Intelligent data analysis and processing system and method based on deep data mining

Abstract

The invention relates to the field of intelligent data analysis, and particularly discloses an intelligent data analysis processing system and method based on deep data mining, which uses an artificial intelligence technology based on the deep learning field to perform feature extraction and coding on original data of a medical record to be detected and living habit records of a patient. And a classification result of whether the patient is a high-risk group of diabetic complications is obtained. Therefore, by intelligently judging whether the patient is a high-risk group of diabetic complications or not, illness state delay can be effectively avoided, and the best opportunity for treating diabetes mellitus is missed.

Description

Technical Field

[0001] The present application relates to the field of intelligent data analysis, and more specifically to a data intelligent analysis processing system and method based on deep data mining. Background Art

[0002] Diabetes is a chronic metabolic disease characterized by persistently high blood sugar levels. Common symptoms of diabetes include polyuria, polydipsia, polyphagia, weight loss, fatigue, blurred vision, and susceptibility to infection. If diabetes is not effectively controlled, it may lead to complications such as cardiovascular disease, kidney disease, neuropathy, eye disease, etc., which are extremely harmful to human health. At present, patients usually go to the hospital for blood sugar tests when they have more serious symptoms and then find out that they have diabetes. This method of examination may delay the disease and miss the best time to treat diabetes.

[0003] Therefore, a data intelligent analysis and processing system and method for diabetes prediction is expected. Summary of the invention

[0004] In order to solve the above technical problems, this application is proposed. The embodiment of the present application provides a data intelligent analysis and processing system and method based on deep data mining, which uses artificial intelligence technology based on the field of deep learning to extract and encode the original data of the medical records to be tested and the patient's living habit records to obtain the classification result of whether the patient is a high-risk group for diabetic complications. In this way, by intelligently judging whether the patient is a high-risk group for diabetic complications, it can effectively avoid delaying the disease and missing the best time to treat diabetes.

[0005] According to one aspect of the present application, a data intelligent analysis and processing system based on deep data mining is provided, which includes:

[0006] Patient data acquisition module, used to obtain the original data of the medical records to be tested and the patient's living habit records;

[0007] A patient data processing module, used for processing the raw data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector;

[0008] The diagnosis result generating module is used to pass the multi-scale patient comprehensive feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is a high-risk group for diabetic complications.

[0009] According to another aspect of the present application, a data intelligent analysis and processing method based on deep data mining is also provided, which includes:

[0010] Obtain the original data of the medical records to be tested and the patient's living habits records;

[0011] Processing the raw data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector;

[0012] The multi-scale patient comprehensive feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is a high-risk group for diabetic complications.

[0013] In summary, the data intelligent analysis and processing system and method based on deep data mining provided by the present application uses artificial intelligence technology based on the field of deep learning to extract features and encode the original data of the medical records to be tested and the patient's living habit records, so as to obtain the classification result of whether the patient is a high-risk group for diabetic complications. In this way, by intelligently judging whether the patient is a high-risk group for diabetic complications, it is possible to effectively avoid delaying the disease and thus missing the best time to treat diabetes. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art description are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings, the same reference numerals generally represent the same components or steps.

[0015] Figure 1 It is a block diagram of a data intelligent analysis and processing system based on deep data mining according to an embodiment of the present application.

[0016] Figure 2 It is a block diagram of a patient data processing module in a data intelligent analysis and processing system based on deep data mining according to an embodiment of the present application.

[0017] Figure 3 It is a flowchart of a data intelligent analysis and processing method based on deep data mining according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] Below, in order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] Figure 1 FIG. 1 is a block diagram of a data intelligent analysis and processing system based on deep data mining according to an embodiment of the present application. Figure 1 As shown, according to the data intelligent analysis and processing system 100 based on deep data mining according to the embodiment of the present application, it includes: a patient data acquisition module 110, which is used to acquire the original data of the medical records to be detected and the patient's living habit records; a patient data processing module 120, which is used to process the original data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector; and a diagnosis result generation module 130, which is used to pass the multi-scale patient comprehensive feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is at high risk of diabetic complications.

[0020] In the above-mentioned data intelligent analysis and processing system 100 based on deep data mining, the patient data acquisition module 110 is used to obtain the original data of the medical records to be tested and the patient's living habit records. As mentioned in the above-mentioned background technology, at present, patients generally go to the hospital for blood sugar and other tests when they have more serious symptoms to find out that they have diabetes. This examination method may delay the disease and miss the best time to treat diabetes. Therefore, a data intelligent analysis and processing system and method for diabetes prediction are expected.

