Method and system for online interaction of cardiovascular disease medical data based on artificial intelligence

Abstract

The present invention relates to the field of medical data processing technology, and in particular to a method and system for online interaction of cardiovascular disease medical data based on artificial intelligence, the method comprising: obtaining and preprocessing a number of historical medical text data and historical medical image data; obtaining initial text features based on initial text data and performing standardization processing; identifying and extracting cardiovascular disease medical features in a number of initial image data; judging the correlation between standard text features and initial image features to construct an initial comprehensive feature set; converting a number of comprehensive feature sets into numerical feature vectors and dividing them, and constructing an initial prediction model based on a training set and a test set; evaluating the accuracy of the initial prediction model based on a validation set, adjusting the initial prediction model parameters according to the accuracy evaluation results, or updating the initial comprehensive feature set to obtain a target prediction model. The present invention improves the accuracy of the online interaction results of cardiovascular disease medical data.

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 online interaction of cardiovascular disease medical data based on artificial intelligence. Background Art

[0002] In the course of cardiovascular disease, due to the long-term shortage of primary medical resources, mistakes in primary medical care are common. Mistakes in initial diagnosis can delay the disease, leading to worsening of the disease, missing the best treatment opportunity and other serious consequences.

[0003] The patent document with Chinese patent application publication number CN112102954A discloses a big data analysis cloud platform system capable of providing smart medical services, the system comprising a general database unit, a human-computer interaction unit and a medical service unit; the general database unit is used to input, update and store information of various diseases and medical resources in real time to form multiple sub-databases; the human-computer interaction unit is used to provide users with a channel to access the system; the medical service unit is used to provide users with online services such as disease query and medical resource query.

[0004] The existing technology only performs simple entry, update and storage of information on various diseases and medical resources, without in-depth preprocessing and feature extraction of the data, and is unable to accurately extract key feature information, resulting in inaccurate online interaction results of medical data. Summary of the invention

[0005] To this end, the present invention provides a method for online interaction of cardiovascular disease medical data based on artificial intelligence. Through multiple links such as text and image data processing, feature standardization, correlation judgment, model construction and evaluation adjustment, the prediction model can be continuously optimized and improved, which can solve the problem of inaccurate output results of online interactive medical data of cardiovascular disease.

[0006] To achieve the above object, the present invention provides a method for online interaction of cardiovascular disease medical data based on artificial intelligence, the method comprising:

[0007] Acquire some historical medical text data and historical medical image data of cardiovascular diseases, pre-process some historical medical text data to obtain some initial text data, pre-process some historical medical images to obtain some initial image data;

[0008] Obtaining initial text features based on the initial text data, matching the initial text features with a standard medical data vocabulary, and performing standard processing on the initial text features according to the matching results to generate standard text features;

[0009] Identifying and extracting medical features of cardiovascular diseases in the initial image data to obtain initial image features, and analyzing the initial image features to obtain disease characterization features according to the analysis results;

[0010] Determine the correlation between the disease representation feature and the standard text feature based on the disease representation database, and construct an initial comprehensive feature set based on the correlation judgment result;

[0011] Taking the initial text data and the initial image data as input sets, taking the initial comprehensive feature sets as output sets, constructing a model construction set based on the input sets and the output sets, dividing the model construction set into a training set, a test set and a validation set, and constructing an initial prediction model based on the training set and the test set;

[0012] An accuracy evaluation is performed on an output result of the initial prediction model based on the validation set, and the initial prediction model is calibrated according to the accuracy evaluation result to obtain a target prediction model.

[0013] Furthermore, the step of preprocessing a number of historical medical text data includes:

[0014] Identify a number of corresponding paragraphs in any of the medical text data, divide the number of paragraphs based on punctuation marks, and obtain a number of sentences;

[0015] The plurality of sentences are processed by a word segmentation tool to obtain a plurality of initial word segments, and the plurality of word segments are used as the initial text data corresponding to any of the medical text data.

[0016] Furthermore, the step of preprocessing a plurality of historical medical images includes:

[0017] Identify a number of edge contour features in any of the historical medical images;

[0018] Identify the pixel values ​​corresponding to the center positions of the initial regions included in the edge contour features, obtain the center pixel value according to the identification result, and determine the pixel value range of the initial region based on the center pixel value;

[0019] Calibrate the initial area based on the pixel value range of the initial area to obtain a plurality of calibration areas;

[0020] A historical heart region and a historical blood vessel region in the historical medical image are determined based on edge contours of a plurality of calibration regions.

