CN117558391B - Postoperative condition deep learning method and system based on aortic dissection medical record data - Google Patents

Abstract

The invention discloses a postoperative condition deep learning method and system based on aortic dissection medical history data, which relate to the technical field of postoperative condition deep learning and comprise the following steps: constructing an aortic dissection medical record large database; classifying the information types of the aortic dissection medical record information to obtain total information classification; classifying the information types under the total information classification to obtain sub-information classification; judging the total classification indexes of different total information classifications and calculating the postoperative disease index of the patient; deep learning is performed based on the postoperative disease index of big data; the method is used for solving the problems that the existing deep learning technology of the postoperative condition of the aortic dissection patient also lacks a method for deep learning the abnormal condition of the postoperative condition by combining the data of the integral stage of the aortic dissection medical record and has insufficient effective evaluation of the postoperative condition of the aortic dissection patient.

Description

Postoperative condition deep learning method and system based on aortic dissection medical record data

Technical Field

The invention relates to the technical field of postoperative condition deep learning, in particular to a postoperative condition deep learning method and system based on aortic dissection medical history data.

Background

The existing postoperative condition deep learning technology refers to a method for analyzing and evaluating the postoperative condition change and recovery condition of a patient by applying a deep learning algorithm, wherein the deep learning is a machine learning technology, and related features can be automatically learned and extracted from a large amount of postoperative data by constructing a multi-layer neural network model so as to evaluate and predict the postoperative condition of the patient; aortic dissection refers to the fact that blood in an aortic cavity enters an aortic media from an aortic intima tear to separate the media, and the aortic media is expanded along the long axis direction of the aorta to form a true and false two-cavity separation state of the aortic wall, which induces more causes, such as hypertension, arteriosclerosis, connective tissue diseases, congenital cardiovascular diseases, injuries and the like, and more symptoms of clinical manifestations, such as pain, wherein the pain also divides chest pain, back pain and abdominal pain, the presentation modes of examination project results are different, and examination is usually imaging examination, blood examination and the like; therefore, in the postoperative condition analysis of aortic dissection, more influencing factors need to be considered.

The existing postoperative condition deep learning technology usually evaluates or predicts the postoperative condition of a patient based on the current detection result of the patient, the adopted data is usually only the body data of the patient in the current stage, and the data analysis of the whole treatment stage is lacking, so that the analysis result is inaccurate and cannot be suitable for the postoperative condition analysis of the aortic dissection patient with complicated condition; for example, in chinese patent application publication No. CN111897857a, a method for predicting ICU duration after aortic dissection heart surgery is disclosed, and the method only analyzes clinical data generated during the surgery stage when performing deep learning analysis, lacks multiple-aspect data analysis of patients in different stages, and different symptoms of the disease condition of the patients in different stages can influence the result of the postoperative disease condition to a certain extent, so multiple-aspect consideration is required to ensure accurate analysis results, and the prior art lacks a method for performing deep learning on abnormal conditions of the postoperative disease condition by combining data of the whole stage of aortic dissection medical record, so that doctors and patients lack effective evaluation means after aortic dissection surgery.

Disclosure of Invention

The invention aims to solve at least one of the technical problems in the prior art to a certain extent, the total information classification is divided based on an aortic dissection medical record collection table, medical record information under the total information classification is classified to obtain sub-information classification, different deep learning branches are established based on different first symptoms, in the deep learning branches, the total information classification and the sub-information classification are analyzed to obtain total classification indexes, then the postoperative condition indexes of a patient are calculated according to the total classification indexes, finally, the deep learning is carried out on big data of the postoperative condition indexes, an intelligent index interval is constructed, the range of the index interval and the abnormal parameter probability of each postoperative human parameter corresponding to the index interval are judged, and the problem that the existing deep learning technology of the postoperative condition of an aortic dissection patient lacks a method for carrying out deep learning on abnormal conditions of the postoperative condition by combining data of the whole stage of the aortic dissection medical record, and the effective evaluation of the postoperative condition of the aortic dissection is insufficient is solved.

To achieve the above object, in a first aspect, the present application provides a method for post-operative condition deep learning based on aortic dissection medical history data, comprising the steps of:

Constructing an aortic dissection medical record large database, and recording aortic dissection medical record information;

classifying the information types of the aortic dissection medical record information to obtain total information classification;

classifying the information types under the total information classification to obtain sub-information classification;

analyzing the total information classification and the sub information classification, judging the total classification indexes of different total information classifications, and calculating the postoperative disease index of the patient;

and performing deep learning based on the postoperative condition index of the big data, and judging the parameter abnormality rate of each human parameter of the patient after the operation.

Further, constructing an aortic dissection medical record large database, recording aortic dissection medical record information, wherein the aortic dissection medical record information comprises the steps of establishing data connection with a hospital medical record database, acquiring the aortic dissection medical record therein, and simultaneously acquiring a large amount of aortic dissection medical records through Internet large data and storing the aortic dissection medical record in the aortic dissection medical record large database.

Further, classifying the information types of the aortic dissection medical record information to obtain the total information classification comprises the following sub-steps:

reading a hospital medical record database to obtain an aortic dissection medical record collection table;

classifying the information types of the aortic dissection medical record information based on the aortic dissection medical record collection table to obtain total information classification; the total information classification includes current history data, past history data, imaging exam data, laboratory exam data, intraoperative data, and postoperative data.

Further, classifying the information types under the total information classification to obtain sub-information classification includes the following sub-steps:

acquiring information types of all information under the total information classification;

searching a data form in the information type, and classifying the data form to obtain sub-information classification; the sub-information classification comprises a digital type, a non-digital type and a hierarchical type;

the information types with the data form of numbers are generalized to digital, the information types with the data form of yes or no are generalized to whether the information types are of a formula, and the information types with the data form of different grades are generalized to a grade formula.

