A method and system for predicting the probability distribution of pathological characteristics based on pathological data

Abstract

The present invention provides a method for predicting the probability distribution of pathological features based on pathological data, comprising: step S1, collecting the pathological data of a patient, performing dimensionality reduction processing on the pathological data to obtain corresponding feature data, and pre-configuring multiple feature labels; step S2, for each feature label, predicting the prior probability corresponding to the feature label according to the feature data and at least one statistical learning model, then obtaining the first posterior probability according to the prior probability processing, and predicting the second posterior probability corresponding to each feature label according to each feature data and at least one machine learning model; step S3, obtaining the probability mean according to the first posterior probability and the second posterior probability processing, and forming the patient's pathological feature probability distribution by all feature labels and the corresponding probability mean. The beneficial effect is to predict the patient's pathological feature probability distribution according to the pathological data, so as to assist doctors in classifying the patient's pathological data and improve the diagnostic efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of signal analysis, and in particular to a method and system for predicting the probability distribution of pathological features based on pathological data. Background Art

[0002] When conducting clinical diagnosis on patients, it is necessary to collect the patient's pathological data in various medical diagnostic methods, including collecting the patient's pathological signals and analyzing the possible diseases of the patient based on the pathological signals. Among various pathological signals, electrocardiogram data is mostly used in the judgment of heart diseases.

[0003] Among them, during the doctor's diagnosis of the patient, the doctor needs to continuously collect the surface electrocardiogram containing the patient's heart characteristics, and analyze the characteristics of the electrocardiogram data contained in the surface electrocardiogram by observing the surface electrocardiogram to make a diagnosis. When analyzing the patient's disease based on the surface electrocardiogram, due to the doctor's professional experience and subjective cognitive limitations, there is a defect of low accuracy of the diagnosis result, and it is difficult to grasp the subtle changes in the electrocardiogram data. Some of the patient's heart abnormalities, such as myocardial ischemia, may even be missed during the detection process. Therefore, processing, predicting and orderly classifying pathological data are of great help in assisting doctors to conduct early analysis and improve work efficiency. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method for predicting the probability distribution of pathological characteristics based on pathological data, comprising:

[0005] Step S1, collecting pathological data of a patient, performing dimensionality reduction processing on the pathological data to obtain corresponding feature data, and pre-configuring multiple feature labels;

[0006] Step S2, for each of the feature labels, predict the prior probability corresponding to the feature label according to the feature data and at least one statistical learning model obtained by pre-training, then obtain the first posterior probability according to the prior probability, and predict the second posterior probability corresponding to each of the feature labels according to each of the feature data and at least one machine learning model obtained by pre-training;

[0007] Step S3, obtaining the probability mean corresponding to each feature label according to the first posterior probability and the second posterior probability, and forming the patient's pathological feature probability distribution by all the feature labels and the corresponding probability means to assist the doctor in classifying the patient's pathological data.

[0008] Preferably, in step S1, Fourier basis transform, Haar base transform or wavelet basis transform is used to perform dimension reduction processing on the pathological data to obtain the characteristic data.

[0009] Preferably, in the step S1, before the dimensionality reduction processing is performed on the pathological data, a preprocessing process is also included, and the preprocessing process includes:

[0010] Step A1, performing format conversion on the pathological data to obtain pre-processed data, and judging whether all the pathological data have completed format conversion according to the format of the pre-processed data;

[0011] If yes, the pre-processed data is used as the pathological data to perform dimension reduction processing to obtain the feature data;

[0012] If not, a signal indicating that the pathological data has a format error is output.

[0013] Preferably, before executing step S2, a statistical learning model training process is also included, including:

[0014] Step B1, obtaining a plurality of first sample pathology data, and obtaining a characteristic probability distribution corresponding to a specific population corresponding to each characteristic label;

[0015] Step B2, obtaining corresponding first sample characteristic data according to the first sample pathological data;

[0016] Step B3, constructing the statistical learning model according to the feature probability distribution, and taking the first sample feature data as input and the first true prior probability corresponding to the first sample feature data as output, to train and obtain the statistical learning model.