[0021] Furthermore, the step of matching the initial text features with a standard medical data vocabulary includes:

[0022] Identify the semantic features of the initial text data, and extract initial symptom features in the initial text data according to the semantic recognition results;

[0023] Performing similarity matching on the initial symptom features and a plurality of standard medical terms in the standard medical data vocabulary to obtain similarity matching results;

[0024] Selecting a corresponding standard medical term with the largest similarity value in the similarity matching results, and mapping the standard medical term to the initial symptom feature to standardize the initial symptom feature to obtain a standard symptom feature;

[0025] Furthermore, the step of identifying and extracting cardiovascular disease medical features from a plurality of the initial image data comprises:

[0026] Comparing the historical heart region with a preset heart image, and marking the abnormal heart region based on the comparison result;

[0027] Comparing the historical vascular region with a preset vascular image, and marking the abnormal vascular region based on the comparison result;

[0028] The abnormal heart region and / or the abnormal blood vessel region are analyzed to obtain disease characterization features.

[0029] Further, the step of analyzing the abnormal heart region and / or the abnormal blood vessel region includes:

[0030] Calculating the abnormal area difference of the abnormal heart region and / or the abnormal blood vessel region;

[0031] determining the abnormal location of the abnormal cardiac region and / or the abnormal vascular region;

[0032] The disease characterization feature is generated based on the abnormal area difference and the abnormal position.

[0033] Furthermore, the step of constructing an initial comprehensive feature set based on the correlation judgment result includes:

[0034] Calculate any of the disease representation features and preset disease representation features in a disease representation database based on a similarity matching algorithm, and extract the preset symptom feature corresponding to the maximum similarity in the calculation results;

[0035] Determine a one-to-one correspondence between the disease characterization feature and the standard symptom feature based on the matching result;

[0036] Based on the relationship determination result, the disease characterization features and the standard symptom features are fused to obtain the initial comprehensive feature set.

[0037] Furthermore, the step of evaluating the accuracy of the output result of the initial prediction model based on the validation set includes:

[0038] Inputting any verification data in the verification set into the initial prediction model to obtain an initial output result;

[0039] Comparing the initial output result with the corresponding preset output result in the verification set, and determining whether it is accurate based on the comparison result;

[0040] The accuracy of the initial prediction model is calculated based on the number of accurate results in the initial output results and the total number of verification data.

[0041] Further, the step of calibrating the initial prediction model according to the accuracy evaluation result includes:

[0042] When the accuracy rate is less than or equal to a first preset accuracy rate, adjusting the initial prediction model parameters to calibrate the initial prediction model;

[0043] When the accuracy is greater than a first preset accuracy, the initial prediction model is not calibrated.

[0044] On the other hand, the present invention also provides a system for an artificial intelligence-based method for online interaction of cardiovascular disease medical data, the system comprising:

[0045] An acquisition module is used to acquire some historical medical text data and historical medical image data of cardiovascular diseases, pre-process some historical medical text data to obtain some initial text data, pre-process some historical medical images to obtain some initial image data;

[0046] A text processing module, connected to the acquisition module, for obtaining initial text features based on the initial text data, matching the initial text features with a standard medical data vocabulary, and performing standard processing on the initial text features according to the matching results to generate standard text features;

[0047] An image recognition module, connected to the acquisition module, for recognizing and extracting medical features of cardiovascular diseases in a number of the initial image data, obtaining initial image features, and analyzing the initial image features to obtain disease characterization features according to the analysis results;

[0048] A feature construction module, used to determine the correlation between the disease representation feature and the standard text feature based on the disease representation database, and to construct an initial comprehensive feature set based on the correlation judgment result;

[0049] A model construction module, connected to the feature construction module, used to take the initial text data and the initial image data as input sets, take the initial comprehensive feature sets as output sets, construct a model construction set based on the input sets and the output sets, divide the model construction set into a training set, a test set and a validation set, and construct an initial prediction model based on the training set and the test set;

[0050] The target acquisition module is connected to the model building module and is used to evaluate the accuracy of the output result of the initial prediction model based on the verification set, and calibrate the initial prediction model according to the accuracy evaluation result to obtain the target prediction model.

