Lung cancer multi-disease co-patient prediction method and system based on co-disease network model

Abstract

The invention provides a lung cancer multi-disease co-patient prediction method and system based on a co-disease network model, and relates to the field of intelligent medical treatment, and the method comprises the steps: obtaining clinical diagnosis and treatment data of a lung cancer patient; preprocessing the clinical diagnosis and treatment data; performing feature extraction and feature fusion on the preprocessed clinical diagnosis and treatment data to determine fusion features; determining common disease association according to the fusion features; establishing a common disease network model based on a graph neural network by taking the common disease association as an edge and each disease as a node; performing feature learning on each node of the common disease network model by adopting a graph convolutional neural network, and determining deep features of the common disease network model; and carrying out co-disease risk prediction on the deep-level features by using a risk assessment algorithm. According to the method, the accuracy and efficiency of lung cancer co-disease prediction are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medicine, and particularly to a method and system for predicting multiple co - morbidities of lung cancer based on a co - morbidity network model. Background Art

[0002] A co - morbidity network is a model that represents the co - morbidity relationships between diseases in a graphical form. In this model, diseases are regarded as nodes in the network, and the co - morbidity relationship between two diseases is modeled as an edge between the nodes. The method for predicting multiple co - morbidities of lung cancer based on the co - morbidity network model is a method for disease prediction by constructing and analyzing the complex association network of lung cancer and its related co - morbidities. Based on the multi - modal clinical data of patients, each disease is modeled as a node in the network, the co - morbidity relationship is modeled as an edge between the nodes, and potential associations between diseases are mined through technologies such as graph neural networks (GNN).

[0003] Lung cancer is one of the most common malignant tumors globally. Its complex pathological characteristics and high - incidence complex co - morbidity patterns make the diagnosis and treatment of lung cancer patients extremely difficult. With the development of medical technology, more and more studies have shown that co - morbidities have a significant impact on the treatment and prognosis of lung cancer. The need for co - morbidity management of lung cancer patients is becoming increasingly urgent. Therefore, exploring how to effectively utilize existing data to analyze the co - morbidity patterns of lung cancer patients and provide diagnosis and treatment support has become an important technical direction in the current medical field.

[0004] However, existing lung cancer prediction methods usually rely on single clinical features, ignoring the potential value of multi - modal information, resulting in insufficient prediction accuracy. Most existing co - morbidity networks adopt static network models based on nodes and edges, failing to effectively reflect multi - dimensional and dynamic relationships, making the efficiency of lung cancer prediction too low. Existing lung cancer prediction methods use decision tree models to analyze co - morbidity networks. When it is difficult to efficiently analyze high - dimensional data, it is difficult to effectively capture complex co - morbidity patterns. Summary of the Invention

[0005] In order to solve the technical problems that existing lung cancer prediction methods usually rely on single clinical features, ignore the potential value of multi - modal information, resulting in insufficient prediction accuracy; most existing co - morbidity networks adopt static network models based on nodes and edges, failing to effectively reflect multi - dimensional and dynamic relationships, making the efficiency of lung cancer prediction too low; existing lung cancer prediction methods use decision tree models to analyze co - morbidity networks, and it is difficult to effectively capture complex co - morbidity patterns when it is difficult to efficiently analyze high - dimensional data, the present invention provides a method and system for predicting multiple co - morbidities of lung cancer based on a co - morbidity network model.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First aspect:

[0008] A method for predicting co - morbidity of lung cancer based on a co - morbidity network model provided by an embodiment of the present invention includes:

[0009] S1: Obtain the clinical diagnosis and treatment data of lung cancer patients;

[0010] S2: Pre - process the clinical diagnosis and treatment data;

[0011] S3: Extract and fuse features from the pre - processed clinical diagnosis and treatment data to determine the fused features;

[0012] S4: Determine the co - morbidity associations according to the fused features;

[0013] S5: Use the co - morbidity associations as edges and each disease as a node to establish a co - morbidity network model based on a graph neural network;

[0014] S6: Use a graph convolutional neural network to perform feature learning on each node of the co - morbidity network model to determine the deep - level features of the co - morbidity network model;

[0015] S7: Use a risk assessment algorithm to predict the co - morbidity risk for the deep - level features.