Further, analyzing the total information classification and the sub-information classification, judging the total classification index of different total information classifications and calculating the postoperative condition index of the patient comprises the following sub-steps:

acquiring the first symptoms of a patient, establishing different deep learning branches aiming at different first symptoms, marking the deep learning branches as DLBn, wherein n is a positive integer greater than or equal to 1;

connecting a characteristic normal range database, and reading a characteristic normal range;

analyzing DLBn, traversing and reading total information classification in DLBn, and aiming at each total information classification, acquiring the number of sub information classifications under the directory of the total information classification, and marking the number as the classification information number;

Acquiring data of medical records under the sub-information classification directory, and marking the data as medical record data;

acquiring medical record data, judging sub-information classification of the medical record data, and outputting a range detection signal if the medical record data is digital; if the signal is of the formula, outputting a detection signal; if the signal is the grade type, outputting a grade detection signal;

and carrying out different processing on the medical record data according to different detection signals, and outputting analysis signals after the processing is finished.

Further, the medical record data is processed differently according to different detection signals, and the analysis signal is output after the processing is completed, which comprises the following sub-steps:

if the range detection signal is output, acquiring a characteristic normal range corresponding to the medical record data, searching whether the medical record data is in the characteristic normal range, and if so, outputting a normal range signal; if not, outputting an abnormal range signal;

if the detection signal is output, acquiring text characters of medical record data, and if the text characters are yes, outputting a characteristic abnormal signal; if the text characters are 'no', outputting a characteristic normal signal;

if the grade detection signal is output, reading whether the medical record data is zero grade, and if the grade of the medical record data is zero grade, outputting a grade normal signal; outputting a grade abnormality signal if the grade of the medical record data is not zero;

Analyzing and calculating the analysis signals, and judging the total classification index of each total information classification; the analysis signal includes a normal range signal, an abnormal range signal, a characteristic abnormal signal, a characteristic normal signal, a grade normal signal, and a grade abnormal signal.

Further, the analyzing and calculating are carried out on the analysis signals, and the judging of the total classification index of each total information classification comprises the following substeps:

setting a range classification index, namely initially zero, marking the range classification index as SCI, receiving an abnormal range signal, and adding one to the SCI if the abnormal range signal is received;

setting a characteristic classification index, namely initially zero, marking the characteristic classification index as FCI, receiving a characteristic abnormal signal, and adding one to FCI if the characteristic abnormal signal is received;

setting a grade classification index, namely, initially setting the grade classification index as zero, marking the grade classification index as GCI, receiving a grade abnormality signal, acquiring the disease grade of medical record data if the grade abnormality signal is received, and increasing the GCI by 1+ (X-1) multiplied by 0.2 if the disease grade is X grade, wherein X is a constant and is a positive integer;

receiving all analysis signals, and obtaining SCI, FCI and GCI after receiving;

searching the quantity of the sub-information types of the total information classification, which are digital, whether or not and the class type, and marking the quantity as a range classification standard, a feature classification standard and a class classification standard;

Calculating through a total classification index calculation formula to obtain a total classification index;

the total classification index calculation formula is configured as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, AC is the total classification index, SCC is the range classification standard, FCC is the characteristic classification standard, GCC is the grade classification standard;

and (3) analyzing and calculating all the total information classifications, and calculating the total classification index to obtain the postoperative condition index of the patient.

Further, the method comprises the following sub-steps of:

analyzing and calculating total classification indexes of current medical history data, past medical history data, imaging examination data, laboratory examination data and intraoperative data, which are respectively named as total current medical history indexes, total past medical history indexes, total imaging indexes, total laboratory indexes and total intraoperative indexes, and are sequentially marked as AC1 to ACm, and m is a constant and a positive integer;

obtaining the minimum value and the maximum value in the ACs 1 to ACm, and marking the minimum value and the maximum value as an abnormal index and a normal index respectively;

setting an abnormality index, and setting the abnormality index as a first preset index if the abnormality index is a total current medical history index; if the abnormality index is the total past history index, setting the abnormality index as a second preset index; if the abnormality index is the total imaging index, setting the abnormality index as a third preset index; if the abnormality index is the total laboratory index, setting the abnormality index as a fourth preset index; if the abnormality index is the total operation index, setting the abnormality index as a fifth preset index;

Setting a normal index, and if the normal index is a total current medical history index, setting the normal index as a first preset index; if the normal index is the total past history index, setting the normal index as a second preset index; if the normal index is the total imaging index, setting the normal index as a third preset index; if the normal index is the total laboratory index, setting the normal index as a fourth preset index; if the normal index is the total intraoperative index, setting the normal index as a fifth preset index;

calculating by using a postoperative condition index calculation formula to obtain postoperative condition indexes of patients;

the postoperative condition index calculation formula is configured as follows:the method comprises the steps of carrying out a first treatment on the surface of the Pte is an index of postoperative condition, T is an abnormality index, and Q is a normal index.

Further, deep learning is performed based on the postoperative condition index of big data, and the judgment of the parameter abnormality rate of each human parameter after the operation of the patient comprises the following sub-steps:

calculating postoperative disease indexes of all aortic dissection medical records based on the aortic dissection medical records large database, numbering the postoperative disease indexes, wherein the marks are Ps, s is a constant and is a positive integer;

establishing a postoperative condition index coordinate system by taking Ps as an X axis and postoperative condition indexes as a Y axis, and inputting all postoperative condition indexes into the postoperative condition index coordinate system; marking the number of postoperative disease indexes as index numbers;

Setting an index range, taking a postoperative condition index as a starting point, increasing the index range, and simultaneously acquiring abnormal items of human parameters in postoperative data in aortic dissection medical records corresponding to the postoperative condition index in the index range, wherein the abnormal items are marked as abnormal parameters;

counting all postoperative disease indexes in an index range, counting the number of different abnormal parameters, dividing the number of the abnormal parameters by the index number to obtain a parameter abnormal rate, increasing the index range, and outputting a range determining signal when the index number in the index range is greater than or equal to a first number threshold;

calculating the parameter anomaly rate, if the index number is greater than or equal to a second number threshold value and any parameter anomaly rate is equal to a first anomaly probability, continuing to increase the index range, monitoring whether the corresponding parameter anomaly rate is reduced, if the parameter anomaly rate is reduced, outputting a range determination signal, and if the parameter anomaly rate is increased, continuing to increase the index range;

if the range determining signal is output, stopping increasing the index range, and simultaneously taking the maximum value of the current index range as a starting point, and continuously analyzing the next index range until the index range comprises all postoperative disease indexes;

outputting the parameter anomaly rates of different human parameters corresponding to different index ranges.