[0051] Compared with the prior art, the beneficial effect of the present invention is that, by preprocessing historical medical text data and image data, a solid foundation is provided for subsequent feature extraction and model training, data quality and model accuracy are improved, and initial text features are matched and standardized with standard medical data vocabulary, which can eliminate information ambiguity caused by inconsistent terminology use, making text features more standardized and unified, which helps the model to better understand and utilize text information, improves the generalization ability and accuracy of the model, automatically identifies and extracts medical features of cardiovascular diseases in images, can make full use of the rich information in image data, and further improve the diagnostic accuracy and efficiency of the model. By judging the correlation between standard text features and initial image features, more critical and effective feature combinations for disease prediction are screened out, feature redundancy is reduced, and the prediction efficiency and accuracy of the model are improved. The comprehensive feature set is converted into a numerical feature vector and divided into a training set, a test set, and a validation set, which helps to systematically evaluate and optimize model performance. The accuracy of the model is ensured by constructing a training set and a test set. The accuracy of the model output results is evaluated based on the validation set, and the model parameters are adjusted or the comprehensive feature set is updated according to the evaluation results. Through continuous iteration and optimization, the model gradually approaches the optimal solution, which improves the accuracy and reliability of the model in practical applications.

[0052] In particular, by identifying a number of edge contour features in any of the historical medical images, a basis is provided for subsequent regional segmentation, feature extraction, etc. By identifying edge contours, the processing range is limited to the key parts of the image, reducing the calculation of non-key areas and improving processing efficiency. By considering the central pixel value and its range, the initial area is defined more accurately, reducing misidentification due to noise or poor image quality, providing a reference standard for subsequent regional calibration, making the calibration process more accurate, and the calibrated area closer to the real heart and blood vessel area, providing more accurate input for subsequent regional segmentation, and performing segmentation based on the edge contour of the calibration area to more accurately extract the heart and blood vessel areas, providing a reliable basis for subsequent model construction, and improving the accuracy of model construction.

[0053] In particular, by matching the symptom features in the initial text data with the terms in the standard medical data vocabulary, we ensure that medical information from different sources and in different expressions is unified and standardized, thereby improving the consistency and accuracy of medical data and laying a solid foundation for subsequent data analysis and mining. Standardized symptom features enable disease diagnosis and prediction models to understand and utilize medical data more accurately. When the input data is consistent with the data used in model training in terms of terminology, the model's prediction results will be more reliable and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A schematic flow chart of a method for online interaction of cardiovascular disease medical data based on artificial intelligence provided by an embodiment of the present invention;

[0055] Figure 2 A schematic diagram of a process for preprocessing a plurality of historical medical images in a method for online interaction of cardiovascular disease medical data based on artificial intelligence provided by an embodiment of the present invention;

[0056] Figure 3 A schematic diagram of a process for identifying and extracting medical features of cardiovascular diseases in a method for online interaction of cardiovascular disease medical data based on artificial intelligence provided by an embodiment of the present invention;

[0057] Figure 4 A structural block diagram of a system for online interaction of cardiovascular disease medical data based on artificial intelligence provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0059] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0060] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be understood as a limitation on the present invention.

[0061] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0062] See also Figure 1 As shown, an embodiment of the present invention provides a method for online interaction of cardiovascular disease medical data based on artificial intelligence, the method comprising:

[0063] Step S100, obtaining some historical medical text data and historical medical image data of cardiovascular diseases, preprocessing some historical medical text data to obtain some initial text data, and preprocessing some historical medical images to obtain some initial image data;

[0064] Step S200, obtaining initial text features based on the initial text data, matching the initial text features with a standard medical data vocabulary, and standardizing the initial text features according to the matching results to generate standard text features;

[0065] Step S300, identifying and extracting medical features of cardiovascular diseases in a number of the initial image data to obtain initial image features, and analyzing the initial image features to obtain disease characterization features according to the analysis results;

[0066] Step S400, determining the correlation between the disease representation feature and the standard text feature based on the disease representation database, and constructing an initial comprehensive feature set based on the correlation determination result;

[0067] Step S500, taking the plurality of the initial text data and the plurality of the initial image data as input sets, taking the plurality of the initial comprehensive feature sets as output sets, constructing a model construction set based on the input sets and the output sets, dividing the model construction set into a training set, a test set and a validation set, and constructing an initial prediction model based on the training set and the test set;

[0068] Step S600: perform accuracy evaluation on the output result of the initial prediction model based on the validation set, and calibrate the initial prediction model according to the accuracy evaluation result to obtain a target prediction model.

