Construction method and system of occupational noise hearing loss risk screening model

Abstract

The invention relates to the technical field, and discloses a construction method and system of an occupational noise hearing loss risk screening model, which are independent of hearing evaluation and direct noise exposure measurement and do not depend on traditional hearing detection equipment and data thereof. The method comprises the following steps: constructing a high-precision occupational noise hearing loss risk screening model by using demographic characteristic data, biochemical metabolism index data and hematology index data of occupational noise hearing loss people and noise exposure hearing normal people and a machine learning algorithm; the model can efficiently and accurately screen occupational noise-induced hearing loss risks, assists in realizing accurate occupational noise-induced hearing loss diagnosis and treatment decision, effectively saves manpower and material resources, and has a huge application prospect in screening and early prevention of noise-induced hearing loss.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a method and system for constructing a risk screening model for occupational noise-induced hearing loss. Background Art

[0002] Occupational noise-induced hearing loss (ONIHL) is a common occupational health problem in industries such as industrial production, construction, and manufacturing. Long-term exposure to high-intensity noise can lead to irreversible hearing damage, severely impacting workers' ability to work and quality of life. Currently, the diagnosis of ONIHL relies primarily on pure-tone audiometry (PTA), which requires multiple hearing tests to assess changes in a patient's hearing threshold. However, this traditional diagnostic method has significant limitations. First, it relies on high-precision audiometric equipment, which is expensive to purchase and maintain. Second, the testing process requires highly trained professionals, which increases time costs and limits its widespread application in resource-limited settings. These factors collectively make large-scale screening and early intervention for ONIHL difficult to implement effectively.

[0003] In recent years, the application of machine learning technology in healthcare has provided new possibilities for risk screening for ONIHL. Previous studies have attempted to use algorithmic models to screen for the impact of noise exposure on hearing. However, most of these models still rely on individual hearing test data (such as PTA results) or precise noise exposure monitoring data (such as noise dosimeter records) for training and screening, failing to truly break free from the constraints of traditional detection methods. This dependence limits the models' applicability and makes them difficult to promote in real-world work environments that lack comprehensive hearing testing facilities or long-term noise monitoring data.

[0004] Therefore, developing a risk screening model for occupational noise-induced hearing loss that is independent of traditional audiometric testing equipment and can efficiently screen for the risk of ONIHL is crucial for early screening and intervention for ONIHL in occupational populations. This model is expected to reduce diagnostic costs, improve screening efficiency, and provide a more convenient and accessible decision-making support tool for occupational health management. Summary of the Invention

[0005] The present invention provides a method for constructing an occupational noise-induced hearing loss risk screening model and a system thereof, so as to overcome the defect that existing occupational noise-induced hearing loss assessment methods are difficult to meet the large-scale screening needs of people exposed to occupational noise.

[0006] The present invention provides a method for constructing a risk screening model for occupational noise-induced hearing loss, comprising: Obtain demographic data, biochemical metabolic index data, and hematological index data of the target population, including those with occupational noise-induced hearing loss and those with normal hearing who are exposed to noise; Based on the demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population, a machine learning algorithm is used to enable the model to learn the relationship between the demographic characteristics data, biochemical metabolic index data, and hematological index data of people with occupational noise-induced hearing loss and people with normal hearing who are exposed to noise, and obtain an occupational noise-induced hearing loss risk screening model.

[0007] According to a method for constructing an occupational noise-induced hearing loss risk screening model provided by the present invention, the demographic characteristic data includes gender and / or age.

[0008] According to a method for constructing an occupational noise-induced hearing loss risk screening model provided by the present invention, the biochemical metabolic index data includes any one of the following or any combination thereof: protein metabolism index data, sugar metabolism index data, lipid metabolism index data, liver function index data, and kidney function index data.

[0009] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, protein metabolism index data includes any one of the following items or any combination thereof: total protein (TP), albumin (ALB), globulin (GLB), and albumin / globulin ratio (A / G).

[0010] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, the glucose metabolism index data includes any one of the following items or any combination thereof: glucose (GLU), glucose / HDL ratio (GLU / HDL), and triglyceride-glucose index (TyG).

[0011] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, lipid metabolism indicator data includes any one of the following or any combination thereof: total cholesterol (CHO), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and platelet / HDL ratio (PLT / HDL).

[0012] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, liver function index data includes any one of the following or any combination thereof: total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), alanine aminotransferase (ALT), and aspartate aminotransferase (AST).

[0013] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, renal function index data includes any one of the following or any combination thereof: blood urea nitrogen (BUN), serum creatinine (Scr), uric acid (UA), and estimated glomerular filtration rate (eGFR).

[0014] According to a method for constructing an occupational noise-induced hearing loss risk screening model provided by the present invention, the hematological index data includes any one of the following or any combination thereof: red blood cell system index data, white blood cell system index data, and platelet system index data.

[0015] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, the red blood cell system indicator data include any one of the following items or any combination thereof: hemoglobin (Hb), red blood cell count (RBC), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), red blood cell distribution width coefficient of variation (RDW-CV), and red blood cell distribution width standard deviation (RDW-SD).

[0016] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, the leukocyte system indicator data includes any one of the following items or any combination thereof: white blood cell count (WBC), neutrophil count (GRANC), lymphocyte count (LYMPHC), monocyte count (MOC), eosinophil count (EOC), basophil count (BAC), neutrophil percentage (GRANP), lymphocyte percentage (LYMPHP), monocyte percentage (MOP), eosinophil percentage (EOP), basophil percentage (BAP), and neutrophil / lymphocyte ratio (NLR).

