade width: 0.05–0.2 mm, uniform sampling;

[0092] Dentin thickness: fixed at 1.4 mm;

[0093] Working tip diameter: 0.5, 0.7, 0.9 mm;

[0094] Water cooling: 0 (off), 1 (on);

[0095] Continuous operation time: 1.5, 2.0, 2.5, 3.0 seconds;

[0096] Pause time: 0, 1, 2 seconds.

[0097] Using the Monte Carlo method, 1000 parameter combinations were randomly sampled from the aforementioned parameter space, and each sample was preprocessed using standardization. Then, the ExtraTrees model was called to predict the temperature rise risk level for each sample, and low-risk combinations with prediction results of category 0 (i.e., temperature rise less than 8℃) were selected, resulting in 75 such combinations.

[0098] S3, Visual Analysis: Recommended Parameter Distribution:

[0099] Statistical analysis was performed on the 75 low-risk parameter combinations selected, and the frequency distribution of each key operational parameter was extracted. The results showed that:

[0100] Lower engine speeds (such as 8000 rpm) occur significantly more frequently in low-risk combinations;

[0101] With water cooling enabled (cooling=1), the probability of safety is significantly improved.

[0102] Shorter continuous operation times (e.g., 1.5–2 seconds) are more conducive to controlling temperature rise;

[0103] Smaller blade spacing (0.2–0.3 mm) and smaller blade width (0.05–0.1 mm) are more common in low-temperature combinations.

[0104] Please see Figure 10 The above distribution data, presented in bar charts and box plots, can provide data support for doctors in formulating preoperative plans.

[0105] S4, Parameter Sensitivity Analysis (SHAP):

[0106] Building upon the multi-model feature importance analysis in Example 1, this invention further incorporates SHAP value interpretation analysis based on the ExtraTrees model. By employing two different perspectives on feature evaluation methods, the consistency and stability of the parameters' influence on temperature rise prediction are verified, enhancing the model's credibility and clinical guidance significance. Results show that:

[0107] The parameters that contribute the most to temperature rise prediction are: blade spacing, water cooling status, continuous operation time, and rotational speed.

[0108] Secondly: working tip diameter, interval time, cutting width, and taper;

[0109] Because the dentin thickness is fixed in this embodiment, the feature importance score for dentin width in this model is 0. The ranking of all features is as follows: Figure 9 As shown.

[0110] Figure 11 and Figure 12 The accompanying explanatory graphs help clinicians gain a deeper understanding of the sensitivity of each parameter to temperature rise control.

[0111] S5, Recommended Safety Parameters Example:

[0112] Based on the screening results, the system can recommend frequently occurring low-risk parameter combinations to doctors, such as:

[0113] Recommended parameter combination example:

[0114] rotation_speed: 8000.000000;

[0115] blade_spacing: 0.201728;

[0116] Taper: 8.000000;

[0117] blade_width: 0.051248;

[0118] dentin_thickness: 1.400000;

[0119] tip_diameter: 0.500000;

[0120] Cooling: 1.000000;

[0121] continuous_time: 1.500000;

[0122] pause_time: 1.000000.

[0123] Using the above combination helps to achieve personalized, low-risk parameter presets before the procedure, improves the safety and controllability of the root canal treatment process, and reduces the risk of damage to tooth tissues due to overheating during the procedure.

[0124] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A system for predicting and optimizing the temperature rise during root canal preparation, characterized in that, Including: User input module: Used to receive actual clinical operation parameters input by the operator, including instrument parameters, rotation speed and cooling conditions; Temperature rise prediction module: used to call the optimal extreme random tree model trained based on large sample experimental data to automatically predict the temperature rise range under the combination of input parameters; Risk assessment module: Based on the temperature rise prediction results and combined with clinical safety thresholds, the risk level is determined and presented in the form of color and graded labels; Parameter optimization suggestion module: used to automatically output suggested parameter combinations for adjustment when the prediction result is high risk; Decision Support Module: This module combines clinical goals with backpropagation based on established models to output recommended parameter combinations within the optimal or acceptable range for personalized reference. After determining the clinical goal, the auxiliary decision-making module calls the ET model and combines it with the Monte Carlo sampling method to screen out parameter combinations with low risk prediction results in the feature space, further statistically analyzes the distribution of its key features, and outputs personalized recommendation schemes. The extreme random tree model in the temperature rise prediction module and the auxiliary decision-making module is the optimal model selected after training and comparing nine machine learning models in parallel based on multi-dimensional parameter combination temperature rise data collected by a standardized high-throughput experimental platform. The multi-dimensional parameters include at least rotational speed, dentin thickness, cooling method, taper, working tip diameter, blade spacing, blade width, continuous operation time, and interval time. Hierarchical K-fold cross-validation and nested cross-validation are used for model evaluation and parameter tuning during model training.

2. The system for predicting and optimizing the temperature rise value of root canal post preparation according to claim 1, characterized in that: The risk levels output by the temperature rise prediction module include low risk, medium risk, and high risk. The high-risk result will trigger the system to automatically generate a real-time warning and simultaneously display the recommended parameter combination for adjustment.

Images