[0069] Specifically, the sources of historical medical text data and historical image data in the embodiment of the present invention may include medical records, electronic health records, medical images, laboratory test results, etc.

[0070] Specifically, the initial ratio of the training set, the test set, and the validation set in the embodiment of the present invention is 1:3:1.

[0071] Specifically, the embodiment of the present invention provides a solid foundation for subsequent feature extraction and model training by preprocessing historical medical text data and image data, improves data quality and model accuracy, matches and standardizes initial text features with standard medical data vocabulary, can eliminate information ambiguity caused by inconsistent terminology, makes text features more standardized and unified, helps the model better understand and utilize text information, improves the generalization ability and accuracy of the model, automatically identifies and extracts medical features of cardiovascular diseases in images, can make full use of rich information in image data, and further improves the diagnostic accuracy and efficiency of the model. By judging the correlation between standard text features and initial image features, more critical and effective feature combinations for disease prediction are screened out, feature redundancy is reduced, and the prediction efficiency and accuracy of the model are improved. The comprehensive feature set is converted into a numerical feature vector and divided into a training set, a test set, and a validation set, which helps to systematically evaluate and optimize model performance. The accuracy of the model is ensured by constructing a training set and a test set. The accuracy of the model output result is evaluated based on the validation set, and the model parameters are adjusted or the comprehensive feature set is updated according to the evaluation results. Through continuous iteration and optimization, the model gradually approaches the optimal solution, thereby improving the accuracy and reliability of the model in practical applications.

[0072] Specifically, the step of preprocessing a number of historical medical text data includes:

[0073] Identify a number of corresponding paragraphs in any of the medical text data, divide the number of paragraphs based on punctuation marks, and obtain a number of sentences;

[0074] The plurality of sentences are processed by a word segmentation tool to obtain a plurality of initial word segments, and the plurality of word segments are used as the initial text data corresponding to any of the medical text data.

[0075] Specifically, the word segmentation tool in the embodiment of the present invention may be HanLP, Jieba, etc.

[0076] Specifically, the embodiments of the present invention decompose medical text into smaller and easier-to-process units by identifying paragraphs and dividing sentences based on punctuation marks, which facilitates subsequent data processing and analysis and improves the efficiency and accuracy of the entire processing flow. Word segmentation can split the text into meaningful vocabulary units, remove some unnecessary spaces, punctuation marks and other noise, and also reduce redundant information, which helps the model focus more on key information and improves the accuracy of predictions.

[0077] See also Figure 2As shown, the step of preprocessing several historical medical images includes:

[0078] Step S110, identifying a number of edge contour features in any of the historical medical images;

[0079] Step S120, identifying pixel values ​​corresponding to the center positions of the initial regions included in the edge contour features, obtaining the center pixel value according to the identification result, and determining the pixel value range of the initial region based on the center pixel value;

[0080] Step S130, calibrating the initial area based on the pixel value range of the initial area to obtain a plurality of calibration areas;

[0081] Step S140, determining a historical heart region and a historical blood vessel region in the historical medical image based on edge contours of a plurality of calibration regions.

[0082] Specifically, the embodiment of the present invention can identify a number of edge contour features in any of the historical medical images by using edge detection algorithms such as Canny, Sobel, Prewitt, etc.

[0083] Specifically, the center position of the initial area in the embodiment of the present invention can be obtained by calculating the geometric center of the contour, and can be obtained by calculating the weighted average of all pixel coordinates in the contour.

[0084] Specifically, the initial area pixel value range in the embodiment of the present invention is the center pixel value ±10.