[0017] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, platelet system indicator data includes any one of the following or any combination thereof: platelet count (PLT), mean platelet volume (MPV), platelet distribution width (PDW), platelet packed cell (PCT), and platelet / lymphocyte ratio (PLR).

[0018] According to the present invention, a method for constructing a risk screening model for occupational noise-induced hearing loss comprises the following steps after obtaining demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population: Oversampling processing was performed on the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population.

[0019] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, the machine learning algorithm includes any one of the following or any combination thereof: logistic regression (LR), random forest (RF), support vector machine (SVM), K-nearest neighbor (KNN), and extreme gradient boosting (XGBoost).

[0020] According to the present invention, a method for constructing a risk screening model for occupational noise-induced hearing loss is provided. The method comprises: using a machine learning algorithm based on demographic characteristic data, biochemical metabolic index data, and hematological index data of a target population, so that the model learns the relationship between the demographic characteristic data, biochemical metabolic index data, and hematological index data of a population with occupational noise-induced hearing loss and a population with normal hearing who are exposed to noise, thereby obtaining a risk screening model for occupational noise-induced hearing loss. The method comprises: Based on the demographic characteristics, biochemical metabolic index data, and hematological index data of the target population, different machine learning algorithms were used to train several candidate occupational noise-induced hearing loss risk screening models; Using evaluation indicators, the performance of several candidate occupational noise-induced hearing loss risk screening models was evaluated, and the optimal occupational noise-induced hearing loss risk screening model was obtained as the first occupational noise-induced hearing loss risk screening model. Among them, the machine learning algorithm used by the optimal occupational noise-induced hearing loss risk screening model is extreme gradient boosting.

[0021] According to the present invention, a method for constructing a risk screening model for occupational noise-induced hearing loss is provided. The method comprises: using a machine learning algorithm based on demographic characteristic data, biochemical metabolic index data, and hematological index data of a target population, so that the model learns the relationship between the demographic characteristic data, biochemical metabolic index data, and hematological index data of a population with occupational noise-induced hearing loss and a population with normal hearing who are exposed to noise, thereby obtaining a risk screening model for occupational noise-induced hearing loss. The method comprises: During the training process of the occupational noise-induced hearing loss risk screening model, five-fold cross-validation was used to prevent overfitting.

[0022] According to the present invention, a method for constructing a risk screening model for occupational noise-induced hearing loss is provided. The method comprises: using a machine learning algorithm based on demographic characteristic data, biochemical metabolic index data, and hematological index data of a target population, so that the model learns the relationship between the demographic characteristic data, biochemical metabolic index data, and hematological index data of a population with occupational noise-induced hearing loss and a population with normal hearing who are exposed to noise, thereby obtaining a risk screening model for occupational noise-induced hearing loss. The method comprises: The demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population were screened for importance to obtain the characteristic data related to the risk screening of occupational noise-induced hearing loss of the target population; The occupational noise-induced hearing loss risk screening-related characteristic data of the target population are used to replace the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, and the first occupational noise-induced hearing loss risk screening model is retrained to obtain the second occupational noise-induced hearing loss risk screening model; The performance of the secondary occupational noise-induced hearing loss risk screening model was evaluated using evaluation indicators. The performance evaluation results of the first occupational noise-induced hearing loss risk screening model were compared with the performance evaluation results of the second occupational noise-induced hearing loss risk screening model, and the model with better performance evaluation results was used as the final occupational noise-induced hearing loss risk screening model.

[0023] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, the importance screening of demographic characteristic data, biochemical metabolic index data, and hematological index data of a target population is performed to obtain characteristic data related to the risk screening of occupational noise-induced hearing loss of the target population, including: Obtaining manually selected feature data based on demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population, and conducting importance analysis related to occupational noise-induced hearing loss risk screening; Based on the demographic characteristics, biochemical metabolic index data, and hematological index data of the target population, principal component analysis (PCA) was used to conduct an importance analysis related to the risk screening of occupational noise-induced hearing loss, and PCA selected feature data were obtained; Based on the demographic characteristics, biochemical metabolic index data, and hematological index data of the target population, the maximum relevance and minimum redundancy method (mRMR) was used to conduct an importance analysis related to the risk screening of occupational noise-induced hearing loss, and the mRMR selection feature data was screened. Manually selected feature data, PCA selected feature data, and mRMR selected feature data were used as relevant feature data for occupational noise-induced hearing loss risk screening of the target population, so as to be used for reconstructing the first occupational noise-induced hearing loss risk screening model.

[0024] According to the present invention, a method for constructing a risk screening model for occupational noise-induced hearing loss is provided. The method comprises: using a machine learning algorithm based on demographic characteristic data, biochemical metabolic index data, and hematological index data of a target population, so that the model learns the relationship between the demographic characteristic data, biochemical metabolic index data, and hematological index data of a population with occupational noise-induced hearing loss and a population with normal hearing who are exposed to noise, thereby obtaining a risk screening model for occupational noise-induced hearing loss. The method comprises: The demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population were replaced by the manually selected feature data, PCA selected feature data, and mRMR selected feature data of the target population, respectively, and the first occupational noise-induced hearing loss risk screening model was retrained to obtain the manually selected feature-occupational noise-induced hearing loss risk screening model, the PCA selected feature-occupational noise-induced hearing loss risk screening model, and the mRMR selected feature-occupational noise-induced hearing loss risk screening model; Decision curve analysis was used to evaluate the net benefits of the manually selected features-occupational noise-induced hearing loss risk screening model, PCA selected features-occupational noise-induced hearing loss risk screening model, and mRMR selected features-occupational noise-induced hearing loss risk screening model at different thresholds.