In a second aspect, the application provides a postoperative condition deep learning system based on aortic dissection medical record data, which comprises a big data collection module, an information classification module, a postoperative condition analysis module and a postoperative condition deep learning module; the big data collection module, the information classification module and the postoperative condition deep learning module are respectively connected with the postoperative condition analysis module in a data mode;

the big data collection module is used for constructing an aortic dissection medical record big database and recording aortic dissection medical record information;

the information classification module is used for classifying the information types of the aortic dissection medical record information to obtain total information classification; classifying the information types under the total information classification to obtain sub-information classification;

the postoperative condition analysis module is used for analyzing the total information classification and the sub-information classification, judging the total classification indexes of different total information classifications and calculating the postoperative condition index of the patient;

the postoperative condition deep learning module is used for carrying out deep learning based on the postoperative condition index of big data and judging the parameter abnormality rate of each human parameter after the operation of a patient.

The invention has the beneficial effects that: according to the invention, the complicated medical record information is divided into different total information classifications by referring to the aortic dissection medical record collection table, and then the medical record information in the total information classifications is divided into different sub-information classifications;

According to the invention, different abnormal judgment is carried out on different sub-information classifications, and then the total classification index of the total information classification is calculated according to the result of the abnormal judgment, so that the method has the advantages that the data types of medical record information of the aortic dissection are different, the accurate result is difficult to obtain by adopting the unified judgment standard, the different data types are different judgment standards, and the total classification index of the total information classification is calculated according to the judgment result, so that the physical condition of the aortic dissection patient in the corresponding treatment stage can be effectively reflected, a data basis is provided for further data analysis, and the accuracy and the comprehensiveness of postoperative condition analysis are further improved;

according to the invention, the total classification index is calculated to obtain the postoperative condition index, so that the current physical condition of the aortic dissection patient is reflected, statistics is carried out through big data, a postoperative condition index coordinate system is drawn, and deep learning is carried out according to the postoperative condition index coordinate system and postoperative data of the patient, so that the parameter abnormality rates of different human parameters in different index ranges are judged;

According to the invention, by setting different index ranges, the index ranges are expanded until the index number in the index ranges is greater than or equal to a first quantity threshold or the abnormal rate of any parameter is greater than a first abnormal probability and then is reduced, and the method has the advantages that the data base in the current index ranges can be ensured to be sufficient through the first quantity threshold if no abnormal rate of the parameter can be greater than or equal to the first abnormal probability, the current index ranges can be ensured to have directivity when the abnormal rate of any parameter is greater than the first abnormal probability and then is reduced, and the accuracy and the reliability of postoperative condition analysis of aortic dissection patients can be improved.

Additional features and advantages of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application. The objectives and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.

Drawings

FIG. 1 is a flow chart of the steps of the method of the present invention;

FIG. 2 is a flowchart illustrating steps for calculating a total classification indicator according to the present invention;

FIG. 3 is a schematic diagram of a post-operative condition index coordinate system according to the present invention;

Fig. 4 is a functional block diagram of the system of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Embodiment 1 referring to fig. 1, in a first aspect, the present application provides a method for post-operative condition deep learning based on aortic dissection medical history data, comprising the steps of:

step S1, constructing an aortic dissection medical record large database, and recording aortic dissection medical record information; constructing an aortic dissection medical record large database, wherein recording the aortic dissection medical record information comprises establishing data connection with a hospital medical record database, acquiring the aortic dissection medical record therein, and simultaneously acquiring a large amount of aortic dissection medical records through Internet large data and storing the large amount of aortic dissection medical records into the aortic dissection medical record large database;

in specific implementations, the partial data of the aortic dissection medical record large database are shown in table 1:

Table 1 aortic dissection medical records information statistics table

Wherein, N in hypertension (grading) represents zero level, I is first level, II is second level, III is third level; n in the other medical record data represents no, and Y represents yes.

Step S2, classifying the information types of the aortic dissection medical record information to obtain total information classification; step S2 comprises the following sub-steps:

step S201, reading a hospital medical record database and obtaining an aortic dissection medical record collection table;

step S202, classifying information types of aortic dissection medical record information based on an aortic dissection medical record collection table to obtain total information classification; the total information classification comprises current history data, past history data, imaging examination data, laboratory examination data, intraoperative data and postoperative data;

in particular implementations, the aortic dissection medical records collection table part table is shown in table 2:

table 2 aortic dissection medical records collection table

Step S3, classifying the information types under the total information classification to obtain sub information classification; step S3 comprises the following sub-steps:

step S301, obtaining information types of all information under the total information classification;

step S302, searching a data form in the information type, and classifying the data form to obtain sub-information classification; the sub information classification includes digital, whether or not, and hierarchical;

Step S303, the information types with the data form of numbers are generalized to digital, the information types with the data form of yes or no are generalized to whether the information types are of a formula, and the information types with the data form of different grades are generalized to a grade formula;

in a specific implementation, the information types in the obtained past history data comprise hypertension (grading), ventricular blood pressure, heart rate, diabetes, coronary heart disease, cerebrovascular disease, chronic renal insufficiency and COPD; the data form of the hypertension (grading) is obtained by searching and is the grade, and the hypertension (grading) is generalized into the grade; searching to obtain the data form of the diabetes mellitus as Y or N, namely, if yes or no, and inducing the diabetes mellitus into a formula or not; searching for the data form of the heart rate, and summarizing the heart rate into a digital form; referring to the method, the digital data of sub-information classification obtained by analysis comprises the blood pressure and heart rate of the person entering the room; the ranking of the sub-information categories includes hypertension (ranking); whether or not the sub-information classification includes diabetes, coronary heart disease, cerebrovascular disease, chronic renal insufficiency, and COPD.

