Intelligent method and system for rapid screening and triage of aortic dissection

Abstract

The invention relates to an intelligent method and system for rapid screening and triage of aortic dissection. The method comprises the following steps: collecting text information data sets indicating chief complaint symptoms, first symptoms, past medical history and high-risk signs; based on the text information data set, risk factor identification is carried out by using a feature extraction and classification algorithm combining natural language processing with machine learning; through a correlation analysis technology, correlation between the risk factor and the chief complaint symptom and between the risk factor and the first symptom is calculated, and whether the patient belongs to an AD high-risk group or not is preliminarily evaluated by setting a threshold value according to correlation strength; starting a first-aid green channel for the target patient which is preliminarily assessed as AD high risk, measuring blood pressure of four limbs, and recording generated blood pressure data; comparing the blood pressure data of the target patient with a preset threshold value, and determining the possibility that the target patient has suspected AD through logical judgment; and performing priority triage based on the suspected AD possibility evaluation result, the severity of the clinical symptoms and the availability of the medical resources.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to an intelligent method and system for rapid screening and triage of aortic dissection. Background Art

[0002] Aortic dissection is a serious cardiovascular emergency, and its timely and accurate screening and triage are crucial to the patient's life safety. However, although traditional medical imaging technology is accurate, it is complicated and costly to operate. For patients with acute illness, time is tight and a faster and more convenient screening method is needed.

[0003] At present, clinical screening for aortic dissection mainly relies on the patient's clinical manifestations and signs, as well as high-risk medical history. However, these symptoms and signs are not specific and are easily interfered by other diseases, leading to misdiagnosis and missed diagnosis. In order to solve this problem, researchers have tried to apply artificial intelligence algorithms to the screening of aortic dissection. By collecting patient data such as the patient's chief complaint, first symptoms, past medical history, and high-risk signs, these data are used for deep learning and analysis to discover the relationship between risk factors and symptoms, thereby preliminarily assessing whether the patient belongs to the high-risk group for AD.

[0004] However, although this method can theoretically improve the accuracy and efficiency of screening, it still faces some challenges in practical application. First, high-quality labeled data is often difficult to obtain in the medical field, which limits the training and optimization of the model. Second, even with sufficient labeled data, the generalization ability of the model may be limited, especially when faced with new, unlabeled data. Therefore, in order to overcome the limitations of existing screening methods and meet the needs of clinical practice for rapid and accurate screening of aortic dissection, it is urgent to provide an intelligent method and system for rapid screening and triage of aortic dissection, so as to further improve the accuracy of aortic dissection screening. Summary of the invention

[0005] In order to improve the accuracy of aortic dissection screening, the purpose of the present invention is to provide an intelligent method and system for rapid screening and triage of aortic dissection. The technical solutions adopted are as follows:

[0006] In a first aspect, the present application discloses an intelligent method for rapid screening and triage of aortic dissection, the method comprising:

[0007] S1. Collect text information datasets that indicate the main symptoms, first symptoms, past medical history, and high-risk signs;

[0008] S2. Based on the text information data set, risk factors are identified using natural language processing combined with machine learning feature extraction and classification algorithms;

[0009] S3. Calculate the association between risk factors and the main symptoms and the first symptoms through correlation analysis technology, and set a threshold based on the strength of the association to preliminarily assess whether the patient belongs to the high-risk group for AD;

[0010] S4. For target patients initially assessed as being at high risk of AD, blood pressure in the limbs should be measured while the emergency green channel is activated, and the resulting blood pressure data should be recorded;

[0011] S5, comparing the blood pressure data of the target patient with a preset threshold, and determining the possibility that the target patient has suspected AD through logical judgment;

[0012] S6. Priority triage should be performed based on the target patient’s suspected AD possibility assessment results, the severity of clinical symptoms, and the availability of medical resources.

[0013] In one embodiment, in step S2, the identification of risk factors based on the text information dataset using natural language processing combined with machine learning feature extraction and classification algorithms includes:

[0014] S21, performing text preprocessing based on the text information data set to obtain a preprocessed text data set;

[0015] S22. Based on natural language processing technology, extract key features related to the risk of aortic dissection from the preprocessed text dataset through syntactic analysis combined with semantic annotation;

[0016] S23. Inputting the key features into an unsupervised machine learning model based on a clustering algorithm, and training the model to identify risk factors associated with aortic dissection.

[0017] In one embodiment, in step S22, the key features related to the risk of aortic dissection are extracted from the preprocessed text dataset by syntactic analysis combined with semantic annotation based on natural language processing technology, including:

[0018] S221, parsing the preprocessed text data set based on a syntactic analyzer to obtain a syntactic tree;

[0019] S222, semantically annotating the words in the text according to the pre-trained semantic model to obtain a semantic annotation result;

[0020] S223. Construct a knowledge graph based on the syntactic tree and the semantic annotation results, and perform relationship path analysis in the knowledge graph to extract key features related to the risk of aortic dissection.