[0025] According to a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention, the evaluation indicators include any one of the following or any combination thereof: sensitivity, specificity, balanced accuracy, area under the receiver operating characteristic curve (AUC), area under the precision-recall curve (PR-AUC), F1-score, and precision.

[0026] The present invention also provides an auxiliary assessment system for occupational noise-induced hearing loss risk, comprising: A data receiving module, configured to receive screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in any of the above-mentioned methods for constructing a risk screening model for occupational noise-induced hearing loss, or any combination thereof; An occupational noise-induced hearing loss risk screening module, configured to obtain an occupational noise-induced hearing loss risk screening result for a subject based on screening factor data of the subject and an occupational noise-induced hearing loss risk screening model obtained by any of the above-mentioned methods for constructing an occupational noise-induced hearing loss risk screening model; The data output module is used to output the risk screening result of occupational noise-induced hearing loss of the subject to be tested to at least one terminal.

[0027] It should be noted that a terminal refers to an input and output device connected to a computer system. Depending on their functions, terminals can be divided into various types: smart terminals or intelligent terminals, dumb terminals, interactive terminals or online terminals. Terminals can specifically be various mobile communication devices, such as mobile phones, tablet computers, etc. The purpose of this article is to provide users with the function of inputting and outputting data.

[0028] The present invention also provides an electronic device comprising a processor and a memory storing a computer program, wherein when the processor executes the computer program, the method for constructing any of the above-mentioned occupational noise-induced hearing loss risk screening models is implemented.

[0029] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for constructing any of the above-mentioned occupational noise-induced hearing loss risk screening models is implemented.

[0030] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute any of the above-mentioned methods for constructing an occupational noise-induced hearing loss risk screening model.

[0031] The present invention provides a method and system for constructing a risk screening model for occupational noise-induced hearing loss. These methods are independent of hearing assessment and direct noise exposure measurement, and do not rely on traditional hearing testing equipment and its data. They utilize demographic data, biochemical metabolic index data, hematological index data, and machine learning algorithms of people with occupational noise-induced hearing loss and people with normal noise-exposed hearing to construct a high-precision risk screening model for occupational noise-induced hearing loss. This model can efficiently and accurately screen for the risk of occupational noise-induced hearing loss, assist in the accurate diagnosis and treatment decision-making of occupational noise-induced hearing loss, effectively save manpower and material resources, and has great application prospects in the screening and early prevention of noise-induced hearing loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 This is a flow chart of a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention.

[0034] Figure 2 Performance of occupational noise-induced hearing loss risk screening models trained using logistic regression (LR), random forest (RF), support vector machine (SVM), K-nearest neighbor (KNN), and extreme gradient boosting (XGBoost) on the D1 validation set. A: ROC curve; B: Precision-recall curve.

[0035] Figure 3 Figure 2 shows feature selection for the final model using PCA, manual screening, and mRMR. A: ROC curve of the model constructed based on features selected using PCA, manual screening, and mRMR on the validation set of dataset D1; B: Precision-recall curve of the aforementioned model; C: Comparison of sensitivity, specificity, and balanced accuracy of the model before and after feature selection on the validation set; D: ROC curve of the model constructed using the selected features on the test set of dataset D2; E: Precision-recall curve of the aforementioned model; F: Comparison of other evaluation metrics of the model on the independent test set D2.

[0036] Figure 4 Shows the decision curve analysis (DCA) results of models built based on three different feature selection methods.

[0037] Figure 5 A structural diagram of an auxiliary assessment of occupational noise-induced hearing loss risk provided by the invention.

[0038] Figure 6 This is a schematic structural diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the embodiments described are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0040] Figure 1 This is a flow chart of a method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention. The method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention can be executed by any applicable terminal-side device or network-side device.

[0041] See also Figure 1 The present invention provides a method for constructing a risk screening model for occupational noise-induced hearing loss, which may include: S110. Obtain demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, wherein the target population includes people with occupational noise-induced hearing loss and people with normal hearing who are exposed to noise.

[0042] In one embodiment, the demographic characteristic data includes gender and age.

[0043] Biochemical metabolism index data include: protein metabolism index data, sugar metabolism index data, lipid metabolism index data, liver function index data, and kidney function index data.

[0044] Among them, protein metabolism index data include: total protein (TP), albumin (ALB), globulin (GLB), albumin / globulin ratio (A / G); carbohydrate metabolism index data include: glucose (GLU), glucose / HDL ratio (GLU / HDL), triglyceride-glucose index (TyG); lipid metabolism index data include: total cholesterol (CHO), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), platelet / HDL ratio (PLT / HDL); liver function index data include: total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST); renal function index data include: blood urea nitrogen (BUN), serum creatinine (Scr), uric acid (UA), estimated glomerular filtration rate (eGFR).

[0045] Hematology index data include: red blood cell system index data, white blood cell system index data, and platelet system index data.