[0021] Specifically, in the technical solution of the present application, it is considered that the medical record is a document in which the doctor records the patient's diagnosis and treatment process, and is used to record the patient's medical history, symptoms, physical examination results, diagnosis, treatment plan, follow-up and other information. Medical records are an important basis for doctors to make diagnosis and treatment, and are also important evidence for medical quality assessment and medical dispute resolution. Traditional medical records are often written by doctors. However, paper medical records have the disadvantages of inconvenient storage, difficult to find, easy to lose and difficult to share. With the development of computer technology, more and more medical institutions have begun to adopt electronic medical record systems to replace traditional paper medical records to improve the management efficiency and security of medical information. At the same time, with the increasing popularity of electronic medical records, it has laid the foundation for the development of data mining in the medical field. At the same time, with the increasing pursuit of accurate diagnostic results, auxiliary analysis is carried out by obtaining multi-source data.

[0022] In response to the above technical problems, a data intelligent analysis and processing system based on deep data mining is proposed. It uses artificial intelligence technology based on the field of deep learning to extract and encode the original data of the medical records to be tested and the patient's living habits records, so as to obtain the classification result of whether the patient is a high-risk group for diabetic complications. In this way, by intelligently judging whether the patient is a high-risk group for diabetic complications, it can effectively avoid delaying the disease and missing the best time to treat diabetes.

[0023] At present, deep learning and neural networks have been widely used in computer vision, natural language processing, speech signal processing and other fields. In addition, deep learning and neural networks have also shown a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields.

[0024] In recent years, the development of deep learning and neural networks has provided new solutions and plans for data intelligent analysis and processing systems based on deep data mining.

[0025] Specifically, first, obtain the original data of the medical records to be tested and the patient's living habit records. The electronic medical records include: the patient's personal information, physiological indicators, examination history, medical records, medical history and other data; the patient's living habits, including: eating habits, exercise habits. By obtaining and comprehensively considering medical record data and living habit records, more comprehensive patient information can be obtained. This helps to identify potential risk factors and disease development trends, and provide more accurate assessments and predictions. For example, some patients may not have obvious medical record indicators, but their bad living habits may increase the risk of diabetic complications.

[0026] In the above-mentioned data intelligent analysis and processing system 100 based on deep data mining, the patient data processing module 120 is used to process the original data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector. In order to extract and encode important information related to diabetic complications, and represent it as a comprehensive feature vector, the original data of the medical records to be detected and the patient's living habit records are processed to obtain a multi-scale patient comprehensive feature vector. Considering that the original data and living habit records may contain a large amount of information, but not all information is critical for determining whether the patient is a high-risk group for diabetic complications. By processing and analyzing these data, important features related to diabetic complications can be extracted, redundancy and noise can be removed, thereby reducing the dimension and complexity of the features. By processing and optimizing features, the original data and living habit records can be converted into more expressive and discriminative feature vectors. Such feature vectors can better capture the patient's characteristic patterns and associated information, and improve the performance and accuracy of the classifier. The occurrence and progression of diabetic complications is a complex process affected by many factors. By processing features of different scales, the patient's medical data and living habit records can be comprehensively considered to obtain more comprehensive and multi-angle information. This will help improve the ability to identify people at high risk of diabetic complications.

[0027] Specifically, Figure 2 FIG. 1 is a block diagram of a patient data processing module in a data intelligent analysis and processing system based on deep data mining according to an embodiment of the present application. Figure 2As shown, the patient data processing module 120 includes: a patient original data processing unit 121, used to perform semantic feature extraction on the original data of the medical record to be detected and the patient's living habit record to obtain a medical record context semantic feature vector and a habit context semantic feature vector; and a patient feature extraction unit 122, used to perform feature extraction and encoding on the medical record context semantic feature vector and the habit context semantic feature vector to obtain the multi-scale patient comprehensive feature vector.

[0028] Specifically, in an embodiment of the present application, the patient raw data processing unit is used to: respectively pass the raw data of the medical record to be detected and the patient's living habit records through a converter-based patient medical record semantic encoder to obtain the medical record context semantic feature vector and the habit context semantic feature vector.

[0029] Specifically, in an embodiment of the present application, the patient medical record semantic encoder is a converter-based context encoder including an embedding layer.