[0085] Specifically, the embodiments of the present invention provide a basis for subsequent region segmentation, feature extraction, etc. by identifying a number of edge contour features in any of the historical medical images. By identifying edge contours, the processing range is limited to the key parts of the image, which reduces the calculation of non-key areas and improves processing efficiency. By considering the central pixel value and its range, the initial area is defined more accurately, reducing misidentification due to noise or poor image quality, providing a reference standard for subsequent region calibration, making the calibration process more accurate, and the calibrated area is closer to the real heart and blood vessel area, providing more accurate input for subsequent region segmentation, and performing segmentation based on the edge contour of the calibration area to more accurately extract the heart and blood vessel areas, providing a reliable basis for subsequent model construction, and improving the accuracy of model construction.

[0086] Specifically, the step of matching the initial text features with the standard medical data vocabulary includes:

[0087] Identify the semantic features of the initial text data, and extract initial symptom features in the initial text data according to the semantic recognition results;

[0088] Performing similarity matching on the initial symptom features and a plurality of standard medical terms in the standard medical data vocabulary to obtain similarity matching results;

[0089] Selecting a corresponding standard medical term with the largest similarity value in the similarity matching results, and mapping the standard medical term to the initial symptom feature to standardize the initial symptom feature to obtain a standard symptom feature;

[0090] Specifically, the step of extracting initial symptom features from initial text data according to the semantic recognition result in the embodiment of the present invention includes:

[0091] Perform part-of-speech tagging on several participles and identify the part of speech of each word;

[0092] According to the results of part-of-speech tagging, adjectives and verb phrases in several participles are screened to obtain an initial vocabulary group, several participles are screened again according to the symptom database, and the screening results are combined with the initial vocabulary group to obtain a comprehensive vocabulary group;

[0093] The complete symptom description phrase is extracted by identifying the dependency relationship related to any vocabulary in the comprehensive vocabulary group in the initial text data to obtain the initial symptom feature.

[0094] Specifically, the parts of speech include nouns, verbs, and adjectives;

[0095] Adjectives include painful, uncomfortable, etc., and verb phrases include feeling dizzy, nausea, etc.;

[0096] The symptom database is derived from relevant symptom descriptions in cardiovascular medical disease-related literature, relevant books, and relevant medical records;

[0097] Identifying the dependency relationship in the initial text data related to any vocabulary in the comprehensive vocabulary group can be done through a dependency syntactic analysis tool (Stanford Parser, spaCy, CoreNLP, etc.), wherein the dependency relationship includes subject, predicate, object, etc.

[0098] Specifically, the embodiments of the present invention ensure that medical information from different sources and in different expressions is unified and standardized by performing similarity matching between symptom features in the initial text data and terms in a standard medical data vocabulary, thereby improving the consistency and accuracy of medical data and laying a solid foundation for subsequent data analysis and mining. Standardized symptom features enable disease diagnosis and prediction models to understand and utilize medical data more accurately. When the input data is consistent with the data used in model training in terms of terminology, the model's prediction results will be more reliable and accurate.

[0099] See also Figure 3As shown, the step of identifying and extracting cardiovascular disease medical features in the initial image data includes:

[0100] Step S310, comparing the historical heart region with a preset heart image, and marking the abnormal heart region based on the comparison result;

[0101] Step S320, comparing the historical blood vessel region with a preset blood vessel image, and marking the abnormal blood vessel region based on the comparison result;

[0102] Step S330, analyzing the abnormal heart region and / or the abnormal blood vessel region to obtain disease characterization features.

[0103] Specifically, the preset heart image in the embodiment of the present invention is a normal heart image in the user's historical detection data;

[0104] The preset blood vessel image is a normal blood vessel image in the historical detection data of the user.

[0105] Specifically, the embodiments of the present invention can quickly and accurately identify abnormal parts in historical heart areas by comparing with preset normal heart images, provide key information for subsequent abnormal area analysis, improve the efficiency and accuracy of identification, conduct in-depth analysis of abnormal areas of the heart and blood vessels, more accurately judge abnormal characteristics, and provide an accurate data basis for the construction of subsequent models.

[0106] Specifically, the step of analyzing the abnormal heart region and / or the abnormal blood vessel region includes:

[0107] Calculating the abnormal area difference of the abnormal heart region and / or the abnormal blood vessel region;

[0108] determining the abnormal location of the abnormal cardiac region and / or the abnormal vascular region;

[0109] The disease characterization feature is generated based on the abnormal area difference and the abnormal position.