[0017] Preferably, before executing step S2, a machine learning model training process is also included, including:

[0018] Step C1, obtaining a plurality of second sample pathology data, and obtaining a second true prior probability and a corresponding true posterior probability corresponding to a specific population corresponding to the second sample pathology data;

[0019] Step C2, obtaining corresponding second sample characteristic data according to the second sample pathological data;

[0020] Step C3, taking each of the second sample feature data as input and the true posterior probability corresponding to the second sample feature data as output, to train and obtain the machine learning model.

[0021] Preferably, before executing step S2, a model verification process is also included, including:

[0022] Step D1, constructing a validation set, wherein the validation set includes a plurality of validation pathology data and a true probability distribution corresponding to the validation pathology data;

[0023] Step D2, inputting the verification pathology data into the statistical learning model to predict the corresponding test prior probability, and processing the test prior probability to obtain the third posterior probability, and then inputting the verification pathology data into the machine learning model to obtain the corresponding fourth posterior probability;

[0024] Step D3, obtaining a corresponding test probability mean according to the third posterior probability and the fourth posterior probability;

[0025] Step D4, forming a test probability distribution by all the feature labels and the corresponding test probability means, obtaining a corresponding error value according to the test probability distribution and the true probability distribution, and judging the accuracy of the statistical learning model prediction and the machine learning model according to the error value and the pre-configured error range.

[0026] Preferably, a pathological feature probability distribution prediction system based on pathological data is also provided, applying any of the pathological feature probability distribution prediction methods described above, and the pathological feature probability distribution prediction system comprises:

[0027] A preprocessing module is used to collect pathological data of a patient, perform dimensionality reduction processing on the pathological data to obtain corresponding feature data, and pre-configure multiple feature labels;

[0028] A prediction module, connected to the preprocessing module, for predicting, for each of the feature labels, a priori probability corresponding to the feature label according to the feature data and at least one statistical learning model obtained by pre-training, then obtaining a first posterior probability according to the prior probability, and predicting a second posterior probability corresponding to each of the feature labels according to each of the feature data and at least one machine learning model obtained by pre-training;

[0029] An analysis module is connected to the prediction module and is used to obtain the probability mean corresponding to each of the feature labels according to the first posterior probability and the second posterior probability, and form the patient's pathological feature probability distribution by all the feature labels and the corresponding probability means, so as to assist the doctor in classifying the patient's pathological data.

[0030] Preferably, the preprocessing module comprises:

[0031] A data acquisition unit, used for acquiring the pathological data of the patient;

[0032] A dimensionality reduction processing unit is connected to the data acquisition unit, and is used to perform format conversion on the pathological data to obtain preprocessed data. When it is determined according to the format of the preprocessed data that all the pathological data have completed format conversion, the preprocessed data is used as the pathological data for dimensionality reduction processing to obtain the feature data.

[0033] Preferably, it further comprises a first model training module connected to the prediction module, wherein the first model training module comprises:

[0034] A first acquisition unit is used to obtain a plurality of first sample pathology data, and obtain a characteristic probability distribution corresponding to a specific population corresponding to each of the characteristic labels;

[0035] a first processing unit, connected to the first acquisition unit, configured to process the first sample pathological data to obtain corresponding first sample characteristic data;

[0036] A first training unit is connected to the first processing unit and is used to construct the statistical learning model according to the feature probability distribution, and to train the statistical learning model using the first sample feature data as input and the first true prior probability corresponding to the first sample feature data as output.

[0037] Preferably, it further comprises a second model training module connected to the prediction module, wherein the second model training module comprises:

[0038] A second acquisition unit is used to obtain a plurality of second sample pathology data, and obtain a second true prior probability and a corresponding true posterior probability corresponding to a specific population corresponding to the second sample pathology data;

[0039] a second processing unit, connected to the second acquisition unit, configured to process the first sample pathological data to obtain corresponding second sample characteristic data;

[0040] A second training unit is connected to the second acquisition unit, and is used to take each of the second sample feature data as input and the true posterior probability corresponding to the second sample feature data as output, so as to train and obtain the machine learning model.

[0041] The above technical solution has the following advantages or beneficial effects:

[0042] (1) According to the pathological data processing, the probability mean corresponding to each feature label is obtained, and the pathological feature probability distribution is formed by each feature label and the corresponding probability mean. The pathological feature probability distribution plays an auxiliary role in the doctor's diagnosis process of the patient, thereby improving the doctor's diagnosis efficiency;

[0043] (2) The probability mean is obtained by processing the first posterior probability and the second posterior probability. Compared with the probability distribution of pathological characteristics obtained by processing the first posterior probability or the second posterior probability alone, the prediction result obtained by combining the first posterior probability and the second posterior probability is more accurate.