[0030] Specifically, in order to facilitate subsequent fusion and processing, the original data of the medical record to be detected and the patient's living habit record are respectively passed through a converter-based patient medical record semantic encoder to obtain the medical record context semantic feature vector and the habit context semantic feature vector. The patient medical record semantic encoder is a converter-based context encoder including an embedding layer. Considering that the original data and living habit records usually contain a large amount of information, which contains rich semantic and context information. By using a converter-based context encoder, useful features of these data can be extracted and converted into more representative and expressible feature vectors. These feature vectors can capture important information in medical record data and living habit records, which is helpful for subsequent processing and analysis. The original data and living habit records often have different data types and dimensions. By using an embedding layer, these different types of data can be converted into feature vectors of the same dimension. This dimensional unification is helpful for subsequent data fusion and processing, so that different types of data can be effectively combined and compared. The converter-based context encoder can learn the semantic representation of the data. This means that the encoded feature vector can better capture the association and semantic information between the data. For example, in medical record data, certain diagnoses and symptoms may be associated with the risk of diabetic complications, while in lifestyle records, certain bad habits may be associated with the development of the disease. The feature vectors learned by the encoder can better represent these associations and semantic information, improving the accuracy and effectiveness of subsequent processing.

[0031] Specifically, in an embodiment of the present application, the patient feature extraction unit includes: a patient information fusion subunit, used to perform kernel association fusion based on dimensionality degradation projection on the medical record context semantic feature vector and the habit context semantic feature vector to obtain a patient comprehensive feature vector; a patient information mixed convolution coding subunit, used to perform mixed feature extraction on the patient comprehensive feature vector to obtain the multi-scale patient comprehensive feature vector.

[0032] Specifically, in order to make full use of different types of information, a more comprehensive and accurate patient feature representation is provided by fusing the medical record context semantic feature vector and the habit context semantic feature vector into a comprehensive patient feature vector. The medical record context semantic feature vector and the habit context semantic feature vector represent the patient's medical information and living habit information, respectively. These information are important in assessing the patient's risk of diabetic complications. By fusing them together, the information from both medical and living aspects can be comprehensively considered to obtain a more comprehensive and accurate patient feature representation. It is also considered that the medical record context semantic feature vector and the habit context semantic feature vector often contain different types of information. The medical record context semantic feature vector may contain medical indicators such as diagnosis results and treatment records, while the habit context semantic feature vector may contain living habit information such as diet and exercise. By fusing these two feature vectors, different types of information can complement each other. For example, some patients may not have obvious medical record indicators, but their bad living habits may increase the risk of diabetic complications. The fused comprehensive feature matrix can better reflect this comprehensive effect. By fusing the medical record context semantic feature vector and the habit context semantic feature vector, richer and more comprehensive feature information can be provided. This helps to improve the performance and accuracy of the classification algorithm. For example, the fused feature matrix can be used to train and predict machine learning or deep learning algorithms to better identify people at high risk of diabetic complications.

[0033] In particular, considering that the medical record context semantic feature vectors and habit context semantic feature vectors may be distributed in different feature spaces, the performance degradation when directly applying the source domain model to the target domain is because medical record data and life habit records essentially represent two different types of input information. Medical record data usually contains professional medical information such as medical diagnosis, treatment history, and laboratory test results, while life habit records cover lifestyle information such as personal daily activity patterns, dietary preferences, and exercise frequency. These two types of data not only differ significantly in content, but also in the semantic context and time scale they express. After being processed by a transformer-based context encoder containing an embedding layer, although their respective semantic feature vectors can be extracted, these feature vectors still retain the characteristics of the original data. Specifically, the medical record context semantic feature vectors tend to capture static or semi-static features related to the disease, such as long-term health conditions and development trends; while the habit context semantic feature vectors reflect more individual behavior patterns and their dynamic impact on health. Therefore, even if both are encoded as high-dimensional feature vectors, they still exist in their own independent and different feature spaces, because each feature vector carries unique information and statistical properties of a specific domain. When trying to directly transfer knowledge from one domain (for example, a model trained only on medical record data) to another domain (i.e., a new environment that integrates medical records and living habits), it is difficult for the model to directly and effectively transfer its learned knowledge due to the feature space differences between the different modalities mentioned above. Based on this, in the technical solution of the present application, the medical record context semantic feature vector and the habit context semantic feature vector are fused by kernel association based on dimensionality degradation projection to obtain a comprehensive patient feature vector.