[0110] Specifically, the abnormal area difference can be calculated by taking the area of ​​the normal heart / blood vessel area as a reference, calculating the area of ​​the abnormal area, and comparing it with the area of ​​the normal area to obtain the area difference;

[0111] Determining the abnormal position may use an image coordinate system or a medical image annotation tool to locate the abnormal region, and divide the image into different anatomical regions or segments to more accurately describe the abnormal position.

[0112] Specifically, an example of an embodiment of the present invention is: assuming that the area of ​​the left ventricle of a normal heart is 100 square centimeters, and the area of ​​the abnormal area of ​​the left ventricle of a patient is 120 square centimeters, the abnormal area difference is 20 square centimeters. In the heart image, the abnormal area is located in the middle and lower segment of the anterior wall of the left ventricle, adjacent to the left anterior descending branch of the coronary artery. The above analysis results are integrated into one or more disease characterization parameters, such as "myocardial hypertrophy of 120 square centimeters in the middle and lower segment of the anterior wall of the left ventricle". Based on the previous analysis, the following disease characterization feature is generated: "There is an abnormal thickening area of ​​20 square centimeters in the middle and lower segment of the anterior wall of the left ventricle, adjacent to the starting point of the left anterior descending branch of the coronary artery."

[0113] Specifically, the embodiments of the present invention quantitatively evaluate the severity of heart or vascular lesions by accurately calculating the abnormal area difference, provide a basis for subsequent model establishment, comprehensively analyze the abnormal area and location, generate comprehensive disease characterization features, and help improve the accuracy of model prediction.

[0114] Specifically, the step of constructing an initial comprehensive feature set based on the correlation judgment result includes:

[0115] Calculate any of the disease representation features and preset disease representation features in a disease representation database based on a similarity matching algorithm, and extract the preset image symptom feature corresponding to the maximum similarity in the calculation results;

[0116] Determine a one-to-one correspondence between the disease characterization feature and the standard symptom feature based on the matching result;

[0117] Based on the relationship determination result, the disease characterization features and the standard symptom features are fused to obtain the initial comprehensive feature set.

[0118] Specifically, the similarity matching algorithm may be cosine similarity or Euclidean distance, etc., which are existing technologies and will not be described in detail here.

[0119] Specifically, an example of an embodiment of the present invention may be that the disease characterization feature is the presence of an abnormally thickened area of ​​20 square centimeters in the middle and lower segments of the left ventricular anterior wall, adjacent to the starting point of the left anterior descending coronary artery, and the preset symptom features obtained by matching the characterization features are chest pain or angina pectoris, shortness of breath, palpitations or arrhythmia after exercise or fatigue, and the standard symptom features are chest pain, shortness of breath, and arrhythmia, then the initial comprehensive feature set is [there is an abnormally thickened area of ​​20 square centimeters in the middle and lower segments of the left ventricular anterior wall, adjacent to the starting point of the left anterior descending coronary artery; chest pain, shortness of breath, and arrhythmia].

[0120] Specifically, the step of evaluating the accuracy of the output result of the initial prediction model based on the validation set includes:

[0121] Inputting any verification data in the verification set into the initial prediction model to obtain an initial output result;

[0122] Comparing the initial output result with the corresponding preset output result in the verification set, and determining whether it is accurate based on the comparison result;

[0123] The accuracy of the initial prediction model is calculated based on the number of accurate results in the initial output results and the total number of verification data.

[0124] Specifically, the accuracy of the initial prediction model described in the embodiment of the present invention = the number of accurate results / the total number of verification data.

[0125] Specifically, the steps for calibrating the initial prediction model based on the accuracy evaluation results include:

[0126] When the accuracy rate is less than or equal to a first preset accuracy rate, adjusting the initial prediction model parameters to calibrate the initial prediction model;

[0127] When the accuracy is greater than a first preset accuracy, the initial prediction model is not calibrated.

[0128] Specifically, the step of adjusting the initial prediction model parameters in the embodiment of the present invention includes:

[0129] Determine the range of parameters that need to be adjusted, such as the weight decay coefficient or the number of hidden layer nodes in a neural network;

[0130] The cross-validation method was used to test different parameter combinations to determine the best parameter combination;

[0131] Based on the cross-validation results, fine-tune the parameters to improve the prediction accuracy of the model;

[0132] Repeat the above steps until the accuracy of the model reaches or exceeds the first preset accuracy.