[0016] Second aspect:

[0017] A system for predicting co - morbidity of lung cancer based on a co - morbidity network model provided by an embodiment of the present invention includes:

[0018] A processor;

[0019] A memory, on which computer - readable instructions are stored. When the computer - readable instructions are executed by the processor, the method for predicting co - morbidity of lung cancer based on a co - morbidity network model as described in the first aspect is implemented.

[0020] Third aspect:

[0021] A computer - readable storage medium provided by an embodiment of the present invention, on which a computer program is stored. When the program is executed by the processor, the method for predicting co - morbidity of lung cancer based on a co - morbidity network model as described in the first aspect is implemented.

[0022] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0023] In the embodiments of the present invention, through comorbidity network modeling, lung cancer and its related comorbidities are presented in a networked form, which can deeply explore the complex association patterns among multiple diseases, effectively reflect multi-dimensional and dynamic relationships, enhance the understanding of the comorbidity status of lung cancer patients, improve the accuracy and flexibility of comorbidity prediction, use a graph convolutional neural network for deep feature learning, extract the potential associations between diseases, improve the accuracy of comorbidity risk prediction, and find the optimal balance between efficacy and side effects through a risk assessment algorithm, significantly improving the accuracy and efficiency of comorbidity prediction, enabling the recommendation of personalized treatment plans for patients, improving the treatment effect, and reducing the risk of side effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0025] Figure 1 It is a schematic flowchart of a method for predicting co-occurrence of multiple diseases in lung cancer based on a comorbidity network model provided by an embodiment of the present invention;

[0026] Figure 2 It is a schematic structural diagram of a system for predicting co-occurrence of multiple diseases in lung cancer based on a comorbidity network model provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] The following will describe the technical solutions in the present invention with reference to the accompanying drawings.

[0028] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0029] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same.

[0030] In the embodiments of the present invention, sometimes subscripts such as W1 It may be written in a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0031] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0032] Refer to the attached Figure 1 , which shows a schematic flowchart of a method for predicting co-occurrence of multiple diseases in lung cancer based on a comorbidity network model provided by an embodiment of the present invention.

[0033] An embodiment of the present invention provides a method for predicting co-occurrence of multiple diseases in lung cancer based on a comorbidity network model. The method includes:

[0034] S1: Obtain the clinical diagnosis and treatment data of lung cancer patients.

[0035] It should be noted that comprehensively collecting multi-modal clinical diagnosis and treatment data of lung cancer patients provides a comprehensive and reliable data basis for subsequent feature extraction and diagnosis and treatment decision-making.

[0036] In a possible implementation manner, the clinical diagnosis and treatment data specifically includes: doctor's order information, treatment information, medical record text records, DICOM images, surgical information, and laboratory results.

[0037] Among them, doctor's order information refers to the medical instructions issued by doctors for patients, including medications, examinations, treatment arrangements, etc. Treatment information refers to the treatment records of patients, including chemotherapy, radiotherapy, surgery, immunotherapy and other treatment plans and their implementation situations. Medical record text records are the records written by doctors during the diagnosis and treatment process, including initial diagnosis medical records, follow-up records, diagnostic opinions, etc. DICOM images refer to Digital Imaging and Communications in Medicine, which is used to store and transmit medical image data. Surgical information is the detailed record of the surgery received by the patient, including the type of surgery, process, intraoperative findings, and postoperative recovery. Laboratory results are the test results from the patient's biological samples (such as blood, urine, tissue), which are used for diagnosis and treatment guidance.

[0038] In a possible embodiment, S1 is specifically:

[0039] Extract the clinical diagnosis and treatment data of lung cancer patients from the hospital information system and electronic medical records.