[0021] In one embodiment, in step S23, the key features are input into an unsupervised machine learning model based on a clustering algorithm, and the model is trained to identify risk factors associated with aortic dissection, including:

[0022] S231, inputting the key features into an unsupervised machine learning model based on a clustering algorithm, and clustering the key features using a K-means clustering algorithm to group similar features into one category;

[0023] S232, for each cluster obtained by the final classification, determine the potential risk factor based on the similarity and correlation between the characteristics within the cluster;

[0024] S233. Perform medical evidence-based verification processing based on various potential risk factors to obtain risk factors associated with aortic dissection.

[0025] Furthermore, in step S3, the correlation analysis technique is used to calculate the correlation between the risk factor and the main complaint symptom and the first symptom, including:

[0026] S31. Obtain statistical data related to risk factors, presenting symptoms, and first symptoms;

[0027] S32, quantifying the statistical data to obtain standardized values ​​that can be used for statistical analysis;

[0028] S33, based on the standardized values, respectively calculating the correlation coefficients between the risk factors and the main symptoms, and the first symptoms;

[0029] S34. Based on the two correlation coefficients obtained, significance tests and effect size assessments were performed to obtain the degree of association between risk factors and the main symptoms and the first symptoms.

[0030] Furthermore, in step S5, the blood pressure data of the target patient is compared with a preset threshold value, and a logical judgment is performed to determine the possibility that the target patient has suspected AD, including:

[0031] S51, obtaining blood pressure data of target patients, and setting a blood pressure threshold range associated with AD risk based on data mining and analysis of a large-scale clinical database;

[0032] S52: When it is determined that the blood pressure data falls within the blood pressure threshold range, the possibility of suspected AD in the target patient is calculated based on the risk assessment model.

[0033] Furthermore, in step S6, the priority triage is performed based on the target patient's suspected AD possibility assessment result, the severity of clinical symptoms, and the availability of medical resources, including:

[0034] S61, classifying the target patient into different risk levels according to the suspected AD possibility assessment result;

[0035] S62. According to the risk level, the severity of clinical symptoms and the availability of medical resources are comprehensively considered, and priority triage is carried out based on the comprehensive assessment results.

[0036] In the second aspect, the present application discloses an intelligent system for rapid screening and triage of aortic dissection, the system comprising a patient information collection module, a feature extraction and classification module, an association analysis module, a blood pressure measurement and recording module, a suspected AD judgment module, and a priority triage module, wherein:

[0037] The patient information collection module is used to collect a text information data set indicating the main symptoms, the first symptoms, the past medical history, and the high-risk signs;

[0038] The feature extraction and classification module is used to identify risk factors based on the text information data set by using a feature extraction and classification algorithm combining natural language processing with machine learning;

[0039] The association analysis module is used to calculate the association between the risk factor and the main symptoms and the first symptoms through correlation analysis technology, and to preliminarily assess whether the patient belongs to the high-risk group of AD by setting a threshold according to the strength of the association;

[0040] The blood pressure measurement and recording module is used to measure the blood pressure of the four limbs of the target patient who is initially assessed to be at high risk of AD, while activating the emergency green channel, and record the generated blood pressure data;

[0041] The suspected AD judgment module is used to compare the blood pressure data of the target patient with a preset threshold value, and to determine the possibility of suspected AD in the target patient through logical judgment;

[0042] The priority triage module is used to perform priority triage based on the target patient's suspected AD possibility assessment result, the severity of clinical symptoms, and the availability of medical resources.

[0043] In a third aspect, the present application discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the intelligent method for rapid screening and triage of aortic dissection.

[0044] In a fourth aspect, the present application discloses an intelligent computing control device for rapid screening and triage of aortic dissection, comprising a communication interface, a memory, a communication bus and a processor, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0045] The memory is used to store computer programs;

[0046] The processor is used to implement the steps of the intelligent method for rapid screening and triage of aortic dissection when executing the program stored in the memory.

[0047] The present invention has the following beneficial effects:

[0048] 1) Using natural language processing and machine learning algorithms for feature extraction and classification can automatically and efficiently identify risk factors associated with aortic dissection (AD), reducing the errors and subjectivity of manual judgment;

[0049] 2) By quantifying the strength of the association between risk factors and symptoms through correlation analysis technology and setting thresholds for preliminary evaluation, it is possible to quickly screen out people at high risk of AD, thus improving the efficiency and accuracy of diagnosis;

[0050] 3) Priority triage is carried out based on the suspected AD possibility assessment results, the severity of clinical symptoms and the availability of medical resources, ensuring that medical resources can be effectively utilized and improving the overall efficiency and quality of medical services. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 A method flow chart of an intelligent method for rapid screening and triage of aortic dissection provided by one embodiment of the present invention;

[0053] Figure 2 A system structure diagram of an intelligent system for rapid screening and triage of aortic dissection provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation, structure, features and effects of an intelligent method and system for rapid screening and triage of aortic dissection proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0055] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0056] The specific scheme of an intelligent method and system for rapid screening and triage of aortic dissection provided by the present invention is described in detail below in conjunction with the accompanying drawings.