[0133] Specifically, the first preset accuracy rate in the embodiment of the present invention is 95%.

[0134] Specifically, the embodiments of the present invention continuously iterate and optimize the initial prediction model and the comprehensive feature set to ensure that the final model can accurately identify and predict cardiac and vascular abnormalities, thereby significantly improving the accuracy and reliability of the model.

[0135] See also Figure 4 As shown, an embodiment of the present invention also provides a system for a method for online interaction of cardiovascular disease medical data based on artificial intelligence, the system comprising:

[0136] An acquisition module 10 is used to acquire a number of historical medical text data and historical medical image data of cardiovascular diseases, pre-process the number of historical medical text data to obtain a number of initial text data, and pre-process the number of historical medical images to obtain a number of initial image data;

[0137] A text processing module 20, connected to the acquisition module 10, for obtaining initial text features based on a number of the initial text data, matching the initial text features with a standard medical data vocabulary, and performing standard processing on the initial text features according to the matching results to generate standard text features;

[0138] An image recognition module 30, connected to the acquisition module 10, is used to recognize and extract the medical features of cardiovascular diseases in the initial image data to obtain initial image features, and analyze the initial image features to obtain disease characterization features according to the analysis results;

[0139] A feature construction module 40 is used to determine the correlation between the disease representation feature and the standard text feature based on the disease representation database, and to construct an initial comprehensive feature set based on the correlation determination result;

[0140] A model construction module 50, connected to the feature construction module 40, is used to take the initial text data and the initial image data as input sets, take the initial comprehensive feature sets as output sets, construct a model construction set based on the input sets and the output sets, divide the model construction set into a training set, a test set and a validation set, and construct an initial prediction model based on the training set and the test set;

[0141] The target acquisition module 60 is connected to the model building module 50 and is used to evaluate the accuracy of the output result of the initial prediction model based on the verification set, and calibrate the initial prediction model according to the accuracy evaluation result to obtain the target prediction model.

[0142] Specifically, the system of a method for online interaction of cardiovascular disease medical data based on artificial intelligence provided by an embodiment of the present invention can execute the above-mentioned method for online interaction of cardiovascular disease medical data based on artificial intelligence to achieve the same technical effect, which will not be repeated here.

[0143] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0144] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for online interaction of cardiovascular disease medical data based on artificial intelligence, characterized in that: include: Acquire some historical medical text data and historical medical image data of cardiovascular diseases, pre-process some historical medical text data to obtain some initial text data, pre-process some historical medical image data to obtain some initial image data; Obtaining initial text features based on the initial text data, matching the initial text features with a standard medical data vocabulary, and performing standard processing on the initial text features according to the matching results to generate standard symptom features; Identifying and extracting medical features of cardiovascular diseases in the initial image data to obtain initial image features, and analyzing the initial image features to obtain disease characterization features according to the analysis results; The step of identifying and extracting cardiovascular disease medical features from a plurality of the initial image data comprises: Compare the historical heart region with the preset heart image, and mark the abnormal heart region based on the comparison result; Compare the historical vascular area with the preset vascular image, and mark the abnormal vascular area based on the comparison result; Analyzing the abnormal cardiac region and / or the abnormal vascular region to obtain disease characterization features; The step of analyzing the abnormal heart region and / or the abnormal blood vessel region comprises: Calculating the abnormal area difference of the abnormal heart region and / or the abnormal blood vessel region; determining the abnormal location of the abnormal cardiac region and / or the abnormal vascular region; generating the disease characterization feature based on the abnormal area difference and the abnormal position; Determine the correlation between the disease representation features and the standard symptom features based on the disease representation database, and construct an initial comprehensive feature set based on the correlation judgment result; The step of constructing an initial comprehensive feature set based on the correlation judgment result comprises: Calculate any of the disease representation features and preset disease representation features in a disease representation database based on a similarity matching algorithm, and extract the preset symptom feature corresponding to the maximum similarity in the calculation results; Determine a one-to-one correspondence between the disease characterization feature and the standard symptom feature based on the matching result; Based on the relationship determination result, the disease characterization features and the standard symptom features are integrated to obtain the initial comprehensive feature set; Taking the initial text data and the initial image data as input sets, taking the initial comprehensive feature sets as output sets, constructing a model construction set based on the input sets and the output sets, dividing the model construction set into a training set, a test set and a validation set, and constructing an initial model based on the training set and the test set; An accuracy evaluation is performed on the output result of the initial model based on the validation set, and the initial model is calibrated according to the accuracy evaluation result to obtain a target model.