[0034] Specifically, in an embodiment of the present application, the medical record context semantic feature vector and the habit context semantic feature vector are subjected to kernel association fusion based on dimensionality degradation projection to obtain a comprehensive feature vector of the patient, including: mapping the medical record context semantic feature vector and the habit context semantic feature vector to the intrinsic decomposition space to obtain a medical record context semantic intrinsic feature vector and a habit context semantic intrinsic feature vector; performing inter-domain feature calibration on the medical record context semantic intrinsic feature vector and the habit context semantic intrinsic feature vector to obtain a medical record context semantic modulated intrinsic feature vector and a habit context semantic modulated intrinsic feature vector; constructing a medical record-habit context fine-grained semantic information field between the medical record context semantic modulated intrinsic feature vector and the habit context semantic modulated intrinsic feature vector; and fusing the medical record context semantic modulated intrinsic feature vector and the habit context semantic modulated intrinsic feature vector based on the medical record-habit context fine-grained semantic information field to obtain the comprehensive feature vector of the patient.

[0035] More specifically, in an embodiment of the present application, a medical record-habit context fine-grained semantic information field is constructed between the intrinsic feature vector after the medical record context semantics modulation and the intrinsic feature vector after the habit context semantics modulation, including: multiplying the transposed vector of the intrinsic feature vector after the medical record context semantics modulation and the intrinsic feature vector after the habit context semantics modulation, and then dividing by the scale of the intrinsic feature vector after the habit context semantics modulation to obtain a medical record-habit context fine-grained interaction matrix; performing convolution encoding on the medical record-habit context fine-grained matrix to obtain the medical record-habit context fine-grained semantic information field.

[0036] More specifically, in an embodiment of the present application, the intrinsic feature vector after semantic modulation of the medical record context and the intrinsic feature vector after semantic modulation of the habit context are fused based on the medical record-habit context fine-grained semantic information field to obtain a comprehensive feature vector of the patient, including: multiplying the medical record-habit context fine-grained semantic information field by the intrinsic feature vector after semantic modulation of the medical record context to obtain a medical record context semantic feature vector after information field mapping; multiplying the medical record-habit context fine-grained semantic information field by the intrinsic feature vector after semantic modulation of the habit context to obtain a habit context semantic feature vector after information field mapping; weighted fusion of the medical record context semantic feature vector after information field mapping and the habit context semantic feature vector after information field mapping to obtain the comprehensive feature vector of the patient.

[0037] In the embodiment of the present application, specifically, the medical record context semantic feature vector and the habit context semantic feature vector are subjected to kernel association fusion based on dimensionality degradation projection to obtain a patient comprehensive feature vector, including: processing the medical record context semantic feature vector and the habit context semantic feature vector according to the following formula to obtain a patient comprehensive feature vector; wherein the formula is:

[0038]

[0039] v c =αv kt +βv st

[0040] Among them, V 1 Represents the semantic feature vector of medical record context, V 2 represents the semantic feature vector of the habit context, PCA(·) represents the principal component analysis, and U 1 represents the sequence of semantic eigendecomposition vectors of medical record context, Λ 1 represents the medical record context semantic diagonal matrix, λ 11 and λ 1mdenote the eigenvalues ​​of the first and mth positions of the medical record context semantic diagonal matrix, respectively. T denotes the transpose of the vector. concat{·;·} denotes vector concatenation. v 11 、v 12 、v 1m Respectively represent the first, second and mth eigenvectors of the sequence of the semantic eigendecomposition vector of the medical record context, V k represents the semantic intrinsic feature vector of the medical record context, U 2 represents the sequence of habitual context semantic eigendecomposition vectors, Λ 2 represents the medical record context semantic diagonal matrix, λ 21 and λ 2m They represent the eigenvalues ​​of the first and mth positions of the custom context semantic diagonal matrix, respectively, and v 21 、v 22 、v 2m Respectively represent the first, second and mth eigenvectors of the sequence of the customary context semantic eigendecomposition vector, V s represents the semantic intrinsic feature vector of the customary context, v i represents the ith eigenvalue of the semantic eigenvector of the medical record context, v j represents the jth eigenvalue of the semantic intrinsic feature vector of the customary context, represents the square of the vector norm, exp represents the natural exponential function, V' k Represents the intrinsic feature vector after the medical record context semantic modulation, V' s represents the intrinsic feature vector after the semantic modulation of the habitual context, represents matrix multiplication, S represents the scale of the vector, M x Represents the medical history-habit context fine-grained interaction matrix, Conv 3×3 (M x ) indicates that M x A convolutional encoding operation based on a convolution kernel of 3×3 is performed, Ω represents the medical history-habit context fine-grained semantic information field, v kt represents the semantic feature vector of the medical record context after information field mapping, v st represents the semantic feature vector of the habitual context after information field mapping, α and β represent weighted hyperparameters, and v c Represents the patient's comprehensive feature vector.