[0057] See also Figure 1 , which shows a method flow chart of an intelligent method for rapid screening and triage of aortic dissection provided by an embodiment of the present invention, the method comprising:

[0058] Step S1, collecting a text information dataset indicating the main symptoms, the first symptoms, the past medical history, and the high-risk signs.

[0059] Step S2, based on the text information data set, risk factors are identified using natural language processing combined with machine learning feature extraction and classification algorithms.

[0060] Step S3, using correlation analysis technology, calculate the correlation between the risk factors and the main symptoms, as well as the first symptoms, and set a threshold based on the strength of the correlation to preliminarily assess whether the patient belongs to the high-risk group for AD.

[0061] Step S4, for the target patient who is initially assessed to be at high risk of AD, while the emergency green channel is activated, the blood pressure of the limbs is measured and the generated blood pressure data is recorded.

[0062] Step S5, comparing the blood pressure data of the target patient with a preset threshold, and determining the possibility of suspected AD in the target patient through logical judgment.

[0063] Step S6, priority triage is performed based on the target patient's suspected AD possibility assessment results, the severity of clinical symptoms, and the availability of medical resources.

[0064] As can be seen from the above, the intelligent method for rapid screening and triage of aortic dissection disclosed in the present application utilizes natural language processing and machine learning algorithms for feature extraction and classification, and can automatically and efficiently identify risk factors associated with aortic dissection (AD), thereby reducing the errors and subjectivity of manual judgment; by quantifying the strength of the association between risk factors and symptoms through correlation analysis technology, and setting thresholds for preliminary evaluation, it can quickly screen out people at high risk of AD, thereby improving the efficiency and accuracy of diagnosis; based on the suspected AD possibility assessment results, the severity of clinical symptoms, and the availability of medical resources, priority triage is performed, ensuring that medical resources can be effectively utilized and improving the overall efficiency and quality of medical services.

[0065] In one of the embodiments, in step S2, the risk factors are identified by using a feature extraction and classification algorithm that combines natural language processing and machine learning based on the text information dataset, including:

[0066] Step S21, perform text preprocessing on the text information dataset to obtain a preprocessed text dataset.

[0067] Specifically, the text preprocessing includes at least one of removing irrelevant characters and stop words, word segmentation, part-of-speech tagging, and removing redundant information. Among them, removing irrelevant characters and stop words includes removing characters (such as punctuation marks) and stop words (such as common but meaningless words like "de" and "le") that are not helpful for subsequent analysis. During the word segmentation process, continuous text is segmented into individual lexical units, and part-of-speech tagging classifies the part of speech of each word, such as nouns, adjectives, verbs, etc. In addition, since there may be redundant information in the text data, it is necessary to remove this redundant information to improve the data quality.

[0068] Step S22, based on natural language processing technology, extract key features related to aortic dissection risk from the preprocessed text dataset through syntactic analysis combined with semantic annotation.

[0069] Specifically, the present application first parses the preprocessed text dataset through a syntactic analyzer to obtain a syntactic tree reflecting the hierarchical and dependency relationships between the components in the sentence; then, uses a pre-trained semantic model to perform semantic annotation on each word in the text, and constructs a knowledge graph in combination with the syntactic tree and the semantic annotation results, and extracts key features related to aortic dissection risk through relationship path analysis in the graph.

[0070] Step S23, input the key features into an unsupervised machine learning model based on a clustering algorithm, and train the model to identify risk factors related to aortic dissection.

[0071] Specifically, the present application inputs these key features into an unsupervised machine learning model, uses the K-means clustering algorithm to classify similar features, and determines potential risk factors by analyzing the statistical relevance between the features within the clustering clusters. Then, based on medical evidence, these potential risk factors are verified, and by analyzing their occurrence frequency, statistical significance, association strength, and consistency in relevant medical literature, clinical trial data, and epidemiological research results, and comprehensively evaluating these indicators, risk factors related to aortic dissection are identified.

[0072] In the above embodiment, on the one hand, by using natural language processing technology, through syntactic analysis and semantic annotation, the text content can be deeply understood and the key features related to the risk of aortic dissection can be accurately extracted. On the other hand, the extracted key features are input into an unsupervised machine learning model based on a clustering algorithm. By training the model, the risk factors related to aortic dissection can be automatically identified and classified. This method not only improves the speed and efficiency of risk factor identification, but also reduces the subjectivity and errors of human judgment, making the results more objective and reliable.