[0046] Among them, the red blood cell system index data include: hemoglobin (Hb), red blood cell count (RBC), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), red blood cell distribution width variation coefficient (RDW-CV), red blood cell distribution width standard deviation (RDW-SD); the white blood cell system index data include: white blood cell count (WBC), neutrophil count (GRANC), lymphocyte count (LYMPHC), monocyte count (MOC), Eosinophil count (EOC), basophil count (BAC), neutrophil percentage (GRANP), lymphocyte percentage (LYMPHP), monocyte percentage (MOP), eosinophil percentage (EOP), basophil percentage (BAP), neutrophil / lymphocyte ratio (NLR); platelet system index data include: platelet count (PLT), mean platelet volume (MPV), platelet distribution width (PDW), platelet packed volume (PCT), platelet / lymphocyte ratio (PLR).

[0047] After obtaining the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population can be oversampled to increase the number of minority class samples and reduce the impact of class imbalance during model training.

[0048] S120. Based on the demographic characteristic data, biochemical metabolic indicator data, and hematological indicator data of the target population, a machine learning algorithm is used to enable the model to learn the relationship between the demographic characteristic data, biochemical metabolic indicator data, and hematological indicator data of the population with occupational noise-induced hearing loss and the population with normal hearing who are exposed to noise, so as to obtain a risk screening model for occupational noise-induced hearing loss.

[0049] In one embodiment, the machine learning algorithm includes any one of the following or any combination thereof: Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Extreme Gradient Boosting (XGBoost).

[0050] Logistic regression is a form of linear regression that uses the Sigmoid function to convert the output into a probability for classification. Random forests consist of multiple independently trained decision trees, and the final screening results are obtained through a voting mechanism between these trees, thereby reducing the risk of overfitting. Support vector machines classify samples by identifying the optimal hyperplane in the feature space and are capable of processing nonlinearly separable data. K-nearest neighbor is an instance-based learning method that classifies samples based on the degree of proximity to their k nearest neighbors. It is particularly suitable for small data sets and is relatively simple to implement. XGBoost is an ensemble method based on decision trees that improves model performance through a gradient boosting framework. It constructs decision trees in an iterative manner to minimize model errors and is particularly suitable for processing large-scale, high-dimensional data sets because of its strong generalization ability and computational efficiency. In this embodiment, different k values were tested in the K-nearest neighbor model, and the optimal performance occurred when k=11.

[0051] In one embodiment, S120 may include: Based on the demographic characteristics, biochemical metabolic index data, and hematological index data of the target population, different machine learning algorithms were used to train several candidate occupational noise-induced hearing loss risk screening models; Using the evaluation indicators, the performance of several candidate occupational noise-induced hearing loss risk screening models is evaluated, and the optimal occupational noise-induced hearing loss risk screening model is obtained as the first occupational noise-induced hearing loss risk screening model; During the training process of the occupational noise-induced hearing loss risk screening model, five-fold cross-validation was used to prevent overfitting.

[0052] In one embodiment, the evaluation metrics include any one of the following or any combination thereof: sensitivity, specificity, balanced accuracy, area under the receiver operating characteristic curve (AUC), area under the precision-recall curve (PR-AUC), F1-score, and precision, which are defined as follows:

[0053]

[0054]

[0055]

[0056]

[0057] True positives (TP) refer to the number of cases of noise-induced hearing loss; false positives (FP) refer to the number of normal subjects incorrectly classified as having occupational noise-induced hearing loss; true negatives (TN) refer to the number of healthy subjects correctly classified as normal; and false negatives (FN) refer to the number of cases of occupational noise-induced hearing loss incorrectly classified as normal. All of the above indicators have a value range of 0 to 1.

[0058] In one embodiment, S120 may include: The demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population were screened for importance to obtain the characteristic data related to the risk screening of occupational noise-induced hearing loss of the target population; The occupational noise-induced hearing loss risk screening-related characteristic data of the target population are used to replace the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, and the first occupational noise-induced hearing loss risk screening model is retrained to obtain the second occupational noise-induced hearing loss risk screening model; The performance of the secondary occupational noise-induced hearing loss risk screening model was evaluated using evaluation indicators. The performance evaluation results of the first occupational noise-induced hearing loss risk screening model were compared with the performance evaluation results of the second occupational noise-induced hearing loss risk screening model, and the model with better performance evaluation results was used as the final occupational noise-induced hearing loss risk screening model.

[0059] In one embodiment, importance screening can be performed manually or through principal component analysis (PCA) or maximum relevance minimum redundancy (mRMR) to obtain manually selected feature data, PCA selected feature data, and mRMR selected feature data. The manually selected feature data, PCA selected feature data, and mRMR selected feature data of the target population can then be used to replace the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, respectively, to retrain the first occupational noise-induced hearing loss risk screening model, thereby obtaining a manually selected feature-occupational noise-induced hearing loss risk screening model, a PCA selected feature-occupational noise-induced hearing loss risk screening model, and a mRMR selected feature-occupational noise-induced hearing loss risk screening model. Decision curve analysis can then be used to evaluate the net benefits of the manually selected feature-occupational noise-induced hearing loss risk screening model, the PCA selected feature-occupational noise-induced hearing loss risk screening model, and the mRMR selected feature-occupational noise-induced hearing loss risk screening model at different thresholds.