[0041] That is, in the technical solution of the present application, the medical record context semantic feature vector and the habitual context semantic feature vector are subjected to kernel association fusion based on dimensionality degradation projection. The process first maps the medical record context semantic feature vector and the habitual context semantic feature vector to the intrinsic decomposition space to obtain the medical record context semantic intrinsic feature vector and the habitual context semantic intrinsic feature vector. A person of ordinary skill in the art should know that the process of mapping to the intrinsic decomposition space is a combination of dimensionality reduction and feature extraction. In a specific example, the eigenvalue decomposition technique of principal component analysis (PCA) is used to find the intrinsic decomposition space, and the intrinsic decomposition space mapping is performed. The medical record context semantic intrinsic feature vector and the habitual context semantic intrinsic feature vector after information field mapping not only represent the projection of the original feature vector in the new coordinate system, but also reflect the main direction of change of the data, which helps to avoid the "dimensionality disaster", that is, the problem that the sample density drops sharply as the number of features increases.

[0042] Next, after mapping the feature vector to the eigendecomposition space, it is necessary to perform inter-domain feature calibration on the medical record context semantics eigenfeature vector and the habitual context semantics eigenfeature vector to adjust the source domain feature vector to better match the data distribution of the target domain. This operation aims to capture and compensate for the systematic deviations between the two domains to reduce the hidden disturbance of the class expression of the local feature distribution of the medical record context semantics eigenfeature vector and the habitual context semantics eigenfeature vector to the overall class expression. After modulation, the eigenfeature vector modulated by the medical record context semantics and the eigenfeature vector modulated by the habitual context semantics can retain the characteristics of the source domain while being more in line with the actual needs of the target domain, thereby enhancing the model's understanding and application of the target domain data.

[0043] After completing the scale modulation of the feature vector, a medical record-habit context fine-grained semantic information field is constructed between the intrinsic feature vector after the medical record context semantic modulation and the intrinsic feature vector after the habit context semantic modulation. The medical record-habit context fine-grained semantic information field is used to describe a more detailed relationship between feature vectors. The introduction of the medical record-habit context fine-grained semantic information field enables the model to go beyond simple point-to-point comparison and consider more complex patterns and contextual information, thereby enhancing the model's sensitivity to subtle differences and improving the quality of feature representation. Finally, the intrinsic feature vector after the medical record context semantic modulation and the intrinsic feature vector after the habit context semantic modulation are fused based on the medical record-habit context fine-grained semantic information field to obtain the patient's comprehensive feature vector.

[0044] Specifically, in an embodiment of the present application, the patient information mixed convolutional coding module is a convolutional neural network model including multiple mixed convolutional layers.

[0045] Specifically, in order to make full use of the multi-scale characteristics of the convolution operation, the patient comprehensive feature vector is passed through the patient information mixed convolution coding module to obtain the multi-scale patient comprehensive feature vector, so as to extract a more comprehensive and multi-level feature representation. The patient information mixed convolution coding module is a convolutional neural network model comprising multiple mixed convolution layers. The patient comprehensive feature vector may contain feature information of different scales. For example, some features may be related to the risk of diabetic complications, while other features may be related to the details of the specific condition. By using multiple mixed convolution layers, features can be extracted and captured at different scales. Each convolution layer can focus on features of different scales, so that the model can more comprehensively understand the information in the patient comprehensive feature vector. The mixed convolution layer can construct a hierarchical feature representation by stacking multiple convolution layers. The lower convolution layer can capture more local and detailed features, while the higher convolution layer can capture more abstract and global features. By combining multiple mixed convolution layers, feature representations of different levels can be obtained. This helps to extract more representative and discriminative features, thereby improving the model's understanding and expression capabilities of the patient's comprehensive features. The convolutional layer in the convolutional neural network has the characteristics of parameter sharing and dimensionality reduction. Parameter sharing means that the same convolution kernel can be used to extract features at different positions, thereby reducing the number of model parameters and improving the efficiency and generalization ability of the model. Dimensionality reduction can reduce the dimension of features and extract more compact and high-dimensional feature representations. These characteristics make the extraction of multi-scale patient comprehensive feature vectors more efficient and effective.