[0073] In one embodiment, in step S22, the key features related to the risk of aortic dissection are extracted from the preprocessed text dataset by syntactic analysis combined with semantic annotation based on natural language processing technology, including:

[0074] Step S221, parsing the preprocessed text data set based on a syntactic analyzer to obtain a syntactic tree.

[0075] Specifically, a syntactic analyzer is a natural language processing tool that can parse the syntactic structure of text data to identify the various components in a sentence and the syntactic relationships between them. The syntactic tree shows the hierarchy and dependency relationships between the various components in a sentence in a tree-like structure. This structured representation helps to understand the grammatical structure of a sentence more clearly, and thus provides strong syntactic support for subsequent natural language processing tasks.

[0076] Step S222: semantically annotate the words in the text according to the pre-trained semantic model to obtain a semantic annotation result.

[0077] Specifically, the pre-trained semantic model can understand and parse the meaning of words in context. It is trained based on a large amount of corpus and annotated data, and can assign corresponding semantic tags to each word in the text. These semantic tags reflect the specific role of the words in a sentence or paragraph, such as the agent of a verb, the modified object of an adjective, etc. In this application, the semantic tags are specifically stored in the form of key-value pairs, where the key represents the word and the value represents the assigned set of semantic tags.

[0078] In one of the embodiments, the pre-trained semantic model adopts a Transformer architecture to capture the contextual information of the vocabulary through bidirectional encoding, so as to more accurately understand the semantics of the vocabulary.

[0079] Step S223, constructing a knowledge graph based on the syntactic tree and the semantic annotation results, and performing relationship path analysis in the knowledge graph to extract key features related to the risk of aortic dissection.

[0080] Specifically, this application will map the syntactic relations in the syntax tree to entity-relationship-entity triples in the knowledge graph, and build the skeleton of the graph based on this. Subsequently, combined with the semantic annotation results, the semantic attributes of each entity are further enriched, and these attributes are attached to the corresponding entity nodes in the form of key-value pairs. In this way, the knowledge graph not only contains the direct connection between sentence components revealed by the syntactic structure, but also incorporates the deep semantic information of the vocabulary in the context.

[0081] Furthermore, in the relationship path analysis, this application uses the entities and relationships in the knowledge graph to construct a path starting from the risk of aortic dissection, which is formed by connecting a series of related entities and relationships. These paths involve the causes, symptoms, diagnostic methods, treatment methods and other aspects of aortic dissection, forming a comprehensive and in-depth understanding framework of the risk of aortic dissection. By analyzing these paths, key features that are directly related to the risk of aortic dissection and frequently appear in the graph can be identified, such as specific symptom descriptions, indicators of risk factors, etc.

[0082] In the above embodiment, on the one hand, the preprocessed text data set is parsed by a syntactic analyzer to obtain a syntactic tree, which helps to deeply understand the syntactic structure of the text. On the other hand, the semantic annotation of the vocabulary in the text using a pre-trained semantic model can further enrich the semantic information in the text, making the understanding of the text content more accurate and comprehensive, and laying a solid foundation for the subsequent knowledge graph construction and feature extraction. Finally, a knowledge graph is constructed based on the syntactic tree and semantic annotation results, and by performing relationship path analysis in the knowledge graph, the association and dependency between various entities and concepts in the text can be clearly revealed, so as to improve the accuracy of key feature extraction.

[0083] In one embodiment, in step S23, the key features are input into an unsupervised machine learning model based on a clustering algorithm, and the model is trained to identify risk factors associated with aortic dissection, including:

[0084] Step S231, input the key features into an unsupervised machine learning model based on a clustering algorithm, and cluster the key features using a K-means clustering algorithm to group similar features into one category.

[0085] Specifically, the present application will first initialize K cluster centers, then calculate the distance between each key feature and each cluster center, and assign each key feature to the cluster center that is closest. Then, the position of the cluster center is updated according to the current allocation result. The above process will be iterated until the position of the cluster center no longer changes or the maximum number of iterations is reached, and then the iteration is stopped. At this point, K clusters will be obtained, among which the key features within each cluster have high similarity and correlation, thereby achieving the purpose of classifying similar features into one category.

[0086] Step S232: for each cluster obtained by the final classification, determine the potential risk factor based on the similarity and correlation between the features in the cluster.

[0087] Specifically, this application will first analyze the distribution of key features within each cluster, including the range of feature values, frequency characteristics and other statistical characteristics. Then, based on these statistical characteristics, similarity measurements between features are performed (such as based on cosine similarity) to evaluate the similarity and correlation between the features within the cluster. By comparing the similarities and correlations between different features, those features that frequently appear in the clusters are identified, and these features are often highly correlated with the risk of aortic dissection, and therefore are considered potential risk factors.