S4, analyzing the total information classification and the sub-information classification, judging the total classification indexes of different total information classifications, and calculating the postoperative disease index of the patient; step S4 comprises the following sub-steps:

Step S401, acquiring first symptoms of a patient, establishing different deep learning branches aiming at different first symptoms, marking the deep learning branches as DLBn, wherein n is a positive integer greater than or equal to 1;

in specific implementation, the first symptoms comprise chest pain, back pain, abdominal pain, chest pain plus back pain, chest pain plus abdominal pain, back pain plus abdominal pain and syncope, 7 first symptoms are counted, 7 deep learning branches are constructed, DLB1 to DLB7 are sequentially constructed, and n is 7;

step S402, connecting a characteristic normal range database and reading a characteristic normal range;

step S403, analyzing DLBn, traversing and reading total information classifications in DLBn, and aiming at each total information classification, obtaining the number of sub information classifications under the directory of the total information classification, and marking the number as the classified information number;

step S404, obtaining the data of the medical record under the sub-information classification directory, and marking the data as the medical record data;

step S405, obtaining medical record data, judging sub-information classification of the medical record data, and outputting a range detection signal if the medical record data is digital; if the signal is of the formula, outputting a detection signal; if the signal is the grade type, outputting a grade detection signal;

in specific implementation, in this embodiment, chest pain is taken as an example to further explain the analysis process of the deep learning branch, and the analysis process of other deep learning branches is performed by referring to the chest pain deep learning branch; the total information classification in DLB1 is read through, the past history data is taken as an example, the number of the obtained classified information is 8, and the digital type of the sub information classification comprises the blood pressure and heart rate of entering a room; the ranking of the sub-information categories includes hypertension (ranking); whether the sub information classification includes diabetes, coronary heart disease, cerebrovascular disease, chronic renal insufficiency and COPD, so that the blood pressure in the room and heart rate output range detection signals, the hypertension (classification) output grade detection signals, and whether the rest of medical record data output detection signals;

Step S406, performing different processing on the medical record data according to different detection signals, and outputting analysis signals after the processing is completed;

step S406 includes the following sub-steps:

step S40601, if the range detection signal is output, acquiring a characteristic normal range corresponding to the medical record data, searching whether the medical record data is in the characteristic normal range, and if so, outputting a normal range signal; if not, outputting an abnormal range signal;

in specific implementation, the blood pressure in the room and the heart rate output range detection signals, wherein the blood pressure in the room is 132/65, the systolic pressure is 132, the diastolic pressure is 65, the systolic pressure normal range in the blood pressure in the room is obtained (90, 140), the diastolic pressure normal range is (60, 90), and the systolic pressure 132 and the diastolic pressure 65 of the blood pressure in the room are obtained through comparison and are in the characteristic normal range, and then the normal range signals are output; the heart rate is obtained to be 62, the heart rate normal range is (50, 100), and the heart rate is obtained to be within the heart rate normal range through comparison, and then a normal range signal is output;

step S40602, if the detection signal is output, text characters of the medical record data are obtained, and if the text characters are yes, a characteristic abnormal signal is output; if the text characters are 'no', outputting a characteristic normal signal;

In the specific implementation, the obtained text characters of diabetes, coronary heart disease, cerebrovascular disease and chronic renal insufficiency are N, namely, if not, 4 times of characteristic normal signals are output, and the obtained text characters in COPD are Y, namely, characteristic abnormal signals are output;

step S40603, if the grade detection signal is output, reading whether the medical record data is zero-grade, and if the grade of the medical record data is zero-grade, outputting a grade normal signal; outputting a grade abnormality signal if the grade of the medical record data is not zero;

in the implementation, when the acquired hypertension (grading) is two-level, outputting a grade abnormality signal;

step S40604, analyzing and calculating the analysis signals, and judging the total classification index of each total information classification; the analysis signals comprise a normal range signal, an abnormal range signal, a characteristic abnormal signal, a characteristic normal signal, a grade normal signal and a grade abnormal signal;

step S40605, setting a range classification index, initially setting to zero, marking as SCI, receiving an abnormal range signal, and adding one to SCI if the abnormal range signal is received;

step S40606, setting a characteristic classification index, namely, initially setting the characteristic classification index as zero, marking the characteristic classification index as FCI, receiving a characteristic abnormal signal, and adding one to the FCI if the characteristic abnormal signal is received;

Step S40607, setting a grade classification index, namely, initially setting the grade classification index as zero, marking the grade classification index as GCI, receiving a grade abnormality signal, acquiring the illness state grade of medical record data if the grade abnormality signal is received, and increasing the GCI by 1+ (X-1) multiplied by 0.2 if the illness state grade is X grade, wherein X is a constant and is a positive integer;

step S40608, receiving all analysis signals, and obtaining SCI, FCI and GCI after receiving;

in specific implementation, if the SCI is initially 0 and no abnormal range signal is output, the SCI is unchanged and is 0; the FCI is initially 0, and when the characteristic abnormal signal is output once, the FCI is added by one, and the FCI is 1; the GCI is initially 0, and when the abnormal signal of the output grade of the hypertension (grade) is obtained, the disease grade of the hypertension (grade) is obtained, and is two-grade, and the GCI is added with 1+ (2-1) multiplied by 0.2, so that the GCI is 1.2;

step S40609, respectively searching the quantity of the sub-information types in the total information classification, namely digital type, whether type and grade type, and respectively marking the quantity as a range classification standard, a characteristic classification standard and a grade classification standard;

step S406410, calculating through a total classification index calculation formula to obtain a total classification index;

the total classification index calculation formula is configured as:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, AC is the total classification index, SCC is the range classification standard, FCC is the characteristic classification standard, GCC is the grade classification standard;

In the specific implementation, searching to obtain a range classification standard SCC of 2, a characteristic classification standard FCC of 5, a grade classification standard GCC of 1, SCI of 0, FCI of 1 and GCI of 1.2, and calculating to obtain a total classification index AC of 1.6;

step S406411, analyzing and calculating all total information classifications, and calculating total classification indexes to obtain postoperative disease indexes of patients;

step S406412, analyzing and calculating total classification indexes of current medical history data, past history data, imaging examination data, laboratory examination data and intraoperative data, which are respectively named as total current medical history indexes, total past history indexes, total imaging indexes, total laboratory indexes and total intraoperative indexes, and are sequentially marked as AC1 to ACm, and m is a constant and a positive integer;

in specific implementation, the total current medical history index, the total past history index, the total imaging index, the total laboratory index and the total intraoperative index are sequentially 2.4, 1.6, 2.3, 1.8 and 2.5, namely, AC1 to AC5 are sequentially 2.4, 1.6, 2.3, 1.8 and 2.5, and m is 5;

step S406413, obtaining the minimum and maximum values of AC1 to ACm, respectively marked as an abnormal index and a normal index;

step S406414, setting an abnormality index, and if the abnormality index is a total current medical history index, setting the abnormality index as a first preset index; if the abnormality index is the total past history index, setting the abnormality index as a second preset index; if the abnormality index is the total imaging index, setting the abnormality index as a third preset index; if the abnormality index is the total laboratory index, setting the abnormality index as a fourth preset index; if the abnormality index is the total operation index, setting the abnormality index as a fifth preset index;