[0046] In the above-mentioned data intelligent analysis and processing system 100 based on deep data mining, the diagnosis result generation module 130 is used to pass the multi-scale patient comprehensive feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is a high-risk group for diabetic complications. The classifier can have the ability to predict by learning the relationship between the feature vector and the high-risk group for diabetic complications. By training the classifier, it can learn the patterns and features related to the high-risk group in the feature vector, so that new patients can be classified and predicted. The classifier can map the optimized feature vector to a specific classification label, such as high risk or low risk, to indicate whether the patient belongs to the high-risk group for diabetic complications.

[0047] Specifically, in the embodiment of the present application, the diagnosis result generation module 130 includes: a fully connected encoding unit, used to use the fully connected layer of the classifier to fully connect the multi-scale patient comprehensive feature vector to obtain a coded classification feature vector; and a diagnosis result generation unit, used to input the coded classification feature vector into the Softmax classification function of the classifier to obtain the classification result.

[0048] In summary, the data intelligent analysis and processing system based on deep data mining according to the embodiment of the present application has been explained, which uses artificial intelligence technology based on the field of deep learning to extract and encode the original data of the medical records to be tested and the patient's living habit records to obtain the classification result of whether the patient is a high-risk group for diabetic complications. In this way, by intelligently judging whether the patient is a high-risk group for diabetic complications, it can effectively avoid delaying the disease and thus missing the best time to treat diabetes.

[0049] As described above, the data intelligent analysis and processing system 100 based on deep data mining according to the embodiment of the present application can be implemented in various terminal devices, such as a server for data intelligent analysis and processing based on deep data mining. In one example, the data intelligent analysis and processing system 100 based on deep data mining according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the data intelligent analysis and processing system 100 based on deep data mining can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the data intelligent analysis and processing system 100 based on deep data mining can also be one of the many hardware modules of the terminal device.

[0050] Alternatively, in another example, the data intelligent analysis and processing system 100 based on deep data mining and the terminal device may also be separate devices, and the data intelligent analysis and processing system 100 based on deep data mining may be connected to the terminal device via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0051] Based on the same inventive concept, an embodiment of the present application also provides a data intelligent analysis and processing method based on deep data mining, which can be used to implement the system described in the above embodiment, as described in the following embodiment.

[0052] Figure 3 Flow chart of a data intelligent analysis and processing method based on deep data mining according to an embodiment of the present application. Figure 3 As shown, according to the data intelligent analysis and processing method based on deep data mining according to the embodiment of the present application, the steps include: S110, acquiring the original data of the medical records to be detected and the patient's living habit records; S120, processing the original data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector; and S130, passing the multi-scale patient comprehensive feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is a high-risk group for diabetic complications.

[0053] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned data intelligent analysis and processing method based on deep data mining have been referred to above. Figure 1 to Figure 2 The invention has been introduced in detail in the description of the data intelligent analysis and processing system based on deep data mining, and therefore, its repeated description will be omitted.

[0054] In summary, the data intelligent analysis and processing method based on deep data mining according to the embodiment of the present application has been explained, which uses artificial intelligence technology based on the field of deep learning to extract and encode the original data of the medical records to be tested and the patient's living habit records to obtain the classification result of whether the patient is a high-risk group for diabetic complications. In this way, by intelligently judging whether the patient is a high-risk group for diabetic complications, it can effectively avoid delaying the disease and thus missing the best time to treat diabetes.

[0055] The "first", "second", "third", "fourth", etc. (if any) referred to in this application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods or devices.

[0056] It should be noted that the descriptions involving "first", "second", etc. in this application are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0057] Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0058] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0059] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0060] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0061] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A data intelligent analysis and processing system based on deep data mining, characterized in that: include: Patient data acquisition module, used to obtain the original data of the medical records to be tested and the patient's living habit records; A patient data processing module, used for processing the raw data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector; The diagnosis result generating module is used to pass the multi-scale patient comprehensive feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is a high-risk group for diabetic complications.