[0088] Step S233, performing verification processing based on medical evidence based on various potential risk factors to obtain risk factors related to aortic dissection.

[0089] Specifically, this application will analyze the frequency of occurrence of each potential risk factor in relevant medical literature, clinical trial data, and epidemiological research results (i.e., how many studies have mentioned or paid attention to the risk factor), statistical significance (i.e., whether the correlation between the risk factor and aortic dissection has reached a statistically significant level), strength of association (i.e., how strong the correlation between the risk factor and aortic dissection is), and consistency (i.e., whether the conclusions of different studies on the correlation between the risk factor and aortic dissection are consistent). By comprehensively evaluating these indicators, the correlation between these potential risk factors and aortic dissection can be verified. Subsequently, a correlation threshold is set, and only risk factors with correlation values ​​higher than the threshold will be considered to have a significant correlation with aortic dissection.

[0090] In one embodiment, the comprehensive evaluation of these indicators to verify the association between these potential risk factors and aortic dissection includes:

[0091] First, a comprehensive evaluation system is constructed, which takes into account the frequency, statistical significance, strength of association, and consistency indicators, and sets corresponding weights for each indicator. The weight setting can be determined based on the importance of the indicator to ensure that the evaluation results can comprehensively and accurately reflect the association between potential risk factors and aortic dissection.

[0092] Next, the comprehensive evaluation system is used to quantify the scores of each potential risk factor. In the scoring process, the specific values ​​and weights of each indicator are used to calculate the comprehensive score of each potential risk factor through weighted average. This comprehensive score will serve as an important basis for the subsequent evaluation of the correlation between potential risk factors and aortic dissection.

[0093] Finally, the quantitative scoring results are input into a pre-built expert system, which uses the logical reasoning and rule matching mechanism within the expert system to assist in evaluating the correlation between potential risk factors and aortic dissection. Specifically, the expert system can deeply analyze and interpret the input quantitative scoring results based on the knowledge and experience of domain experts, and verify, adjust and optimize the scoring results according to preset rules and logic, thereby further improving the accuracy and reliability of the evaluation.

[0094] In the above embodiment, on the one hand, for each cluster, the potential risk factors are determined based on the similarity and correlation between its internal features. Currently, through cluster analysis, it is easier to find those risk factors that may be overlooked or difficult to identify in traditional methods, providing an effective data basis for subsequent verification processing. On the other hand, after the potential risk factors are determined, verification processing based on medical evidence is performed, which further enhances the reliability and scientificity of the results.

[0095] In one embodiment, in step S3, calculating the correlation between the risk factor and the main complaint symptom and the first symptom by correlation analysis technology includes:

[0096] Step S31, obtaining statistical data related to risk factors, main symptoms, and first symptoms.

[0097] Specifically, this application will extract the patient's medical records from the health database, and based on preset keywords and coding rules, extract statistical data related to risk factors, main symptoms, and first symptoms from the medical records. These statistical data reflect the association pattern between different risk factors and specific symptoms in the patient population, the frequency and distribution characteristics of symptoms, and the impact of possible risk factors on the severity of symptoms or the course of the disease.

[0098] Step S32, quantifying the statistical data to obtain standardized values ​​that can be used for statistical analysis.

[0099] Specifically, for non-numeric data (such as symptom descriptions), this application will convert them into numerical data, such as using a Likert scale to score the severity of the symptoms. Afterwards, in order to improve the accuracy and reliability of data analysis and eliminate possible dimensional differences between different variables, this application will also standardize the numerical data, such as using the minimum-maximum standardization method (i.e., linearly converting the original data to a new range of values) to ensure that different variables are compared on the same scale.

[0100] Step S33, based on the standardized values, respectively calculating the correlation coefficients between the risk factors and the main symptoms, and the first symptoms.

[0101] Specifically, the present application uses the Pearson correlation coefficient to calculate the correlation coefficient between each risk factor and the main complaint symptom and the first symptom, respectively, to quantitatively evaluate the linear correlation strength and direction between different risk factors and symptoms.

[0102] Step S34, based on the two obtained correlation coefficients, a significance test and an effect size evaluation are performed to obtain the degree of association between the risk factor and the main complaint symptom and the first symptom.

[0103] Specifically, this application uses statistical software to perform a t-test on the two calculated correlation coefficients to determine whether there is a statistically significant difference between the two. Furthermore, the effect size reflects the closeness of the association between the risk factor and the symptoms. The Pearson correlation coefficient is used in this application to measure the effect size. Finally, this application combines the results of the significance test and the measurement results of the effect size, and obtains the degree of association between the risk factor and the chief complaint symptoms, as well as the first symptoms through logical analysis and quantitative evaluation. Specifically, if it is determined that the correlation coefficient of a certain risk factor with the chief complaint symptoms and the first symptoms is significant, and the effect size measurement is large, then it can be considered that there is a strong association between the risk factor and these symptoms; otherwise, it is considered that the degree of association between it and these symptoms is weak or non-existent.

