Method, equipment and medium for assisting in identifying acute A-type aortic dissection
By combining gas chromatography-ion mobility spectrometry with machine learning algorithms, volatile organic compounds in serum are identified, and a high-performance model is constructed. This solves the problems of high invasiveness and insufficient accuracy in the identification of acute type A aortic dissection in existing technologies, and achieves rapid and accurate identification results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-24
- Publication Date
- 2026-03-24
AI Technical Summary
Existing CT angiography and non-invasive detection methods have problems such as high invasiveness, high cost and insufficient accuracy in the identification of acute type A aortic dissection, making it difficult to promote and apply them on a large scale.
By combining gas chromatography-ion mobility spectrometry with machine learning algorithms, a high-performance classification model is constructed to identify volatile organic compounds in serum, thus assisting in the identification of acute type A aortic dissection.
It enables rapid, accurate, and low-interference identification of acute type A aortic dissection, improving identification efficiency and accuracy while reducing detection time and cost.
Smart Images
Figure CN121725937A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a method, device, and medium for assisting in the identification of acute type A aortic dissection. Background Technology
[0002] Acute type A aortic dissection (ATAAD) is one of the most dangerous cardiovascular diseases of the aorta, belonging to the Stanford classification A. It is characterized by rapid onset, rapid progression, and extremely high mortality; the mortality rate in untreated patients can reach 50% within 48 hours of onset. Due to the high pre-hospital mortality rate and potential for misdiagnosis, rapid and accurate identification of the disease is extremely important.
[0003] Currently, methods for detecting acute type A aortic dissection include electrocardiogram, chest X-ray, echocardiography, CT, magnetic resonance imaging, and laboratory tests. Among these, CT angiography is the gold standard. In addition, laboratory methods such as D-dimer testing are non-invasive, which are rapid, simple, and provide objective results that are not affected by equipment or personnel experience.
[0004] CT angiography is the gold standard for identifying acute type A aortic dissection, but the method is invasive, and the contrast agent used during the examination may cause allergic reactions or acute kidney injury in patients. In addition, it has disadvantages such as complicated operation, high cost, and poor compliance, making it difficult to promote its large-scale application.
[0005] Among non-invasive detection methods, the D-dimer method based on immunology has important negative exclusion value in acute type A aortic dissection, but its identification accuracy is insufficient. Currently, better methods and models are still needed for the identification of acute type A aortic dissection. Summary of the Invention
[0006] This application provides a method, device, and medium for assisting in the identification of acute type A aortic dissection, in order to solve the above-mentioned problems.
[0007] This application employs gas chromatography-ion mobility spectrometry (GC-IMA) to identify volatile organic compounds (VOCs) based on retention index and relative migration time, accurately identifying trace amounts of VOC markers in serum with high sensitivity. Simultaneously, machine learning algorithms are used to analyze and model VOC data. These algorithms automatically learn features and patterns in the data, improving the efficiency and accuracy of substance screening. Furthermore, the importance of VOCs in the random forest algorithm model is ranked by calculating Gini Importance, systematically screening the model's accuracy, and then optimizing the model to construct a high-performance model. This innovative combination of highly sensitive technology and the random forest algorithm offers a better way to assist in the identification of acute type A aortic dissection.
[0008] On the one hand, this application provides a method for assisting in the identification of acute type A aortic dissection, the method comprising the following steps: Step S1: acquiring detection data of volatile organic compounds (VOCs) in serum samples detected by gas chromatography-ion mobility spectrometry (GC-IMS); Step S2: determining the types of VOCs and performing statistical analysis to screen out characteristic VOCs that show statistical differences between the acute type A aortic dissection group and the control group; Step S3: based on the screened characteristic VOCs, constructing a classification model for identifying acute type A aortic dissection using a machine learning algorithm; Step S4: screening the optimal classification model and optimizing the optimal classification model to obtain an optimized identification model; Step S5: processing the serum sample VOCs data of the test subject using the optimized identification model to output the auxiliary identification result of acute type A aortic dissection.
[0009] In one implementation of this application, step S3, determining the type of VOCs, is obtained by judging the VOCs monitoring data based on the retention index RI and relative migration time Dt. Specifically, this includes: comparing the gas chromatography retention time and ion migration spectrum drift time data obtained from the detection with a standard substance database containing retention index and relative migration time data of C4-C9 ketone standard substances; when the retention index and relative migration time of the substance to be identified are consistent with the data of a certain standard substance in the database, the identification of the VOCs is completed.
[0010] In one implementation of this application, step S2 involves statistically analyzing the identified VOCs to screen for characteristic VOCs. Specifically, this includes: using the Mann-Whitney U test to analyze the difference in VOC expression levels between the acute type A aortic dissection group and the control group; screening out VOCs showing statistically significant differences between the two groups; and selecting a total of eight characteristic VOCs with significant differences for subsequent modeling. The control group mainly consists of myocardial infarction patients and healthy individuals.