[0060] The following specific examples demonstrate the method for constructing a risk screening model for occupational noise-induced hearing loss provided by the present invention. The medical examination data (demographic characteristics, biochemical metabolic indicators, and hematological indicators of the target population) for this example were obtained from the occupational health records of the Shenzhen Institute of Occupational Disease Prevention and Treatment from January 2023 to July 2024. The data were divided into two parts, D1 and D2, in chronological order. D1 was used for model training, and D2 was used for independent validation.

[0061] The data preprocessing steps are as follows: Data cleaning: Remove samples containing erroneous or outliers.

[0062] Inclusion and exclusion criteria: Inclusion criteria: According to the Diagnosis of Occupational Noise-Induced Hearing Loss (GBZ49-2014): ① Noise exposure duration ≥ 3 years; ② Average hearing threshold of high frequencies (3000 Hz, 4000 Hz, 6000 Hz) in both ears ≥ 40 dB.

[0063] Exclusion criteria: pseudo-deafness, exaggerated hearing loss, drug-induced deafness, traumatic deafness, infectious deafness, hereditary deafness, Meniere's disease, sudden deafness, acoustic neuroma, and auditory neuropathy.

[0064] Grouping: Samples were divided into a group with occupational noise-induced hearing loss (positive samples) and a group with normal noise-exposed hearing (negative samples). After preprocessing, a total of 3,297 samples were retained, of which D1 and D2 contained 2,868 (107 positive) and 429 (53 positive), respectively.

[0065] Dataset partitioning: D1 is randomly sampled and divided into training set and test set in a ratio of 7:3, and D2 is used as an independent validation set.

[0066] Addressing class imbalance: Given the significant class imbalance across all datasets, we oversampled positive instances in the training set using the ovun.sample() function in the ROSE R package. This function randomly replicates minority class samples to achieve an equal number of positive and negative instances in the training set and a balanced class distribution. This approach effectively increases the number of minority class samples and mitigates the impact of class imbalance during model training. All datasets were Z-score normalized, with the mean and standard deviation used being derived from the training set data.

[0067] Based on the demographic characteristics data, biochemical metabolic index data, and hematological index data of the above-mentioned target population, logistic regression, random forest, support vector machine, K nearest neighbor, and extreme gradient boosting were used to train candidate occupational noise-induced hearing loss risk screening models. Figure 2The XGBoost algorithm achieved the highest AUC (0.942) and PR-AUC (0.791), a high Recall (0.875), and an F1-Score second only to RF, demonstrating excellent overall performance. The LR algorithm also performed well with a good AUC (0.923) and F1-Score (0.921), and a high Balanced Accuracy (0.896), demonstrating excellent overall performance. To maximize the risk of occupational noise-induced hearing loss, this example ultimately selected the XGBoost algorithm to further construct the screening model.

[0068] The five-fold cross-validation results on the training set showed an AUC of 0.999, a PR-AUC of 0.995, a sensitivity of 0.995, and a balanced accuracy of 0.998. In addition, the XGBoost model's reliable performance on the test set (AUC = 0.900, PR-AUC = 0.648) was also verified.

[0069] Although the screening model demonstrated relatively good performance with 48 features, there is still the possibility that redundant information or noisy features could adversely affect the decision-making process. To improve feature utilization and simplify the model, this example combined manual selection, principal component analysis (PCA), and minimal redundancy and maximum correlation (mRMR) methods to extract key features for the final model. During the manual selection process, features showing significant differences between positive and negative samples were first identified. To improve the stability of the screening model, features that caused significant collinearity were removed. Consequently, 16 features were ultimately retained. The number of feature subsets was also fixed to 16 to ensure consistency during PCA and mRMR analyses. Furthermore, feature selection was performed on the training set to reduce the risk of overfitting. Feature importance analysis facilitates interpretation of the screening model and helps identify features closely associated with occupational noise-induced hearing loss. In this context, the importance of each feature is quantified using its corresponding weight coefficient in the XGBoost model.

[0070] High correlations were observed between several pairs of features, such as MCH and RDW-CV, and GRANP and LYMPHP. This could introduce redundant information, potentially affecting the model's decision-making process and stability. Therefore, this example employed manual screening, principal component analysis (PCA), and maximum relevance minimum redundancy (mRMR) to determine the optimal features. Ultimately, each method identified 16 features, based on which the XGBoost model was rebuilt (see Table 1). PCA and mRMR methods both identified seven features, three of which overlapped in their respective top five important features: ALB, RDW-CV, and Scr.

[0071] Table S1 Feature subsets used to construct the final model by PCA, manual screening, and mRMR methods

[0072] The model constructed using mRMR and 16 features selected by PCA performed slightly better on the validation set than the model using all 48 features. Specifically, the PCA model achieved an AUC of 0.957 and a PR-AUC of 0.741, while the mRMR model also achieved an AUC of 0.957 and a PR-AUC of 0.720. In contrast, the model constructed based on manually selected features performed worse, with an AUC of 0.919 and a PR-AUC of 0.540 ( Figure 3 A, B). Similarly, the models constructed by mRMR and PCA also showed improvements in sensitivity and balanced accuracy on the validation set, with the maximum increase of 29.2% and 3.7%, respectively. In contrast, the models based on manual screening performed poorly on all evaluation indicators ( Figure 3 C).