[0104] In the above embodiment, the correlation between the risk factor and the main symptoms and the first symptoms is further determined by the significance test and the effect size evaluation. Among them, the significance test can determine whether the correlation coefficient has a statistically significant difference, and the effect size evaluation reflects the closeness of the correlation. Such operation can ensure the scientificity and reliability of the correlation analysis.

[0105] In one embodiment, in step S5, the blood pressure data of the target patient is compared with a preset threshold value, and a logical judgment is performed to determine the possibility that the target patient has suspected AD, including:

[0106] Step S51, obtaining the blood pressure data of the target patient, and setting a blood pressure threshold range associated with AD risk based on data mining and analysis of a large-scale clinical database.

[0107] Specifically, this application will use data mining technology to screen out features related to AD risk from clinical databases, such as age, gender, genetic factors, living habits, etc. Afterwards, the association between blood pressure data and these features is analyzed using an association rule mining algorithm to determine the association between blood pressure data and AD risk. Finally, based on the aforementioned determined association rules, a blood pressure threshold range associated with AD risk is set by combining statistical analysis with expert consensus. Specifically, this application will use statistical software to quantitatively analyze the association rules to calculate the strength of the correlation between blood pressure data and AD risk under different feature combinations. Afterwards, the opinions of clinical experts and existing research results are combined, and the patient's overall health status and potential risk factors are comprehensively considered to determine a scientific and practical blood pressure threshold range.

[0108] Step S52: when it is determined that the blood pressure data falls within the blood pressure threshold range, the possibility of suspected AD in the target patient is calculated based on the risk assessment model.

[0109] Specifically, the risk assessment model is a composite scoring system that integrates multiple factors (including but not limited to blood pressure data, age, gender, genetic factors, lifestyle habits, etc.). When the blood pressure data is determined to fall within the blood pressure threshold range associated with AD risk, the application will call the risk assessment model, which comprehensively considers the multiple factors covered therein and calculates the probability of the target patient having a suspected AD risk based on the interaction between them.

[0110] In the above embodiment, through data mining and analysis of large-scale clinical databases, it is possible to comprehensively consider a variety of factors related to AD risk, so as to more accurately set the blood pressure threshold range related to AD risk, thereby ensuring the accuracy of subsequent risk assessment. In addition, since the risk assessment model comprehensively considers a variety of factors related to AD risk, including blood pressure data, age, gender, genetic factors, etc. Therefore, the probability of the target patient being suspected of AD risk calculated based on the model is more reliable, and can provide reliable data support for subsequent priority triage.

[0111] In one embodiment, in step S6, the priority triage based on the target patient's suspected AD possibility assessment result, the severity of clinical symptoms, and the availability of medical resources includes:

[0112] Step S61: classify the target patient into different risk levels according to the suspected AD possibility assessment result.

[0113] Step S62, according to the risk level, comprehensively consider the severity of clinical symptoms and the availability of medical resources, and perform priority triage based on the comprehensive evaluation results.

[0114] It should be noted based on step S61 to step S62 that the risk level includes high risk, medium risk and low risk, wherein each level corresponds to a different suspected AD probability range. Specifically, the present application will classify the target patient into the corresponding risk level according to the suspected AD possibility assessment result through the preset assessment criteria. Among them, the assessment criteria are mainly determined based on a large amount of clinical research data, expert experience and statistical analysis, and can objectively reflect the patient's suspected AD possibility. Further, according to the risk level, when comprehensively considering the severity of clinical symptoms and the availability of medical resources, for each patient, the present application will obtain its clinical symptom data and quantify it to obtain a corresponding quantitative score. Afterwards, the quantitative score of clinical symptoms is comprehensively analyzed with the relevant data of medical resources (such as the number of doctors, the configuration of diagnostic and treatment equipment, and the coverage of medical services, etc.) to obtain the corresponding comprehensive evaluation score. Afterwards, different priority standards are set according to the comprehensive evaluation score, such as "urgent", "priority", "routine", etc. According to the set priority standards, triage decisions will be made for each patient. Among them, patients with high risk and severe symptoms will be marked as "urgent" and will be recommended to be arranged to visit medical institutions or experts with experience in AD diagnosis and treatment; medium-risk patients will be marked as "priority" and will be recommended to receive initial consultation through telemedicine services or to be arranged to follow up at primary medical institutions; low-risk patients will be marked as "routine" and will be recommended to undergo routine health monitoring and management.

[0115] Please refer to Figure 2 The present application discloses an intelligent system for rapid screening and triage of aortic dissection, the system comprising a patient information collection module, a feature extraction and classification module, a correlation analysis module, a blood pressure measurement and recording module, a suspected AD judgment module, and a priority triage module, wherein:

[0116] The patient information collection module is used to collect a text information data set indicating the main symptoms, the first symptoms, the past medical history, and the high-risk signs.