[0011] In one implementation of this application, the machine learning algorithm in step S3 is Random Forest (RF), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), Minimum Absolute Shrinkage and Selection Operator (LASSO), and Limiting Gradient Boosting (XGBoost) algorithm.
[0012] In one implementation of this application, constructing a classification model for identifying acute type A aortic dissection includes: dividing a dataset containing characteristic VOCs data into a training set and a validation set proportionally; training a selected machine learning algorithm using the training set to obtain a preliminary classification model; and evaluating the performance of the preliminary classification model using the validation set. The performance evaluation metrics include accuracy, precision, sensitivity, specificity, false positive rate, false negative rate, and the area under the receiver operating characteristic (ROC) curve (AUC).
[0013] In one implementation of this application, the selection of the optimal classification model includes: selecting the best-performing classification model as the optimization direction based on the relevant performance indicators of each classification model. The specific process is as follows: calculating the accuracy, precision, sensitivity, specificity, false positive rate, and false negative rate of the six machine learning algorithms on the validation set, assigning values based on the calculation results, and finally selecting the model constructed by the random forest algorithm as the best-performing model, and performing the next optimization operation on the random forest algorithm model.
[0014] In one implementation of this application, the optimal classification model is further optimized by: using the Gini impurity calculation method to rank the importance of the characteristic VOCs used for modeling; reducing the number of VOCs used for modeling in ascending order of importance scores to construct multiple sub-models containing different numbers of VOCs; selecting the sub-model with the best performance as the optimization direction based on the relevant performance indicators of each sub-model; further optimizing the selected hyperparameters using a grid search method, and determining the optimal hyperparameter combination through cross-validation, finally obtaining the optimized recognition model.
[0015] Additionally, an auxiliary device for identifying acute type A aortic dissection is provided, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the aforementioned method for auxiliary identification of acute type A aortic dissection.
[0016] Finally, a non-volatile computer storage medium is also provided for assisting in the identification of acute type A aortic dissection, storing computer-executable instructions that are executed by a processor to implement the aforementioned method for assisting in the identification of acute type A aortic dissection.
[0017] The method, device, and medium provided in this application for assisting in the identification of acute type A aortic dissection have the following beneficial effects: (1) Efficient processing of high-dimensional complex data. GC-IMS detection generates high-dimensional data containing a large amount of information, covering multiple dimensions of features such as retention time, drift time, and signal strength of different VOCs. Machine learning algorithms can automatically reduce the dimensionality of high-dimensional data, filter out key features, eliminate redundant information, reduce the difficulty of data processing, and avoid omissions and errors caused by manual screening.
[0018] (2) Improved identification and analysis efficiency. GC-IMS itself has the advantages of fast analysis speed (usually completing detection within a few minutes) and no need for complex preprocessing. When combined with machine learning, it can further improve the overall analysis efficiency. After the machine learning model is trained, it can automatically analyze the data detected by GC-IMS in real time and directly output the classification or prediction results without manual intervention, which greatly shortens the time from detection to conclusion.
[0019] (3) Enhance anti-interference ability and improve stability. Machine learning can use feature selection algorithms to select features with strong anti-interference ability, and reduce the impact of interference signals on the results through model optimization. Attached Figure Description
[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a method for assisting in the identification of acute type A aortic dissection provided in this application embodiment; Figure 2 A three-dimensional spectrum of VOCs signal peaks provided in the embodiments of this application; Figure 3(a) shows the confusion matrix of the 8-VOCs-based model constructed by the six machine learning methods provided in the embodiments of this application; Figure 3(b) shows the ROC curves corresponding to the 8-VOCs-based model constructed by the six machine learning methods provided in the embodiments of this application. Figure 4 The model accuracy statistics provided for the embodiments of this application; Figure 5 Sensitivity-specificity curves provided for embodiments of this application; Figure 6This is a schematic diagram of a device for assisting in the identification of acute type A aortic dissection, provided in an embodiment of this application. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0022] This application provides a method, device, and medium for assisting in the identification of acute type A aortic dissection. The technical solution proposed in this application will be described in detail below with reference to the accompanying drawings.