Step S406415, setting a normal index, and if the normal index is a total current medical history index, setting the normal index as a first preset index; if the normal index is the total past history index, setting the normal index as a second preset index; if the normal index is the total imaging index, setting the normal index as a third preset index; if the normal index is the total laboratory index, setting the normal index as a fourth preset index; if the normal index is the total intraoperative index, setting the normal index as a fifth preset index;

in specific implementation, the first preset index is set to 10, the second preset index is set to 20, the third preset index is set to 30, the fourth preset index is set to 40, and the fifth preset index is set to 50; the minimum value and the maximum value in the obtained AC1 to ACm are AC2 and AC5 respectively, namely, the abnormal index and the normal index are the total past history index and the total intraoperative index respectively; the abnormality index is set to 20 and the normal index is set to 50;

step S406416, calculating by a postoperative condition index calculation formula to obtain postoperative condition indexes of the patient;

the postoperative condition index calculation formula is configured as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein Pte is postoperative condition index, T is abnormality index, and Q is normal index;

In specific implementation, the abnormality index T is 20, the normal index Q is 50, and the AC1 to AC5 are 2.4, 1.6, 2.3, 1.8 and 2.5, respectively, so that the postoperative condition index Pte is 4.24 by calculation.

Referring to fig. 2, step S5 is performed to perform deep learning based on the postoperative condition index of the big data, and determine the abnormal parameter rate of each human parameter after the operation performed by the patient; step S5 comprises the following sub-steps:

step S501, calculating postoperative condition indexes of all aortic dissection medical records based on the aortic dissection medical record large database, numbering the postoperative condition indexes, wherein the marks are Ps, and S is a constant and a positive integer;

referring to fig. 3, step S502 establishes a post-operation condition index coordinate system with Ps as the X axis and post-operation condition index as the Y axis, and inputs all post-operation condition indexes into the post-operation condition index coordinate system; marking the number of postoperative disease indexes as index numbers;

in the specific implementation, the number of the postoperative condition indexes is calculated to be 10 ten thousand, s is 1 to 10 ten thousand, ps is taken as an X axis, the postoperative condition indexes are taken as a Y axis, a postoperative condition index coordinate system is established as shown in figure 3, and the index number is used for representing the number of the postoperative condition indexes in an index range;

step S503, setting an index range, taking the postoperative condition index as a starting point, increasing the index range, and simultaneously obtaining abnormal items of human parameters in postoperative data in aortic dissection medical records corresponding to the postoperative condition index in the index range, wherein the abnormal items are marked as abnormal parameters;

In specific implementation, setting an index range, namely an increased index range, namely an index range starting from 0 and continuously expanding, wherein when the index range is expanded to (0,4.25), postoperative disease indexes in the range comprise P1 and P1 is 4.24, the analysis process of P1 is explained, and the analysis process of the rest Ps is carried out with reference to P1; reading an aortic dissection medical record of P1, classifying postoperative data after the postoperative data, reading various human parameters in the postoperative data, and searching abnormal items in the postoperative data, wherein if postoperative hypoxia is marked as abnormal in the postoperative data, the postoperative hypoxia is marked as abnormal parameters;

step S504, counting all postoperative disease indexes in an index range, counting the number of different abnormal parameters, dividing the number of the abnormal parameters by the index number to obtain a parameter abnormal rate, increasing the index range, and outputting a range determining signal when the index number in the index range is greater than or equal to a first number threshold;

step S505, calculating the parameter anomaly rate, if the index number is greater than or equal to a second number threshold value and any parameter anomaly rate is equal to a first anomaly probability, continuing to increase the index range, simultaneously monitoring whether the corresponding parameter anomaly rate is reduced, if the parameter anomaly rate is reduced, outputting a range determination signal, and if the parameter anomaly rate is increased, continuing to increase the index range;

In specific implementation, the first number threshold is set to 10000, the first abnormality probability is set to 0.8, taking postoperative hypoxia as an example, in an index range (0,4.25), the index number is 3824, wherein the number of the postoperative hypoxia is 2321, the parameter abnormality rate obtained by statistics is 2321/3824 and is 0.61, the calculated result is kept in two decimal places, the index range is continuously increased by comparing that the index number is smaller than the first number threshold and the parameter abnormality rate is smaller than the first abnormality probability, when the index range is increased to (0,10.34), the index number at the moment is 8965 and smaller than the first number threshold, the parameter abnormality rate of the postoperative hypoxia is 0.8, the index range is continuously increased, the parameter abnormality rate is continuously increased from 0.8 to 0.83 in the process of increasing the index range to (0,11.25), when the parameter abnormality rate is detected to start decreasing at the moment, the index range is continuously increased, the index range is maintained at (0,11.25) and a range determination signal is output;

step S506, if the range determination signal is output, stopping increasing the index range, and simultaneously taking the maximum value of the current index range as a starting point, and continuing to analyze the next index range until the index range includes all postoperative disease indexes;

Step S507, outputting parameter anomaly rates of different human parameters corresponding to different index ranges;

in the specific implementation, when the range determining signal is output, the increase of the index range is stopped, the index range is 0,11.25, the next index range is continuously analyzed, and the next index range starts from 11.25, and different index ranges are analyzed until the index range includes all postoperative disease indexes; taking the index range (0,11.25) as an example, the partial parameter anomaly rate is as follows: the abnormal rate of leucocyte is 0.23, the abnormal rate of neutral ratio is 0.12, the abnormal rate of granulesten is 0.08, the abnormal rate of platelet is 0.34, and the abnormal rate of postoperative hypoxia is 0.83.