[0117] The feature extraction and classification module is used to identify risk factors based on the text information data set by using a feature extraction and classification algorithm combining natural language processing with machine learning.

[0118] The association analysis module is used to calculate the association between risk factors and the main symptoms and the first symptoms through correlation analysis technology, and to preliminarily evaluate whether the patient belongs to the high-risk group for AD by setting a threshold according to the strength of the association.

[0119] The blood pressure measurement and recording module is used to measure the blood pressure of the limbs of target patients who are initially assessed to be at high risk of AD while activating the emergency green channel, and to record the generated blood pressure data.

[0120] The suspected AD judgment module is used to compare the blood pressure data of the target patient with a preset threshold value, and to determine the possibility of suspected AD in the target patient through logical judgment.

[0121] The priority triage module is used to perform priority triage based on the target patient's suspected AD possibility assessment result, the severity of clinical symptoms, and the availability of medical resources.

[0122] In one embodiment, the above modules are also used to implement an intelligent system method for rapid screening and triage of aortic dissection as described in any of the aforementioned method embodiments, which is not limited in this application.

[0123] As can be seen from the above, the intelligent system for rapid screening and triage of aortic dissection disclosed in the present application utilizes natural language processing and machine learning algorithms for feature extraction and classification, and can automatically and efficiently identify risk factors associated with aortic dissection (AD), thereby reducing the errors and subjectivity of manual judgment; by quantifying the strength of the association between risk factors and symptoms through correlation analysis technology, and setting thresholds for preliminary evaluation, it can quickly screen out people at high risk of AD, thereby improving the efficiency and accuracy of diagnosis; based on the results of suspected AD possibility assessment, the severity of clinical symptoms, and the availability of medical resources, priority triage is performed, thereby ensuring that medical resources can be effectively utilized and improving the overall efficiency and quality of medical services.

[0124] The present application also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the intelligent method for rapid screening and triage of aortic dissection.

[0125] As can be seen from the above, a computer-readable storage medium disclosed in the present application utilizes natural language processing and machine learning algorithms for feature extraction and classification, and can automatically and efficiently identify risk factors associated with aortic dissection (AD), thereby reducing the errors and subjectivity of manual judgment; by quantifying the strength of the association between risk factors and symptoms through correlation analysis technology, and setting thresholds for preliminary evaluation, it is possible to quickly screen out people at high risk of AD, thereby improving the efficiency and accuracy of diagnosis; based on the results of suspected AD possibility assessment, the severity of clinical symptoms, and the availability of medical resources, priority triage is performed, thereby ensuring that medical resources can be effectively utilized and improving the overall efficiency and quality of medical services.

[0126] The present application also discloses an intelligent computing control device for rapid screening and triage of aortic dissection, comprising a communication interface, a memory, a communication bus and a processor, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0127] The memory is used to store computer programs;

[0128] The processor is used to implement the steps of the intelligent method for rapid screening and triage of aortic dissection when executing the program stored in the memory.

[0129] As can be seen from the above, the intelligent computing control device for rapid screening and triage of aortic dissection disclosed in the present application uses natural language processing and machine learning algorithms for feature extraction and classification, and can automatically and efficiently identify risk factors related to aortic dissection (AD), reducing the errors and subjectivity of manual judgment; quantifying the strength of the association between risk factors and symptoms through correlation analysis technology, and setting thresholds for preliminary evaluation, which can quickly screen out people at high risk of AD and improve the efficiency and accuracy of diagnosis; based on the suspected AD possibility assessment results, the severity of clinical symptoms and the availability of medical resources, priority triage is performed to ensure that medical resources can be effectively utilized and improve the overall efficiency and quality of medical services.

[0130] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0131] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent method for rapid screening and triage of aortic dissection, characterized in that: The method comprises: S1. Collect text information datasets that indicate the main symptoms, first symptoms, past medical history, and high-risk signs; S2. Based on the text information data set, risk factors are identified using natural language processing combined with machine learning feature extraction and classification algorithms; S3. Calculate the association between risk factors and the main symptoms and the first symptoms through correlation analysis technology, and set a threshold based on the strength of the association to preliminarily assess whether the patient belongs to the high-risk group for AD; S4. For target patients initially assessed as being at high risk of AD, blood pressure in the limbs should be measured while the emergency green channel is activated, and the resulting blood pressure data should be recorded; S5, comparing the blood pressure data of the target patient with a preset threshold, and determining the possibility that the target patient has suspected AD through logical judgment; S6. Priority triage should be performed based on the target patient’s suspected AD possibility assessment results, the severity of clinical symptoms, and the availability of medical resources.