[0044] (3) By verifying the statistical learning model and the machine learning model, the corresponding error value is obtained by comparing the test probability distribution with the true probability distribution corresponding to the verification set, and the accuracy of the prediction results of the statistical learning model and the machine learning model is judged according to the error value, so as to facilitate the subsequent optimization of the statistical learning model and the machine learning model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 The flowchart of the method for predicting the probability distribution of pathological characteristics in a preferred embodiment of the present invention is as follows;

[0046] Figure 2 The specific flow chart of the preprocessing process in the pathological feature probability distribution prediction method in the preferred embodiment of the present invention;

[0047] Figure 3 A specific flow chart of the statistical learning model training process in the pathological feature probability distribution prediction method in a preferred embodiment of the present invention;

[0048] Figure 4 A specific flow chart of the machine learning model training process in the pathological feature probability distribution prediction method in a preferred embodiment of the present invention;

[0049] Figure 5 The specific flow chart of the model verification process in the pathological feature probability distribution prediction method in the preferred embodiment of the present invention;

[0050] Figure 6 This is a control principle diagram of a pathological feature probability distribution prediction system in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The present invention is not limited to this embodiment, and other embodiments may also fall within the scope of the present invention as long as they conform to the gist of the present invention.

[0052] In a preferred embodiment of the present invention, based on the above problems existing in the prior art, a method for predicting the probability distribution of pathological features based on pathological data is provided. Figure 1 As shown, including:

[0053] Step S1, collecting pathological data of a patient, performing dimensionality reduction processing on the pathological data to obtain corresponding feature data, and pre-configuring multiple feature labels;

[0054] Step S2, for each feature label, predict the prior probability corresponding to the feature label according to the feature data and at least one statistical learning model obtained by pre-training, then obtain the first posterior probability according to the prior probability processing, and predict the second posterior probability corresponding to each feature label according to each feature data and at least one machine learning model obtained by pre-training;

[0055] Step S3, obtaining the probability mean corresponding to each feature label according to the first posterior probability and the second posterior probability, and forming the patient's pathological feature probability distribution by all feature labels and the corresponding probability means to assist doctors in classifying the patient's pathological data.

[0056] Specifically, in this embodiment, the pathological data is the electrocardiogram data of the patient, then:

[0057] In step S1, the electrocardiogram data of the patient is collected, the electrocardiogram data is subjected to dimensionality reduction processing to obtain corresponding feature data, and feature labels corresponding to various heart diseases are pre-configured. The doctor diagnoses the patient's heart disease based on the probability distribution of the pathological features.

[0058] The probability distribution of pathological features includes all feature labels and their corresponding probability means, which provides doctors with a more comprehensive basis for diagnosis and reduces the situation where doctors fail to diagnose other diseases of patients because they fail to find minor abnormalities in the ECG data.

[0059] In a preferred embodiment of the present invention, in step S1, Fourier basis transform, Haar base transform or wavelet basis transform is used to perform dimension reduction processing on the pathological data to obtain characteristic data.

[0060] Specifically, in this embodiment, before predicting the possible heart disease of the patient based on the electrocardiogram data, the patient's electrocardiogram data needs to be obtained from the hospital's internal system, and in the hospital's internal system, the electrocardiogram data is often stored in the XML format. Therefore, after obtaining the electrocardiogram data, the electrocardiogram data needs to be converted to a different format for subsequent analysis.

[0061] In a preferred embodiment of the present invention, in step S1, before the dimensionality reduction processing of the pathological data is performed, a preprocessing process is also included, such as Figure 2 As shown, the preprocessing process includes:

[0062] Step A1, collecting the patient's pathological data;

[0063] Step A2, converting the format of the pathological data to obtain preprocessed data, and judging whether all the pathological data have completed the format conversion according to the format of the preprocessed data;

[0064] If so, the preprocessed data is used as pathological data for dimensionality reduction to obtain feature data;

[0065] If not, a signal indicating that the pathological data has a format error is output.

[0066] Specifically, in this embodiment, before the dimensionality reduction process is performed on the electrocardiogram data:

[0067] First, the data between two data feature points contained in the electrocardiogram data in XML format is extracted, and after being sorted into 12-dimensional data, it is stored as electrocardiogram data in CSV or TXT format.