[0073] This example further evaluated the performance of these models on an independent test set (D2). The test results showed that the manual screening model had an AUC of 0.830, the PCA model had an AUC of 0.837, and the mRMR model performed best with an AUC of 0.872 ( Figure 3 D). In terms of PR-AUC, the manual screening, PCA, and mRMR models were 0.524, 0.540, and 0.594, respectively, with mRMR again performing best ( Figure 3 E). The mRMR model also showed the highest performance in the test set in terms of sensitivity, specificity, and balanced accuracy ( Figure 3 F). Notably, the mRMR model achieved a minimum sensitivity of 75.5% on the D2 test set, while all specificity scores exceeded 78.0%. Overall, the mRMR model performed robustly on the independent test set, demonstrating that the selected core features are sufficient for identifying noise-induced hearing loss in noise-exposed workers.

[0074] To investigate which features contribute most to the risk of occupational noise-induced hearing loss, this example first used the maximum relevance and minimum redundancy (mRMR) method to select 16 important features. An XGBoost model was then constructed based on these features. These features were then ranked according to their weights in the XGBoost model. Furthermore, the importance of the features selected using principal component analysis (PCA) and manual screening was also ranked. The results showed that the top five features by importance were serum albumin (ALB), platelet distribution width (PDW), red blood cell distribution width (CV), serum creatinine (Scr), and lymphocyte percentage (LYMPHP). Further comparisons of the ONIHL group with the normal control group revealed significant differences in ALB, total protein (TP), age, RDW-CV, and PDW between the two groups. These findings were highly consistent with the top-ranked features in the XGBoost model, indicating a strong correlation between these indicators (ALB, PDW, RDW-CV, Scr, and LYMPHP) and ONIHL.

[0075] This example uses decision curve analysis (DCA) to evaluate the net benefits of the three models at different thresholds ( Figure 4 The mRMR model performed best across the entire threshold range, demonstrating a high and stable net benefit, indicating its superior value in clinical decision-making. When no intervention (None) or intervention was applied to all individuals (All), the net benefit was lower than that of the mRMR model across a wide range, demonstrating the superiority of this decision-making model.

[0076] Occupational noise-induced hearing loss (ONIHL) is a significant public health issue worldwide. Despite its complexity, ONIHL is a preventable disease. The U.S. Occupational Safety and Health Administration (OSHA) requires workers exposed to noise levels of 85 decibels or higher to implement hearing conservation programs to protect their hearing health in noisy work environments. Therefore, developing risk screening tools for ONIHL is crucial as a preliminary screening and prevention strategy for workers' occupational noise exposure. In this example, five machine learning (ML) algorithms were used to construct an occupational noise-induced hearing loss risk screening model using demographic, biochemical, and hematological data. The model demonstrated an area under the curve (AUC) exceeding 0.85 in both the validation and independent test datasets, with both accuracy and sensitivity exceeding 0.75. These results demonstrate that the occupational noise-induced hearing loss risk screening model can accurately identify individuals with ONIHL among noise-exposed workers.

[0077] In evaluating model performance on the validation set, the XGBoost model outperformed all other evaluated algorithms, achieving an AUC of 0.942 and a PR-AUC of 0.791. Furthermore, the XGBoost model maintained consistent performance on the test set, achieving an AUC of 0.900 and a PR-AUC of 0.648. Furthermore, our results demonstrate that the XGBoost model outperforms logistic regression (LR), random forest (RF), support vector machine (SVM), and k-nearest neighbor (KNN) in screening performance. This is consistent with previous research showing that traditional logistic regression typically exhibits lower AUC values in receiver operating characteristic (ROC) curve analysis, with higher screening errors and lower performance than more modern methods.

[0078] Traditional hearing loss diagnosis usually relies on audiometric tests, such as pure tone audiometry, which requires professional equipment and well-trained personnel, thereby increasing the time, cost and resources required for diagnosis. In addition, many published machine learning models still usually rely on individual hearing test data or direct measurement of noise exposure levels to build accurate screening models. This dependence increases the complexity of their practical application. In contrast, the model constructed in this embodiment is independent of hearing assessment and direct noise exposure measurement, and focuses on analyzing routine blood and biochemical indicators for screening. This method significantly reduces the dependence on professional equipment and data, thereby saving manpower and material resources, and provides an efficient and convenient alternative, which is expected to play a role in the screening and early prevention of noise-induced hearing loss. In addition, the model is expected to be expanded to screen the risks of other occupational or chronic diseases, thereby providing important support for maintaining and improving public health.

[0079] The following describes the auxiliary assessment system for occupational noise-induced hearing loss risk provided by the present invention. The auxiliary assessment system for occupational noise-induced hearing loss risk described below and the method for constructing the occupational noise-induced hearing loss risk screening model described above can be referenced to each other.

[0080] See also Figure 5 The present invention provides an auxiliary assessment system for occupational noise-induced hearing loss risk, which may include: A data receiving module, configured to receive screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in any of the above-mentioned methods for constructing a risk screening model for occupational noise-induced hearing loss, or any combination thereof; An occupational noise-induced hearing loss risk screening module, configured to obtain an occupational noise-induced hearing loss risk screening result for a subject based on screening factor data of the subject and an occupational noise-induced hearing loss risk screening model obtained by any of the above-mentioned methods for constructing an occupational noise-induced hearing loss risk screening model; The data output module is used to output the risk screening result of occupational noise-induced hearing loss of the subject to be tested to at least one terminal.