2. The method according to claim 1, characterized in that In step S2, based on the text information data set, the risk factors are identified by using natural language processing combined with machine learning feature extraction and classification algorithms, including: S21, performing text preprocessing based on the text information data set to obtain a preprocessed text data set; S22. Based on natural language processing technology, extract key features related to the risk of aortic dissection from the preprocessed text dataset through syntactic analysis combined with semantic annotation; S23. Inputting the key features into an unsupervised machine learning model based on a clustering algorithm, and training the model to identify risk factors associated with aortic dissection.

3. The method according to claim 2, characterized in that In step S22, the key features related to the risk of aortic dissection are extracted from the preprocessed text dataset by syntactic analysis combined with semantic annotation based on natural language processing technology, including: S221, parsing the preprocessed text data set based on a syntactic analyzer to obtain a syntactic tree; S222, semantically annotating the words in the text according to the pre-trained semantic model to obtain a semantic annotation result; S223. Construct a knowledge graph based on the syntactic tree and the semantic annotation results, and perform relationship path analysis in the knowledge graph to extract key features related to the risk of aortic dissection.

4. The method according to claim 2, characterized in that: In step S23, the key features are input into an unsupervised machine learning model based on a clustering algorithm, and the model is trained to identify risk factors associated with aortic dissection, including: S231, inputting the key features into an unsupervised machine learning model based on a clustering algorithm, and clustering the key features using a K-means clustering algorithm to group similar features into one category; S232, for each cluster obtained by the final classification, determine the potential risk factor based on the similarity and correlation between the characteristics within the cluster; S233. Perform medical evidence-based verification processing based on various potential risk factors to obtain risk factors associated with aortic dissection.

5. The method according to claim 1, characterized in that In step S3, the correlation analysis technique is used to calculate the correlation between the risk factor and the main complaint symptom and the first symptom, including: S31. Obtain statistical data related to risk factors, presenting symptoms, and first symptoms; S32, quantifying the statistical data to obtain standardized values ​​that can be used for statistical analysis; S33, based on the standardized values, respectively calculating the correlation coefficients between the risk factors and the main symptoms, and the first symptoms; S34. Based on the two correlation coefficients obtained, significance tests and effect size assessments were performed to obtain the degree of association between risk factors and the main symptoms and the first symptoms.

6. The method according to claim 1, characterized in that In step S5, the blood pressure data of the target patient is compared with a preset threshold value, and a logical judgment is performed to determine the possibility that the target patient has suspected AD, including: S51, obtaining blood pressure data of target patients, and setting a blood pressure threshold range associated with AD risk based on data mining and analysis of a large-scale clinical database; S52: When it is determined that the blood pressure data falls within the blood pressure threshold range, the possibility of suspected AD in the target patient is calculated based on the risk assessment model.

7. The method according to claim 1, characterized in that In step S6, the priority triage is performed based on the target patient's suspected AD possibility assessment result, the severity of clinical symptoms, and the availability of medical resources, including: S61, classifying the target patient into different risk levels according to the suspected AD possibility assessment result; S62. According to the risk level, the severity of clinical symptoms and the availability of medical resources are comprehensively considered, and priority triage is carried out based on the comprehensive assessment results.

8. An intelligent system for rapid screening and triage of aortic dissection, characterized in that: The system includes a patient information collection module, a feature extraction and classification module, a correlation analysis module, a blood pressure measurement and recording module, a suspected AD judgment module, and a priority triage module, wherein: The patient information collection module is used to collect a text information data set indicating the main symptoms, the first symptoms, the past medical history, and the high-risk signs; The feature extraction and classification module is used to identify risk factors based on the text information data set by using a feature extraction and classification algorithm combining natural language processing with machine learning; The association analysis module is used to calculate the association between the risk factor and the main symptoms and the first symptoms through correlation analysis technology, and to preliminarily assess whether the patient belongs to the high-risk group of AD by setting a threshold according to the strength of the association; The blood pressure measurement and recording module is used to measure the blood pressure of the four limbs of the target patient who is initially assessed to be at high risk of AD, while activating the emergency green channel, and record the generated blood pressure data; The suspected AD judgment module is used to compare the blood pressure data of the target patient with a preset threshold value, and to determine the possibility of suspected AD in the target patient through logical judgment; The priority triage module is used to perform priority triage based on the target patient's suspected AD possibility assessment result, the severity of clinical symptoms, and the availability of medical resources.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the intelligent method for rapid screening and triage of aortic dissection is implemented.

10. An intelligent computing control device for rapid screening and triage of aortic dissection, comprising a communication interface, a memory, a communication bus and a processor, wherein: The processor, communication interface and memory communicate with each other via the communication bus; The memory is used to store computer programs; The processor is used to implement the steps of the intelligent method for rapid screening and triage of aortic dissection when executing the program stored in the memory.

Images