[0068] Then, it is determined whether the formats of the electrocardiogram data are successfully converted. When the format conversion of the electrocardiogram data is unsuccessful, an error signal is output to prompt the operator that there is an error in the format of the electrocardiogram data and the prediction cannot be continued.

[0069] After all ECG data are successfully converted, the ECG data are normalized and the ECG data need to be decomposed into singular values ​​to reduce the influence of singular points in the ECG data on the prediction results and achieve the purpose of noise reduction.

[0070] In a preferred embodiment of the present invention, before executing step S2, a statistical learning model training process is also included, such as Figure 3 As shown, including:

[0071] Step B1, obtaining a plurality of first sample pathological data, and obtaining a characteristic probability distribution corresponding to a specific population corresponding to each characteristic label;

[0072] Step B2, obtaining corresponding first sample characteristic data according to the first sample pathological data;

[0073] Step B3, constructing a statistical learning model according to the feature probability distribution, and taking the first sample feature data as input and the first true prior probability corresponding to the first sample feature data as output, to train and obtain the statistical learning model.

[0074] Specifically, in this embodiment, the populations corresponding to the first sample pathological data belong to specific populations, and the first sample pathological data are read in respectively according to the type of specific population to which the population belongs. The first sample pathological data corresponding to each specific population is saved in the corresponding data file respectively. For each data file, the first sample pathological data in the data file is read with a 12-dimensional lead time series with a data dimension of 12 or a single lead time series with a data dimension of 1, and the electrocardiogram data read is a one-dimensional time data sequence. Subsequently, the electrocardiogram data read is normalized.

[0075] The normalization process includes: when there are singular points in the electrocardiogram data, the second-order difference of the electrocardiogram data corresponding to the sequence number between the data length 0.1 and the data length 0.9 in the one-dimensional time data sequence is calculated to obtain the minimum value, and the minimum value is multiplied by the preset threshold to obtain the product value, and all the segments of the electrocardiogram data whose second-order difference is less than the product value are taken out. Subsequently, according to the median corresponding to the ordinate of each segment, the horizontal coordinate of the segment corresponding to the median is screened. Then, the screened segments are cut to obtain a normalized time data sequence with a data length of 0.1, and finally the normalized time data sequence is subjected to dimensionality reduction processing, and the normalized time data sequence after dimensionality reduction processing is output as the first sample feature data.

[0076] In a preferred embodiment of the present invention, before executing step S2, a machine learning model training process is also included, such as Figure 4 As shown, including:

[0077] Step C1, obtaining a plurality of second sample pathology data, and obtaining a second true prior probability and a corresponding true posterior probability corresponding to a specific population corresponding to the second sample pathology data;

[0078] Step C2, obtaining corresponding second sample characteristic data according to the second sample pathological data;

[0079] In step C3, each second sample feature data is used as input, and the true posterior probability corresponding to the second sample feature data is used as output to train a machine learning model.

[0080] Specifically, in this embodiment, the machine learning model includes but is not limited to decision tree-based XGBoost, LightGBM, random forest, etc. and neural network-based LSTM, CNN and other algorithms.

[0081] When predicting the probability distribution of pathological features, for each feature label, the kernel density estimation method is used for the first sample feature data after dimensionality reduction to obtain the prior probability. The calculation formula is as follows:

[0082]

[0083] Where h is the standard deviation of the pre-configured kernel function, K is used to represent the Gaussian function, and x i =1,2,...,n are one-dimensional components corresponding to each first sample feature data, and each feature label and the corresponding prior probability form a prior probability distribution.

[0084] Then, the first posterior probability and the second posterior probability are calculated respectively.

[0085] Among them, for the statistical learning model, the calculation formula of the first posterior probability is as follows:

[0086]

[0087] Among them, p(y|x obs ) is the posterior probability distribution, x obs is the first characteristic sample data, p(x obs |y) is the prior probability distribution, and p(z) is the characteristic probability distribution.

[0088] For machine learning models, the calculation formula for the second posterior probability is as follows:

[0089]

[0090] Among them, p(y|x obs ) is the second posterior probability, x obs is the first feature sample data, f y (x obs ) is the prediction function corresponding to the feature label y obtained by pre-training in the machine learning model, f z (x obs ) is the prediction function corresponding to the specific population z pre-trained in the machine learning model.