[0081] Figure 6 An example of a physical structure diagram of an electronic device is shown below. Figure 6 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to perform the following steps: Receiving screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in any of the above-mentioned methods for constructing a risk screening model for occupational noise-induced hearing loss, or any combination thereof; Obtaining an occupational noise-induced hearing loss risk screening result for the subject based on the screening factor data of the subject and the occupational noise-induced hearing loss risk screening model obtained by any of the above-mentioned methods for constructing an occupational noise-induced hearing loss risk screening model; The occupational noise-induced hearing loss risk screening result of the subject is output to at least one terminal.

[0082] Furthermore, the logic instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0083] In another aspect, the present invention further provides a computer program product, comprising a computer program, which may be stored on a non-transitory computer-readable storage medium, and wherein when the computer program is executed by a processor, the computer is capable of performing the following steps: Receiving screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in any of the above-mentioned methods for constructing a risk screening model for occupational noise-induced hearing loss, or any combination thereof; Obtaining an occupational noise-induced hearing loss risk screening result for the subject based on the screening factor data of the subject and the occupational noise-induced hearing loss risk screening model obtained by any of the above-mentioned methods for constructing an occupational noise-induced hearing loss risk screening model; The occupational noise-induced hearing loss risk screening result of the subject is output to at least one terminal.

[0084] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is configured to execute the following steps when executed by a processor: Receiving screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in any of the above-mentioned methods for constructing a risk screening model for occupational noise-induced hearing loss, or any combination thereof; Obtaining an occupational noise-induced hearing loss risk screening result for the subject based on the screening factor data of the subject and the occupational noise-induced hearing loss risk screening model obtained by any of the above-mentioned methods for constructing an occupational noise-induced hearing loss risk screening model; The occupational noise-induced hearing loss risk screening result of the subject is output to at least one terminal.

[0085] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0086] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for constructing a risk screening model for occupational noise-induced hearing loss, characterized in that: include: Obtain demographic data, biochemical metabolic index data, and hematological index data of the target population, including those with occupational noise-induced hearing loss and those with normal hearing who are exposed to noise; Based on the demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population, a machine learning algorithm is used to enable the model to learn the relationship between the demographic characteristics data, biochemical metabolic index data, and hematological index data of people with occupational noise-induced hearing loss and people with normal hearing who are exposed to noise, and obtain an occupational noise-induced hearing loss risk screening model.

2. The method for constructing a risk screening model for occupational noise-induced hearing loss according to claim 1, characterized in that: Demographic characteristic data include gender and / or age; biochemical metabolic index data include any one of the following or any combination thereof: protein metabolism index data, sugar metabolism index data, lipid metabolism index data, liver function index data, and kidney function index data; hematological index data include any one of the following or any combination thereof: red blood cell system index data, white blood cell system index data, and platelet system index data; and / or, Machine learning algorithms include any one of the following or any combination of them: logistic regression, random forest, support vector machine, K-nearest neighbors, extreme gradient boosting.

3. The method for constructing a risk screening model for occupational noise-induced hearing loss according to claim 2, characterized in that: Protein metabolism index data include any one of the following or any combination thereof: total protein, albumin, globulin, albumin / globulin ratio; carbohydrate metabolism index data include any one of the following or any combination thereof: glucose, glucose / HDL ratio, triglyceride-glucose index; lipid metabolism index data include any one of the following or any combination thereof: total cholesterol, triglycerides, high-density lipoprotein, low-density lipoprotein, platelet / HDL ratio; liver function index data include any one of the following or any combination thereof: total bilirubin, direct bilirubin, indirect bilirubin, alanine aminotransferase, aspartate aminotransferase; renal function index data include any one of the following or any combination thereof: blood urea nitrogen, serum creatinine, uric acid, estimated glomerular filtration rate.

4. The method for constructing a risk screening model for occupational noise-induced hearing loss according to claim 2, wherein: The red blood cell system indicator data include any one of the following items or any combination thereof: hemoglobin, red blood cell count and hematocrit, mean corpuscular volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, coefficient of variation of red blood cell distribution width, and standard deviation of red blood cell distribution width; the white blood cell system indicator data include any one of the following items or any combination thereof: white blood cell count, neutrophil count, lymphocyte count, monocyte count, eosinophil count, basophil count, neutrophil percentage, lymphocyte percentage, monocyte percentage, eosinophil percentage, basophil percentage, and neutrophil / lymphocyte ratio; the platelet system indicator data include any one of the following items or any combination thereof: platelet count, mean platelet volume, platelet distribution width, hematocrit, and platelet / lymphocyte ratio.

5. The method for constructing a risk screening model for occupational noise-induced hearing loss according to any one of claims 1 to 4, characterized in that: After obtaining the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, the following steps are included: Oversampling the demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population; and / or, The method uses a machine learning algorithm based on the demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population to enable the model to learn the relationship between the demographic characteristics data, biochemical metabolic index data, and hematological index data of the population with occupational noise-induced hearing loss and the population with normal hearing who are exposed to noise, thereby obtaining an occupational noise-induced hearing loss risk screening model, including: Based on the demographic characteristics, biochemical metabolic index data, and hematological index data of the target population, different machine learning algorithms were used to train several candidate occupational noise-induced hearing loss risk screening models; Using evaluation indicators, the performance of several candidate occupational noise-induced hearing loss risk screening models was evaluated, and the optimal occupational noise-induced hearing loss risk screening model was obtained as the first occupational noise-induced hearing loss risk screening model. Among them, the machine learning algorithm used by the optimal occupational noise-induced hearing loss risk screening model is extreme gradient boosting.