2. The data intelligent analysis and processing system based on deep data mining according to claim 1 is characterized in that: The patient data processing module comprises: A patient raw data processing unit, used for extracting semantic features from the raw data of the medical record to be detected and the patient's living habit record to obtain a medical record context semantic feature vector and a habit context semantic feature vector; The patient feature extraction unit is used to extract and encode the medical record context semantic feature vector and the habit context semantic feature vector to obtain the multi-scale patient comprehensive feature vector.

3. The data intelligent analysis and processing system based on deep data mining according to claim 2 is characterized in that: The patient raw data processing unit is used for: The original data of the medical record to be detected and the patient's living habit record are respectively passed through a converter-based patient medical record semantic encoder to obtain the medical record context semantic feature vector and the habit context semantic feature vector.

4. The data intelligent analysis and processing system based on deep data mining according to claim 3 is characterized in that: The patient medical record semantic encoder is a transformer-based context encoder that includes an embedding layer.

5. The data intelligent analysis and processing system based on deep data mining according to claim 4 is characterized in that: The patient feature extraction unit comprises: A patient information fusion subunit, used for performing kernel association fusion based on dimensionality degradation projection on the medical record context semantic feature vector and the habit context semantic feature vector to obtain a patient comprehensive feature vector; The patient information mixed convolution encoding subunit is used to perform mixed feature extraction on the patient comprehensive feature vector to obtain the multi-scale patient comprehensive feature vector.

6. The data intelligent analysis and processing system based on deep data mining according to claim 5 is characterized in that: The patient information fusion subunit comprises: A feature eigendecomposition unit, used for mapping the medical record context semantic feature vector and the habit context semantic feature vector to an eigendecomposition space to obtain a medical record context semantic eigenfeature vector and a habit context semantic eigenfeature vector; An inter-domain feature calibration unit, used for performing inter-domain feature calibration on the medical record context semantics intrinsic feature vector and the custom context semantics intrinsic feature vector to obtain a medical record context semantics modulated intrinsic feature vector and a custom context semantics modulated intrinsic feature vector; A fine-grained semantic information field construction unit, used to construct a medical record-habit context fine-grained semantic information field between the intrinsic feature vector after the medical record context semantics modulation and the intrinsic feature vector after the habit context semantics modulation; A patient comprehensive feature generation unit is used to fuse the medical record context semantically modulated intrinsic feature vector and the habit context semantically modulated intrinsic feature vector based on the medical record-habit context fine-grained semantic information field to obtain the patient comprehensive feature vector.

7. The data intelligent analysis and processing system based on deep data mining according to claim 6 is characterized in that: The fine-grained semantic information field construction unit is used to: Multiplying the transposed vector of the intrinsic feature vector after the semantic modulation of the medical record context and the intrinsic feature vector after the semantic modulation of the habit context, and then dividing by the scale of the intrinsic feature vector after the semantic modulation of the habit context to obtain a medical record-habit context fine-grained interaction matrix; The medical record-habit context fine-grained matrix is ​​convolutionally encoded to obtain the medical record-habit context fine-grained semantic information field.

8. The data intelligent analysis and processing system based on deep data mining according to claim 7 is characterized in that: The patient comprehensive feature generating unit is used for: Multiplying the medical record-habit context fine-grained semantic information field by the medical record context semantically modulated intrinsic feature vector to obtain a medical record context semantic feature vector after information field mapping; Multiplying the medical record-habit context fine-grained semantic information field by the habit context semantically modulated intrinsic feature vector to obtain the habit context semantic feature vector after information field mapping; The medical record context semantic feature vector after the information field mapping and the habit context semantic feature vector after the information field mapping are weightedly fused to obtain the patient comprehensive feature vector.

9. The data intelligent analysis and processing system based on deep data mining according to claim 8 is characterized in that: The patient information mixed convolutional coding subunit is used for: The patient comprehensive feature vector is passed through a patient information mixed convolutional coding module to obtain the multi-scale patient comprehensive feature vector, wherein the patient information mixed convolutional coding module is a convolutional neural network model including multiple mixed convolutional layers.

10. A data intelligent analysis and processing method based on deep data mining, characterized in that: include: Obtain the original data of the medical records to be tested and the patient's living habits records; Processing the raw data of the medical records to be detected and the patient's living habit records to obtain a multi-scale patient comprehensive feature vector; The multi-scale patient comprehensive feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the patient is a high-risk group for diabetic complications.

Images