[0023] Figure 1 This is a flowchart illustrating a method for assisting in the identification of acute type A aortic dissection, as provided in an embodiment of this application. Figure 1 As shown, the method mainly includes the following steps: Step S1: Obtain detection data of volatile organic compounds (VOCs) in serum samples based on gas chromatography-ion mobility spectrometry (GC-IMS) technology; Step S2: Determine the types of VOCs and perform statistical analysis to screen out the characteristic VOCs that show statistical differences between the acute type A aortic dissection group and the control group; Step S3: Based on the selected characteristic VOCs, a classification model for identifying acute type A aortic dissection is constructed using a machine learning algorithm; Step S4: Select the optimal classification model, optimize the optimal classification model, and obtain the optimized recognition model; Step S5: Use the optimized recognition model to process the VOCs data of the serum sample of the test subject and output the auxiliary recognition results of acute type A aortic dissection.
[0024] In this application, gas chromatography-ion mobility spectrometry (GC-IMS) is used to detect volatile organic compounds in serum. The serum sample is heated in an incubator and then placed in a sample tray. It is extracted by an injector, first separated by gas chromatography, and then separated by ion mobility spectrometry. Due to the differences in mass, charge, collision interface and spatial configuration of substances, their migration rates in the electric field are different, and the time to reach the detector is also different. The detector collects ion signals to form gas phase ion mobility spectra, and qualitative and quantitative analysis of substances is performed by using the spectra.
[0025] The detected volatile organic compounds (VOCs) were identified based on the retention index (RI) and ion migration time (drift time, Dt) of gas chromatography (GC). The peak positions of the VOCs were confirmed by comparing them with the peak positions of standard substances (C4-C9 ketones), and the RI and Dt of the test substance must be consistent with the data of the standard substance. The three-dimensional signal on the spectrum has a regular peak shape, and the peak intensity is at least three times the baseline noise. The bias of the absolute ion migration time was eliminated by normalizing the RIP. Figure 2 A three-dimensional spectrum consisting of the gas-phase retention time, ion migration time, and signal peak intensity of each sample VOCs is presented.
[0026] After identifying VOCs based on retention index (RI) and relative migration time (drift time, Dt), Mann-Whitney U statistical analysis was used to identify VOCs with statistical significance between the two groups. A p-value < 0.05 was considered statistically significant. A total of 8 VOCs with differences were selected and further incorporated into the model, and the model was constructed.
[0027] Six machine learning methods were employed to construct the model, including Random Forest (RF), Support Vector Machine (SVM), Latent Dirichlet Allocation (LDA), K-Nearest Neighbors (KNN), Least Absolute Shrinkage and Selection Operator (LASSO), and Extreme Gradient Boosting (XGBoost). For each method, 70% of the population was randomly divided into a training set and 30% into a validation set. Receiver Operating Characteristic (ROC) curves were plotted, and relevant performance metrics were calculated, including accuracy {[True Positive (TP) + True Negative (TN)] / TP + TN + False Positive (FP) + False Negative (FN)}, precision (TP / TP + FP), sensitivity (TP / TB + FN), specificity (TN / TN + FP), false positive rate (FPR = FP / FP + TN), and false negative rate (FNR = FN / TP + FN). The comprehensive analysis of the above algorithms revealed that the model built using the Random Forest algorithm exhibited the highest performance, with an AUC of 0.963, a sensitivity of 96.3%, a specificity of 90.5%, a precision of 92.9%, a false positive rate (FPR) of 9.5%, and a false negative rate (FNR) of 3.7%. This model demonstrates good recognition performance. Figure 3 shows the confusion matrix and ROC curves of the 8-VOCs-based model constructed using six machine learning methods.
[0028] Based on the model constructed using the random forest algorithm, the Gini coefficient is further calculated using feature_importances_. The Gini Importance metric ranked the importance of volatile organic compounds (VOCs) in the model. Based on the importance score, the number of VOCs was reduced from low to high, resulting in eight models covering 1-8 VOCs. The model with the highest accuracy (93.75%) was selected as the 8-VOCs combination. Furthermore, in building the new random forest, grid search was used from sklearn calling GridSearchCV. Ten-fold cross-validation was performed for each parameter combination, and the optimal hyperparameter combination was selected for final model optimization. The final model, optimized based on 8-VOCs, achieved an AUC of 0.966, a sensitivity of 90.5%, a specificity of 96.3%, a precision of 95.0%, and an FPR of 3.7% (p<0.001). This optimized model improved the specificity and precision of identification while reducing the FPR. Figure 4 For model accuracy statistics, Figure 5 The graph shows the sensitivity-specificity curve. The optimal combination of VOCs and the machine learning model were selected for the auxiliary identification of acute type A aortic dissection.
[0029] The above describes a method for assisting in the identification of acute type A aortic dissection provided by embodiments of this application. Based on the same inventive concept, embodiments of this application also provide a device for assisting in the identification of acute type A aortic dissection. Figure 6 A schematic diagram of a device for assisting in the identification of acute type A aortic dissection provided in an embodiment of this application is shown below. Figure 6 As shown, the device mainly includes: at least one processor 601; and a memory 602 communicatively connected to the at least one processor; wherein the memory 602 stores instructions that can be executed by the at least one processor 601, and the instructions are executed by the at least one processor 601 to enable the at least one processor 601 to complete the aforementioned method for assisting in the identification of acute type A aortic dissection.