6. The method for constructing a risk screening model for occupational noise-induced hearing loss according to claim 5, characterized in that: The method uses a machine learning algorithm based on the demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population to enable the model to learn the relationship between the demographic characteristics data, biochemical metabolic index data, and hematological index data of the population with occupational noise-induced hearing loss and the population with normal hearing who are exposed to noise, thereby obtaining an occupational noise-induced hearing loss risk screening model, including: The demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population were screened for importance to obtain the characteristic data related to the risk screening of occupational noise-induced hearing loss of the target population; The occupational noise-induced hearing loss risk screening-related characteristic data of the target population are used to replace the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population, and the first occupational noise-induced hearing loss risk screening model is retrained to obtain the second occupational noise-induced hearing loss risk screening model; The performance of the secondary occupational noise-induced hearing loss risk screening model was evaluated using evaluation indicators. The performance evaluation results of the first occupational noise-induced hearing loss risk screening model were compared with the performance evaluation results of the second occupational noise-induced hearing loss risk screening model, and the model with better performance evaluation results was used as the final occupational noise-induced hearing loss risk screening model.

7. The method for constructing a risk screening model for occupational noise-induced hearing loss according to claim 6, characterized in that: The importance screening of the demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population is performed to obtain characteristic data related to the risk screening of occupational noise-induced hearing loss of the target population, including: Obtaining manually selected feature data based on demographic characteristics data, biochemical metabolic index data, and hematological index data of the target population, and conducting importance analysis related to occupational noise-induced hearing loss risk screening; Based on the demographic characteristics, biochemical metabolic index data, and hematological index data of the target population, principal component analysis was used to conduct importance analysis related to the risk screening of occupational noise-induced hearing loss, and PCA selected feature data were obtained; Based on the demographic, biochemical, and hematological data of the target population, a maximum relevance and minimum redundancy method was used to conduct an importance analysis related to occupational noise-induced hearing loss risk screening to obtain mRMR selection feature data. The manually selected feature data, PCA selected feature data, and mRMR selected feature data are used as the target population's occupational noise-induced hearing loss risk screening-related feature data, respectively, to reconstruct the first occupational noise-induced hearing loss risk screening model; Furthermore, the method utilizes a machine learning algorithm based on the demographic data, biochemical metabolic index data, and hematological index data of the target population to enable the model to learn the relationship between the demographic data, biochemical metabolic index data, and hematological index data of the population with occupational noise-induced hearing loss and the population with normal hearing who are exposed to noise, thereby obtaining an occupational noise-induced hearing loss risk screening model, including: The demographic characteristic data, biochemical metabolic index data, and hematological index data of the target population were replaced by the manually selected feature data, PCA selected feature data, and mRMR selected feature data of the target population, respectively, and the first occupational noise-induced hearing loss risk screening model was retrained to obtain the manually selected feature-occupational noise-induced hearing loss risk screening model, the PCA selected feature-occupational noise-induced hearing loss risk screening model, and the mRMR selected feature-occupational noise-induced hearing loss risk screening model; Decision curve analysis was used to evaluate the net benefits of the manually selected features-occupational noise-induced hearing loss risk screening model, PCA selected features-occupational noise-induced hearing loss risk screening model, and mRMR selected features-occupational noise-induced hearing loss risk screening model at different thresholds.

8. An occupational noise-induced hearing loss risk auxiliary assessment system, characterized in that: include: A data receiving module, configured to receive screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in the method for constructing an occupational noise-induced hearing loss risk screening model according to any one of claims 1 to 7, or any combination thereof; An occupational noise-induced hearing loss risk screening module, configured to obtain an occupational noise-induced hearing loss risk screening result for a subject based on screening factor data of the subject and an occupational noise-induced hearing loss risk screening model obtained by the method for constructing an occupational noise-induced hearing loss risk screening model according to any one of claims 1 to 7; The data output module is used to output the risk screening result of occupational noise-induced hearing loss of the subject to be tested to at least one terminal.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the following steps are implemented: Receiving screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in the method for constructing an occupational noise-induced hearing loss risk screening model according to any one of claims 1 to 7, or any combination thereof; Obtaining an occupational noise-induced hearing loss risk screening result for the subject based on the screening factor data of the subject and the occupational noise-induced hearing loss risk screening model obtained by the method for constructing an occupational noise-induced hearing loss risk screening model according to any one of claims 1 to 7; The occupational noise-induced hearing loss risk screening result of the subject is output to at least one terminal.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the following steps are implemented: Receiving screening factor data of a subject from at least one terminal, wherein the screening factor data includes any one of demographic characteristic data, biochemical metabolic index data, and hematological index data used in the method for constructing an occupational noise-induced hearing loss risk screening model according to any one of claims 1 to 7, or any combination thereof; Obtaining an occupational noise-induced hearing loss risk screening result for the subject based on the screening factor data of the subject and the occupational noise-induced hearing loss risk screening model obtained by the method for constructing an occupational noise-induced hearing loss risk screening model according to any one of claims 1 to 7; The occupational noise-induced hearing loss risk screening result of the subject is output to at least one terminal.

Images