[0091] In a preferred embodiment of the present invention, before executing step S2, a model verification process is also included, such as Figure 5 As shown, including:

[0092] Step D1, constructing a validation set, wherein the validation set includes a plurality of validation pathology data and a true probability distribution corresponding to the validation pathology data;

[0093] Step D2, input the verification pathology data into the statistical learning model to predict the corresponding test prior probability, and obtain the third posterior probability according to the test prior probability, and then input the verification pathology data into the machine learning model to obtain the corresponding fourth posterior probability;

[0094] Step D3, obtaining a corresponding test probability mean value according to the third posterior probability and the fourth posterior probability;

[0095] Step D4, forming a test probability distribution by all feature labels and the corresponding test probability means, obtaining the corresponding error value according to the test probability distribution and the true probability distribution, and judging the accuracy of the statistical learning model prediction and the machine learning model according to the error value and the pre-configured error range.

[0096] Specifically, in this embodiment, the process of determining the accuracy of the statistical learning model and the machine learning model includes:

[0097] According to the true probability distribution and the test probability distribution, an evaluation score representing the accuracy of the test probability value is obtained. The calculation formula of the evaluation score is as follows:

[0098]

[0099] Among them, BS is the evaluation score, I is the number of verification pathology data included in the test set, represents the test probability mean of the i-th verification pathology data with feature label y, and the error value is p(y|i)-y i , the error range is [0, 1];

[0100] When the feature label y matches the specific population corresponding to the i-th validation pathology data, y i =1;

[0101] When the feature label y does not match the specific population corresponding to the i-th validation pathology data, y i =0.

[0102] The closer the BS value is to 0, the higher the accuracy is; the closer the BS value is to 1, the lower the accuracy is.

[0103] In another embodiment, the precision, recall and accuracy are calculated respectively, and processed to obtain a comprehensive index, and the accuracy of the statistical learning model and the machine learning model is judged according to the value of the comprehensive index.

[0104] The calculation formula of precision is as follows:

[0105]

[0106] Among them, TP i Indicates that the i-th verification pathological data belongs to the feature label y and actually belongs to the corresponding heart disease, FP i It means that the i-th verification pathology data does not belong to the feature label y, but actually belongs to the corresponding heart disease.

[0107] The calculation formula of recall rate Recall is as follows:

[0108]

[0109] Among them, FN i It means that the i-th verification pathological data does not belong to the feature label y, and actually does not belong to the corresponding heart disease;

[0110] The calculation formula for accuracy is as follows:

[0111]

[0112] Among them, TN iIt means that the i-th test data belongs to the feature label y, but actually does not belong to the corresponding heart disease.

[0113] Comprehensive Index F 1 The calculation formula is as follows:

[0114]

[0115] In a preferred embodiment of the present invention, a pathological feature probability distribution prediction system based on pathological data is also provided, which uses any of the above pathological feature probability distribution prediction methods, such as Figure 6 As shown, the pathological feature probability distribution prediction system includes:

[0116] Preprocessing module 1, used for collecting pathological data of a patient, performing dimension reduction processing on the pathological data to obtain corresponding feature data, and pre-configuring multiple feature labels;

[0117] Prediction module 2, connected to preprocessing module 1, is used to predict, for each feature label, the prior probability corresponding to the feature label according to the feature data and at least one statistical learning model obtained by pre-training, then obtain the first posterior probability according to the prior probability processing, and predict the second posterior probability corresponding to each feature label according to each feature data and at least one machine learning model obtained by pre-training;

[0118] The analysis module 3 is connected to the prediction module 2 and is used to obtain the probability mean corresponding to each feature label according to the corresponding first posterior probability and the second posterior probability, and form the patient's pathological feature probability distribution by all feature labels and the corresponding probability means to assist doctors in classifying the patient's pathological data.

[0119] In a preferred embodiment of the present invention, the preprocessing module 1 comprises:

[0120] A data collection unit 11 is used to collect pathological data of the patient;

[0121] The dimension reduction processing unit 12 is connected to the data acquisition unit and is used to perform format conversion on the pathological data to obtain pre-processed data. When it is determined according to the format of the pre-processed data that all the pathological data have completed the format conversion, the pre-processed data is used as the pathological data for dimension reduction processing to obtain feature data.