Embodiment 2 referring to fig. 4, in a second aspect, the present application provides a postoperative condition deep learning system based on aortic dissection medical history data, which includes a big data collection module, an information classification module, a postoperative condition analysis module, and a postoperative condition deep learning module; the big data collection module, the information classification module and the postoperative condition deep learning module are respectively connected with the postoperative condition analysis module in a data manner;

the big data collection module is used for constructing an aortic dissection medical record big database and recording aortic dissection medical record information; acquiring aortic dissection medical records therein, and simultaneously acquiring a large amount of aortic dissection medical records through Internet big data and storing the aortic dissection medical records into an aortic dissection medical record big database;

The information classification module is used for classifying the information types of the aortic dissection medical record information to obtain total information classification; classifying the information types under the total information classification to obtain sub-information classification; the information classification module comprises a total information classification unit and a sub information classification unit;

the total information classification unit is used for classifying the information types of the aortic dissection medical record information to obtain total information classification;

the total information classification unit is configured with a total information classification policy, which includes:

reading a hospital medical record database to obtain an aortic dissection medical record collection table;

classifying the information types of the aortic dissection medical record information based on the aortic dissection medical record collection table to obtain total information classification; the total information classification comprises current history data, past history data, imaging examination data, laboratory examination data, intraoperative data and postoperative data;

the sub-information classification unit is used for classifying the information types under the total information classification to obtain sub-information classification;

the sub-information classification unit is configured with a sub-information classification policy, the sub-information classification policy including:

acquiring information types of all information under the total information classification;

Searching a data form in the information type, and classifying the data form to obtain sub-information classification; the sub information classification includes digital, whether or not, and hierarchical;

the method comprises the steps of inducing the information types with the data form of numbers into numbers, inducing the information types with the data form of yes or no into a formula of whether or not, and inducing the information types with the data form of different grades into a grade formula;

the postoperative condition analysis module is used for analyzing the total information classification and the sub-information classification, judging the total classification indexes of different total information classifications and calculating the postoperative condition index of the patient; the postoperative condition analysis module comprises a detection scheme analysis unit, a medical record data detection unit, a total classification index calculation unit and a postoperative condition index calculation unit;

the detection scheme analysis unit is configured with a detection scheme analysis strategy including:

acquiring the first symptoms of a patient, establishing different deep learning branches aiming at different first symptoms, marking the deep learning branches as DLBn, wherein n is a positive integer greater than or equal to 1;

connecting a characteristic normal range database, and reading a characteristic normal range;

analyzing DLBn, traversing and reading total information classification in DLBn, and aiming at each total information classification, acquiring the number of sub information classifications under the directory of the total information classification, and marking the number as the classification information number;

Acquiring data of medical records under the sub-information classification directory, and marking the data as medical record data;

acquiring medical record data, judging sub-information classification of the medical record data, and outputting a range detection signal if the medical record data is digital; if the signal is of the formula, outputting a detection signal; if the signal is the grade type, outputting a grade detection signal;

performing different processing on the medical record data according to different detection signals, and outputting analysis signals after the processing is completed;

the medical record data detection unit is configured with a medical record data detection strategy, and the medical record data detection strategy comprises:

if the range detection signal is output, acquiring a characteristic normal range corresponding to the medical record data, searching whether the medical record data is in the characteristic normal range, and if so, outputting a normal range signal; if not, outputting an abnormal range signal;

if the detection signal is output, acquiring text characters of medical record data, and if the text characters are yes, outputting a characteristic abnormal signal; if the text characters are 'no', outputting a characteristic normal signal;

if the grade detection signal is output, reading whether the medical record data is zero grade, and if the grade of the medical record data is zero grade, outputting a grade normal signal; outputting a grade abnormality signal if the grade of the medical record data is not zero;

Analyzing and calculating the analysis signals, and judging the total classification index of each total information classification; the analysis signals comprise a normal range signal, an abnormal range signal, a characteristic abnormal signal, a characteristic normal signal, a grade normal signal and a grade abnormal signal;

the total classification index calculation unit is configured with a total classification index calculation policy including:

setting a range classification index, namely initially zero, marking the range classification index as SCI, receiving an abnormal range signal, and adding one to the SCI if the abnormal range signal is received;

setting a characteristic classification index, namely initially zero, marking the characteristic classification index as FCI, receiving a characteristic abnormal signal, and adding one to FCI if the characteristic abnormal signal is received;

setting a grade classification index, namely, initially setting the grade classification index as zero, marking the grade classification index as GCI, receiving a grade abnormality signal, acquiring the disease grade of medical record data if the grade abnormality signal is received, and increasing the GCI by 1+ (X-1) multiplied by 0.2 if the disease grade is X grade, wherein X is a constant and is a positive integer;

receiving all analysis signals, and obtaining SCI, FCI and GCI after receiving;

searching the quantity of the sub-information types of the total information classification, which are digital, whether or not and the class type, and marking the quantity as a range classification standard, a feature classification standard and a class classification standard;

Calculating through a total classification index calculation formula to obtain a total classification index;

the total classification index calculation formula is configured as:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, AC is the total classification index, SCC is the range classification standard, FCC is the characteristic classification standard, GCC is the grade classification standard;

analyzing and calculating all total information classifications, and calculating total classification indexes to obtain postoperative disease indexes of patients;

the postoperative condition index calculating unit is configured with a postoperative condition index calculating strategy, and the postoperative condition index calculating strategy comprises:

analyzing and calculating total classification indexes of current medical history data, past medical history data, imaging examination data, laboratory examination data and intraoperative data, which are respectively named as total current medical history indexes, total past medical history indexes, total imaging indexes, total laboratory indexes and total intraoperative indexes, and are sequentially marked as AC1 to ACm, and m is a constant and a positive integer;

obtaining the minimum value and the maximum value in the ACs 1 to ACm, and marking the minimum value and the maximum value as an abnormal index and a normal index respectively;

setting an abnormality index, and setting the abnormality index as a first preset index if the abnormality index is a total current medical history index; if the abnormality index is the total past history index, setting the abnormality index as a second preset index; if the abnormality index is the total imaging index, setting the abnormality index as a third preset index; if the abnormality index is the total laboratory index, setting the abnormality index as a fourth preset index; if the abnormality index is the total operation index, setting the abnormality index as a fifth preset index;