2. The method for online interaction of cardiovascular disease medical data based on artificial intelligence according to claim 1, characterized in that: The step of preprocessing a number of historical medical text data comprises: Identify a number of corresponding paragraphs in any of the medical text data, divide the number of paragraphs based on punctuation marks, and obtain a number of sentences; The plurality of sentences are processed by a word segmentation tool to obtain a plurality of initial word segments, and the plurality of word segments are used as the initial text data corresponding to any of the medical text data.

3. The method for online interaction of cardiovascular disease medical data based on artificial intelligence according to claim 2, characterized in that: The step of preprocessing a plurality of historical medical image data comprises: Identify a number of edge contour features in any of the historical medical images; Identify the pixel values ​​corresponding to the center positions of the initial regions included in the edge contour features, obtain the center pixel value according to the identification result, and determine the pixel value range of the initial region based on the center pixel value; Calibrate the initial area based on the pixel value range of the initial area to obtain a plurality of calibration areas; A historical heart region and a historical blood vessel region in the historical medical image are determined based on edge contours of a plurality of calibration regions.

4. The method for online interaction of cardiovascular disease medical data based on artificial intelligence according to claim 3 is characterized in that: The step of matching the initial text features with the standard medical data vocabulary comprises: Identify the semantic features of the initial text data, and extract initial symptom features in the initial text data according to the semantic recognition results; Performing similarity matching on the initial symptom features and a plurality of standard medical terms in the standard medical data vocabulary to obtain similarity matching results; The corresponding standard medical term with the largest similarity value in the similarity matching result is selected, and the standard medical term is mapped to the initial symptom feature to standardize the initial symptom feature to obtain the standard symptom feature.

5. The method for online interaction of cardiovascular disease medical data based on artificial intelligence according to claim 4 is characterized in that: The step of evaluating the accuracy of the output result of the initial model based on the validation set comprises: Input any verification data in the verification set into the initial model to obtain an initial output result; Comparing the initial output result with the corresponding preset output result in the verification set, and determining whether it is accurate based on the comparison result; The accuracy of the initial model is calculated based on the number of accurate results in the initial output results and the total number of verification data.

6. The method for online interaction of cardiovascular disease medical data based on artificial intelligence according to claim 5, characterized in that: The steps to calibrate the initial model based on the accuracy assessment results include: When the accuracy rate is less than or equal to a first preset accuracy rate, adjusting the initial model parameters to calibrate the initial model; When the accuracy is greater than a first preset accuracy, the initial model is not calibrated.

7. A system using the method for online interaction of cardiovascular disease medical data based on artificial intelligence as described in any one of claims 1 to 6, characterized in that: include: An acquisition module, used to acquire a number of historical medical text data and historical medical image data of cardiovascular diseases, pre-process the number of historical medical text data to obtain a number of initial text data, pre-process the number of historical medical image data to obtain a number of initial image data; A text processing module, connected to the acquisition module, for obtaining initial text features based on the initial text data, matching the initial text features with a standard medical data vocabulary, and performing standard processing on the initial text features according to the matching results to generate standard symptom features; An image recognition module, connected to the acquisition module, for recognizing and extracting medical features of cardiovascular diseases in a number of the initial image data, obtaining initial image features, and analyzing the initial image features to obtain disease characterization features according to the analysis results; A feature construction module, used to determine the correlation between the disease representation feature and the standard symptom feature based on the disease representation database, and to construct an initial comprehensive feature set based on the correlation judgment result; A model construction module, connected to the feature construction module, used to take the initial text data and the initial image data as input sets, take the initial comprehensive feature sets as output sets, construct a model construction set based on the input sets and the output sets, divide the model construction set into a training set, a test set and a validation set, and construct an initial model based on the training set and the test set; The target acquisition module is connected to the model building module and is used to evaluate the accuracy of the output results of the initial model based on the verification set, and calibrate the initial model according to the accuracy evaluation results to obtain the target model.

Images