[0122] In a preferred embodiment of the present invention, a first model training module 4 is further included, which is connected to the prediction module 2. The first model training module 4 includes:

[0123] The first acquisition unit 41 is used to acquire a plurality of first sample pathological data and acquire a characteristic probability distribution corresponding to a specific population corresponding to each characteristic label;

[0124] A first processing unit 42, connected to the first acquisition unit 41, is used to process the first sample pathological data to obtain corresponding first sample characteristic data;

[0125] The first training unit 43 is connected to the first processing unit 42, and is used to build a statistical learning model according to the feature probability distribution, and use the first sample feature data as input and the first true prior probability corresponding to the first sample feature data as output to train the statistical learning model.

[0126] In a preferred embodiment of the present invention, a second model training module 5 is further included, connected to the prediction module 2, and the second model training module 5 includes:

[0127] The second acquisition unit 51 is used to obtain a plurality of second sample pathology data, and obtain a second true prior probability and a corresponding true posterior probability corresponding to a specific population corresponding to the second sample pathology data;

[0128] The second processing unit 52 is connected to the second acquisition unit 51 and is used to process the first sample pathological data to obtain corresponding second sample characteristic data;

[0129] The second training unit 53 is connected to the second processing unit 52, and is used to take each second sample feature data as input and the true posterior probability corresponding to the second sample feature data as output, so as to train and obtain a machine learning model.

[0130] Specifically, in this embodiment, the first acquisition unit 41 and the second acquisition unit 51 may be integrated to acquire the third sample pathology data, wherein the third sample pathology data is stored as the first sample pathology data and the second sample pathology data.

[0131] The first processing unit 42 and the second processing unit 52 may also be integrated to process the third sample pathology data to obtain third sample characteristic data, and output the third sample characteristic data as the first sample characteristic data and the second sample characteristic data.

[0132] In summary, by extracting segments, normalizing data, and reducing the dimension of the ECG data to be predicted, the ECG data is converted into normalized time data sequences for subsequent prediction of the probability distribution of pathological features. In addition, for each feature label, the first posterior probability and the second posterior probability are predicted and the average value is calculated to obtain the probability mean, and the probability mean is used as the probability value of the corresponding feature label, which reduces the prediction error and improves the accuracy of the probability distribution of pathological features.

[0133] The above description is only a preferred embodiment of the present invention, and does not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.

Claims

1. A method for predicting the probability distribution of pathological features based on pathological data, characterized in that: include: Step S1, collecting pathological data of a patient, performing dimensionality reduction processing on the pathological data to obtain corresponding feature data, and pre-configuring multiple feature labels; Step S2, for each of the feature labels, predict the prior probability corresponding to the feature label according to the feature data and at least one statistical learning model obtained by pre-training, then obtain the first posterior probability according to the prior probability, and predict the second posterior probability corresponding to each of the feature labels according to each of the feature data and at least one machine learning model obtained by pre-training; Step S3, obtaining the probability mean corresponding to each feature label according to the first posterior probability and the second posterior probability, and forming the patient's pathological feature probability distribution by all the feature labels and the corresponding probability means to assist the doctor in classifying the patient's pathological data.

2. The method for predicting the probability distribution of pathological features according to claim 1, characterized in that: In the step S1, Fourier basis transform, Haar base transform or wavelet basis transform is used to perform dimension reduction processing on the pathological data to obtain the characteristic data.

3. The method for predicting the probability distribution of pathological features according to claim 1, characterized in that: In the step S1, before the dimension reduction process is performed on the pathological data, a preprocessing process is also included, and the preprocessing process includes: Step A1, collecting the pathological data of the patient; Step A2, performing format conversion on the pathological data to obtain pre-processed data, and judging whether all the pathological data have completed format conversion according to the format of the pre-processed data; If yes, the pre-processed data is used as the pathological data to perform dimension reduction processing to obtain the feature data; If not, a signal indicating that the pathological data has a format error is output.

4. The method for predicting the probability distribution of pathological features according to claim 1, characterized in that: Before executing step S2, a statistical learning model training process is also included, including: Step B1, obtaining a plurality of first sample pathology data, and obtaining a characteristic probability distribution corresponding to a specific population corresponding to each characteristic label; Step B2, obtaining corresponding first sample characteristic data according to the first sample pathological data; Step B3, constructing the statistical learning model according to the feature probability distribution, and taking the first sample feature data as input and the first true prior probability corresponding to the first sample feature data as output, to train and obtain the statistical learning model.