[0030] In addition, this application embodiment also provides a non-volatile computer storage medium for assisting in the identification of acute type A aortic dissection, which stores computer-executable instructions that are executed by a processor to implement the aforementioned method for assisting in the identification of acute type A aortic dissection.
[0031] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0032] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0033] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0034] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0035] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0036] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0037] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for assisting in the identification of acute type A aortic dissection, characterized in that, The method includes the following steps: Step S1: Obtain detection data of volatile organic compounds (VOCs) in serum samples based on gas chromatography-ion mobility spectrometry (GC-IMS) technology; Step S2: Determine the types of VOCs and perform statistical analysis to screen out the characteristic VOCs that show statistical differences between the acute type A aortic dissection group and the control group; Step S3: Based on the selected characteristic VOCs, a classification model for identifying acute type A aortic dissection is constructed using a machine learning algorithm; Step S4: Select the optimal classification model, optimize the optimal classification model, and obtain the optimized recognition model; Step S5: Use the optimized recognition model to process the VOCs data of the serum sample of the test subject and output the auxiliary recognition results of acute type A aortic dissection.
2. The method for assisting in the identification of acute type A aortic dissection according to claim 1, characterized in that, In step S3, the types of VOCs are determined based on the retention index RI and relative migration time Dt, according to the VOCs monitoring data. Specifically, this includes: The gas chromatography retention time and ion migration spectrum drift time data obtained from the detection were compared with a standard reference database containing retention index and relative migration time data of C4-C9 n-ketone standard references. The identification of a VOC is complete when the retention index and relative migration time of the substance to be identified are consistent with the data of a certain standard substance in the database.
3. The method for assisting in the identification of acute type A aortic dissection according to claim 1, characterized in that, In step S2, statistical analysis is performed on the identified VOCs to screen out characteristic VOCs, specifically including: The Mann-Whitney U test was used to analyze the differences in VOC expression levels between the acute type A aortic dissection group and the control group; VOCs with statistically significant differences between the two groups were screened out; among them, a total of 8 characteristic VOCs with differences were selected for subsequent modeling.
4. The method for assisting in the identification of acute type A aortic dissection according to claim 1, characterized in that, The machine learning algorithms in step S3 are Random Forest (RF), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), Minimum Absolute Shrinkage and Selection Operator (LASSO), and Limiting Gradient Boosting (XGBoost) algorithm.
5. The method for assisting in the identification of acute type A aortic dissection according to claim 1, characterized in that, Construct a classification model for identifying acute type A aortic dissection, including: The dataset containing characteristic VOCs data is divided into training and validation sets proportionally. The selected machine learning algorithm is trained using the training set to obtain a preliminary classification model; The performance of the preliminary classification model is evaluated using the validation set. The performance evaluation metrics include accuracy, precision, sensitivity, specificity, false positive rate, false negative rate, and area under the receiver operating characteristic (ROC) curve (AUC).
6. The method for assisting in the identification of acute type A aortic dissection according to claim 1, characterized in that, The selection of the optimal classification model includes: selecting the best-performing classification model as the optimization direction based on the relevant performance indicators of each classification model. The specific process is as follows: calculate the accuracy, precision, sensitivity, specificity, false positive rate, and false negative rate of the six machine learning algorithms on the validation set, assign values based on the calculation results, and finally select the model constructed by the random forest algorithm as the best-performing model. Then, perform the next optimization operation on the random forest algorithm model.
7. The method for assisting in the identification of acute type A aortic dissection according to claim 1, characterized in that, Further optimization of the optimal classification model includes: The importance of the characteristic VOCs used for modeling is ranked using the Gini impurity calculation method. The number of VOCs used for modeling is reduced sequentially according to the importance score from low to high, and multiple sub-models containing different numbers of VOCs are constructed. Based on the relevant performance indicators of each sub-model, the sub-model with the best performance is selected as the optimization direction; The selected hyperparameters are further optimized using a grid search method, and the optimal combination of hyperparameters is determined through cross-validation, ultimately obtaining the optimized recognition model.
8. A device for assisting in the identification of acute type A aortic dissection, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform a method for assisting in the identification of acute type A aortic dissection as described in any one of claims 1-7.
9. A non-volatile computer storage medium for assisting in the identification of acute type A aortic dissection, storing computer-executable instructions, characterized in that, The computer-executable instructions are executed by a processor to implement a method for assisting in the identification of acute type A aortic dissection as described in any one of claims 1-7.