Setting a normal index, and if the normal index is a total current medical history index, setting the normal index as a first preset index; if the normal index is the total past history index, setting the normal index as a second preset index; if the normal index is the total imaging index, setting the normal index as a third preset index; if the normal index is the total laboratory index, setting the normal index as a fourth preset index; if the normal index is the total intraoperative index, setting the normal index as a fifth preset index;

calculating by using a postoperative condition index calculation formula to obtain postoperative condition indexes of patients;

the postoperative condition index calculation formula is configured as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein Pte is postoperative condition index, T is abnormality index, and Q is normal index;

the postoperative condition deep learning module is used for carrying out deep learning based on the postoperative condition index of big data and judging the parameter abnormality rate of each human parameter after the operation of a patient;

the postoperative condition deep learning module is configured with a postoperative condition deep learning strategy, and the postoperative condition deep learning strategy comprises:

calculating postoperative disease indexes of all aortic dissection medical records based on the aortic dissection medical records large database, numbering the postoperative disease indexes, wherein the marks are Ps, s is a constant and is a positive integer;

Establishing a postoperative condition index coordinate system by taking Ps as an X axis and postoperative condition indexes as a Y axis, and inputting all postoperative condition indexes into the postoperative condition index coordinate system; marking the number of postoperative disease indexes as index numbers;

setting an index range, taking a postoperative condition index as a starting point, increasing the index range, and simultaneously acquiring abnormal items of human parameters in postoperative data in aortic dissection medical records corresponding to the postoperative condition index in the index range, wherein the abnormal items are marked as abnormal parameters;

counting all postoperative disease indexes in an index range, counting the number of different abnormal parameters, dividing the number of the abnormal parameters by the index number to obtain a parameter abnormal rate, increasing the index range, and outputting a range determining signal when the index number in the index range is greater than or equal to a first number threshold;

calculating the parameter anomaly rate, if the index number is greater than or equal to a second number threshold value and any parameter anomaly rate is equal to a first anomaly probability, continuing to increase the index range, monitoring whether the corresponding parameter anomaly rate is reduced, if the parameter anomaly rate is reduced, outputting a range determination signal, and if the parameter anomaly rate is increased, continuing to increase the index range;

if the range determining signal is output, stopping increasing the index range, and simultaneously taking the maximum value of the current index range as a starting point, and continuously analyzing the next index range until the index range comprises all postoperative disease indexes;

Outputting the parameter anomaly rates of different human parameters corresponding to different index ranges.

In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media having computer-usable program code embodied therein. The storage medium may be implemented by any type or combination of volatile or nonvolatile Memory devices, such as static random access Memory (Static Random Access Memory, SRAM), electrically erasable Programmable Read-Only Memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), erasable Programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), programmable Read-Only Memory (PROM), read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The above-described apparatus embodiments are merely illustrative, for example, the division of the units is merely a logical function division, and there may be other manners of division in actual implementation, and for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some communication interface, device or unit indirect coupling or communication connection, which may be in electrical, mechanical or other form.

Claims (3)

1. The postoperative condition deep learning method based on aortic dissection medical record data is characterized by comprising the following steps of:

constructing an aortic dissection medical record large database, and recording aortic dissection medical record information;

classifying the information types of the aortic dissection medical record information to obtain total information classification;

classifying the information types under the total information classification to obtain sub-information classification;

analyzing the total information classification and the sub information classification, judging the total classification indexes of different total information classifications, and calculating the postoperative disease index of the patient;

Deep learning is carried out based on the postoperative disease index of big data, and the parameter abnormality rate of each human parameter after operation of a patient is judged;

classifying the information types of the aortic dissection medical record information to obtain total information classification comprises the following sub-steps:

reading a hospital medical record database to obtain an aortic dissection medical record collection table;

classifying the information types of the aortic dissection medical record information based on the aortic dissection medical record collection table to obtain total information classification; the total information classification comprises current history data, past history data, imaging examination data, laboratory examination data, intraoperative data and postoperative data;

classifying the information types under the total information classification to obtain sub-information classification, wherein the sub-information classification comprises the following sub-steps:

acquiring information types of all information under the total information classification;

searching a data form in the information type, and classifying the data form to obtain sub-information classification; the sub-information classification comprises a digital type, a non-digital type and a hierarchical type;

the method comprises the steps of inducing the information types with the data form of numbers into numbers, inducing the information types with the data form of yes or no into a formula of whether or not, and inducing the information types with the data form of different grades into a grade formula;

Analyzing the total information classification and the sub-information classification, judging the total classification index of different total information classifications and calculating the postoperative condition index of the patient comprises the following sub-steps:

acquiring the first symptoms of a patient, establishing different deep learning branches aiming at different first symptoms, marking the deep learning branches as DLBn, wherein n is a positive integer greater than or equal to 1;

connecting a characteristic normal range database, and reading a characteristic normal range;

analyzing DLBn, traversing and reading total information classification in DLBn, and aiming at each total information classification, acquiring the number of sub information classifications under the directory of the total information classification, and marking the number as the classification information number;

acquiring data of medical records under the sub-information classification directory, and marking the data as medical record data;

acquiring medical record data, judging sub-information classification of the medical record data, and outputting a range detection signal if the medical record data is digital; if the signal is of the formula, outputting a detection signal; if the signal is the grade type, outputting a grade detection signal;

performing different processing on the medical record data according to different detection signals, and outputting analysis signals after the processing is completed;

performing different processing on the medical record data according to different detection signals, and outputting analysis signals after the processing is finished comprises the following sub-steps:

If the range detection signal is output, acquiring a characteristic normal range corresponding to the medical record data, searching whether the medical record data is in the characteristic normal range, and if so, outputting a normal range signal; if not, outputting an abnormal range signal;

if the detection signal is output, acquiring text characters of medical record data, and if the text characters are yes, outputting a characteristic abnormal signal; if the text characters are 'no', outputting a characteristic normal signal;

if the grade detection signal is output, reading whether the medical record data is zero grade, and if the grade of the medical record data is zero grade, outputting a grade normal signal; outputting a grade abnormality signal if the grade of the medical record data is not zero;

analyzing and calculating the analysis signals, and judging the total classification index of each total information classification; the analysis signals comprise a normal range signal, an abnormal range signal, a characteristic abnormal signal, a characteristic normal signal, a grade normal signal and a grade abnormal signal;

analyzing and calculating the analysis signals, and judging the total classification index of each total information classification comprises the following substeps:

setting a range classification index, namely initially zero, marking the range classification index as SCI, receiving an abnormal range signal, and adding one to the SCI if the abnormal range signal is received;