5. The method for predicting the probability distribution of pathological features according to claim 1, characterized in that: Before executing step S2, a machine learning model training process is also included, including: Step C1, obtaining a plurality of second sample pathology data, and obtaining a second true prior probability and a corresponding true posterior probability corresponding to a specific population corresponding to the second sample pathology data; Step C2, obtaining corresponding second sample characteristic data according to the second sample pathological data; Step C3, taking each of the second sample feature data as input and the true posterior probability corresponding to the second sample feature data as output, to train and obtain the machine learning model.

6. The method for predicting the probability distribution of pathological features according to claim 1, characterized in that: Before executing step S2, a model verification process is also included, including: Step D1, constructing a validation set, wherein the validation set includes a plurality of validation pathology data and a true probability distribution corresponding to the validation pathology data; Step D2, inputting the verification pathology data into the statistical learning model to predict the corresponding test prior probability, and processing the test prior probability to obtain the third posterior probability, and then inputting the verification pathology data into the machine learning model to obtain the corresponding fourth posterior probability; Step D3, obtaining a corresponding test probability mean according to the third posterior probability and the fourth posterior probability; Step D4, forming a test probability distribution by all the feature labels and the corresponding test probability means, obtaining a corresponding error value according to the test probability distribution and the true probability distribution, and judging the accuracy of the statistical learning model prediction and the machine learning model according to the error value and the pre-configured error range.

7. A pathological feature probability distribution prediction system based on pathological data, characterized in that: The pathological feature probability distribution prediction method according to any one of claims 1 to 6 is applied, wherein the pathological feature probability distribution prediction system comprises: A preprocessing module is used to collect pathological data of a patient, perform dimensionality reduction processing on the pathological data to obtain corresponding feature data, and pre-configure multiple feature labels; A prediction module, connected to the preprocessing module, for predicting, for each of the feature labels, a priori probability corresponding to the feature label according to the feature data and at least one statistical learning model obtained by pre-training, then obtaining a first posterior probability according to the prior probability, and predicting a second posterior probability corresponding to each of the feature labels according to each of the feature data and at least one machine learning model obtained by pre-training; An analysis module is connected to the prediction module and is used to obtain the probability mean corresponding to each of the feature labels according to the first posterior probability and the second posterior probability, and form the patient's pathological feature probability distribution by all the feature labels and the corresponding probability means, so as to assist the doctor in classifying the patient's pathological data.

8. The pathological feature probability distribution prediction system according to claim 7, characterized in that: The preprocessing module comprises: A data acquisition unit, used for acquiring the pathological data of the patient; A dimensionality reduction processing unit is connected to the data acquisition unit, and is used to perform format conversion on the pathological data to obtain preprocessed data. When it is determined according to the format of the preprocessed data that all the pathological data have completed format conversion, the preprocessed data is used as the pathological data for dimensionality reduction processing to obtain the feature data.

9. The pathological feature probability distribution prediction system according to claim 7, characterized in that: It also includes a first model training module connected to the prediction module, and the first model training module includes: A first acquisition unit is used to obtain a plurality of first sample pathology data, and obtain a characteristic probability distribution corresponding to a specific population corresponding to each of the characteristic labels; a first processing unit, connected to the first acquisition unit, configured to process the first sample pathological data to obtain corresponding first sample characteristic data; A first training unit is connected to the first processing unit and is used to construct the statistical learning model according to the feature probability distribution, and to train the statistical learning model using the first sample feature data as input and the first true prior probability corresponding to the first sample feature data as output.

10. The pathological feature probability distribution prediction system according to claim 9, characterized in that: It also includes a second model training module connected to the prediction module, and the second model training module includes: A second acquisition unit is used to obtain a plurality of second sample pathology data, and obtain a second true prior probability and a corresponding true posterior probability corresponding to a specific population corresponding to the second sample pathology data; a second processing unit, connected to the second acquisition unit, configured to process the first sample pathological data to obtain corresponding second sample characteristic data; A second training unit is connected to the second acquisition unit, and is used to take each of the second sample feature data as input and the true posterior probability corresponding to the second sample feature data as output, so as to train and obtain the machine learning model.

Images