Setting a characteristic classification index, namely initially zero, marking the characteristic classification index as FCI, receiving a characteristic abnormal signal, and adding one to FCI if the characteristic abnormal signal is received;

setting a grade classification index, namely, initially setting the grade classification index as zero, marking the grade classification index as GCI, receiving a grade abnormality signal, acquiring the disease grade of medical record data if the grade abnormality signal is received, and increasing the GCI by 1+ (X-1) multiplied by 0.2 if the disease grade is X grade, wherein X is a constant and is a positive integer;

receiving all analysis signals, and obtaining SCI, FCI and GCI after receiving;

searching the quantity of the sub-information types of the total information classification, which are digital, whether or not and the class type, and marking the quantity as a range classification standard, a feature classification standard and a class classification standard;

calculating through a total classification index calculation formula to obtain a total classification index;

the total classification index calculation formula is configured as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein, AC is the total classification index, SCC is the range classification standard, FCC is the characteristic classification standard, GCC is the grade classification standard;

analyzing and calculating all total information classifications, and calculating total classification indexes to obtain postoperative disease indexes of patients;

analyzing and calculating all total information classifications, and calculating total classification indexes to obtain postoperative disease indexes of patients, wherein the method comprises the following substeps:

Analyzing and calculating total classification indexes of current medical history data, past medical history data, imaging examination data, laboratory examination data and intraoperative data, which are respectively named as total current medical history indexes, total past medical history indexes, total imaging indexes, total laboratory indexes and total intraoperative indexes, and are sequentially marked as AC1 to ACm, and m is a constant and a positive integer;

obtaining the minimum value and the maximum value in the ACs 1 to ACm, and marking the minimum value and the maximum value as an abnormal index and a normal index respectively;

setting an abnormality index, and setting the abnormality index as a first preset index if the abnormality index is a total current medical history index; if the abnormality index is the total past history index, setting the abnormality index as a second preset index; if the abnormality index is the total imaging index, setting the abnormality index as a third preset index; if the abnormality index is the total laboratory index, setting the abnormality index as a fourth preset index; if the abnormality index is the total operation index, setting the abnormality index as a fifth preset index;

setting a normal index, and if the normal index is a total current medical history index, setting the normal index as a first preset index; if the normal index is the total past history index, setting the normal index as a second preset index; if the normal index is the total imaging index, setting the normal index as a third preset index; if the normal index is the total laboratory index, setting the normal index as a fourth preset index; if the normal index is the total intraoperative index, setting the normal index as a fifth preset index;

Calculating by using a postoperative condition index calculation formula to obtain postoperative condition indexes of patients;

the postoperative condition index calculation formula is configured as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein Pte is postoperative condition index, T is abnormality index, and Q is normal index;

deep learning is carried out based on postoperative disease indexes of big data, and the parameter abnormality rate of each human parameter after operation of a patient is judged, wherein the method comprises the following sub-steps:

calculating postoperative disease indexes of all aortic dissection medical records based on the aortic dissection medical records large database, numbering the postoperative disease indexes, wherein the marks are Ps, s is a constant and is a positive integer;

establishing a postoperative condition index coordinate system by taking Ps as an X axis and postoperative condition indexes as a Y axis, and inputting all postoperative condition indexes into the postoperative condition index coordinate system; marking the number of postoperative disease indexes as index numbers;

setting an index range, taking a postoperative condition index as a starting point, increasing the index range, and simultaneously acquiring abnormal items of human parameters in postoperative data in aortic dissection medical records corresponding to the postoperative condition index in the index range, wherein the abnormal items are marked as abnormal parameters;

counting all postoperative disease indexes in an index range, counting the number of different abnormal parameters, dividing the number of the abnormal parameters by the index number to obtain a parameter abnormal rate, increasing the index range, and outputting a range determining signal when the index number in the index range is greater than or equal to a first number threshold;

Calculating the parameter anomaly rate, if the index number is greater than or equal to a second number threshold value and any parameter anomaly rate is equal to a first anomaly probability, continuing to increase the index range, monitoring whether the corresponding parameter anomaly rate is reduced, if the parameter anomaly rate is reduced, outputting a range determination signal, and if the parameter anomaly rate is increased, continuing to increase the index range;

if the range determining signal is output, stopping increasing the index range, and simultaneously taking the maximum value of the current index range as a starting point, and continuously analyzing the next index range until the index range comprises all postoperative disease indexes;

outputting the parameter anomaly rates of different human parameters corresponding to different index ranges.

2. The post-operative condition deep learning method based on aortic dissection medical record data according to claim 1, wherein constructing an aortic dissection medical record big database, recording aortic dissection medical record information comprises establishing data connection with a hospital medical record database, acquiring the aortic dissection medical record therein, and simultaneously acquiring a large number of aortic dissection medical records through internet big data and storing the large number of aortic dissection medical records in the aortic dissection medical record big database.

3. The system suitable for the postoperative condition deep learning method based on aortic dissection medical history data according to any one of claims 1 or 2, which is characterized by comprising a big data collection module, an information classification module, a postoperative condition analysis module and a postoperative condition deep learning module; the big data collection module, the information classification module and the postoperative condition deep learning module are respectively connected with the postoperative condition analysis module in a data mode;

The big data collection module is used for constructing an aortic dissection medical record big database and recording aortic dissection medical record information;

the information classification module is used for classifying the information types of the aortic dissection medical record information to obtain total information classification; classifying the information types under the total information classification to obtain sub-information classification;

the postoperative condition analysis module is used for analyzing the total information classification and the sub-information classification, judging the total classification indexes of different total information classifications and calculating the postoperative condition index of the patient;

the postoperative condition deep learning module is used for carrying out deep learning based on the postoperative condition index of big data and judging the parameter abnormality rate of each human parameter after the operation of a patient.