[0040] Among them, the hospital information system (HIS) refers to a comprehensive system used in hospitals to manage patient medical records, finances, resource allocation, etc. The electronic medical record (EMR) refers to the patient's diagnosis and treatment information stored in a digital manner, including medical record text, test results, image reports, etc.

[0041] It should be noted that extracting the clinical diagnosis and treatment data of lung cancer patients from the hospital information system (HIS) and electronic medical records (EMR) can efficiently and comprehensively obtain the multimodal information of patients (such as doctor's orders, images, test results, etc.). This automated data extraction method improves the information collection efficiency, avoids the errors of manual entry, and at the same time ensures the integrity and timeliness of the data.

[0042] S2: Preprocess the clinical diagnosis and treatment data.

[0043] In a possible implementation, S2 is specifically:

[0044] Preprocess the clinical diagnosis and treatment data including data cleaning and data standardization.

[0045] Among them, data cleaning refers to processing the missing values, outliers and duplicate values in the clinical data to ensure data integrity and consistency. Data standardization is to convert the data into a form with the same distribution and scale for unified analysis.

[0046] It should be noted that preprocessing the clinical diagnosis and treatment data through data cleaning and data standardization can effectively improve the quality and usability of the data. The cleaned data eliminates errors and noise, ensuring the reliability of the analysis results. The standardized processing enables data with different features to be compared on the same scale, avoiding the interference of feature differences on model learning.

[0047] Data cleaning specifically includes:

[0048] Use the interpolation algorithm or mean substitution method to fill in the missing clinical diagnosis and treatment data.

[0049] Use the filtering method to eliminate the abnormal clinical diagnosis and treatment data.

[0050] Data standardization is specifically:

[0051]

[0052] Among them, x i ′ represents the standardized clinical diagnosis and treatment data, x i represents the clinical diagnosis and treatment data, μ represents the mean of the clinical diagnosis and treatment data, and σ represents the standard deviation of the clinical diagnosis and treatment data.

[0053] S3: Extract features and fuse features from the preprocessed clinical diagnosis and treatment data to determine the fused features.

[0054] In a possible implementation, S3 specifically includes:

[0055] S301: Extract features from the preprocessed clinical diagnosis and treatment data to determine data features, where the data features include text features, imaging features, and numerical features.

[0056] S302: Integrate the text features, imaging features, and numerical features to obtain data features:

[0057] X = concat([f text (T), f image (I), f num (N)])

[0058] where X represents the integrated feature, concat represents the feature concatenation function, f text (T) represents the text feature, f image (I) represents the imaging feature, f num (N) represents the numerical feature.

[0059] It should be noted that by extracting features and integrating features from the preprocessed clinical diagnosis and treatment data, multi-modal information such as text, imaging, and numerical values can be effectively integrated into a unified feature vector, fully exploring the potential value of various types of data.

[0060] In the present invention, the system can quantify the indicators (such as white blood cell count, liver function, etc.) in the blood test data to generate structured data. At the same time, the imaging data (such as CT, MRI images, etc.) extracts lesion features through a convolutional neural network (CNN) to more accurately describe the imaging features of the patient. These features are used in subsequent models for risk prediction and comorbidity network construction.

[0061] S4: Determine the comorbidity association according to the integrated feature.

[0062] Among them, it refers to the association between two or more diseases, usually quantified by statistical methods (such as support and confidence).

[0063] It should be noted that determining the comorbidity association according to the integrated feature can quantify the association strength between diseases, intuitively display the comorbidity pattern, and provide a scientific basis for accurate diagnosis and personalized treatment.

[0064] In a possible implementation, S4 specifically includes:

[0065] S401: Determine the comorbidity association based on the integrated feature.

[0066] S402: Use the support formula and confidence formula to quantify the comorbidity association:

[0067]

[0068] Among them, Support represents the support function, Confidence represents the confidence function, Count represents the calculation function, A and B represent disease combinations, and Total represents the total number of patients.

[0069] S5: Use the comorbidity association as the edge and each disease as a node to establish a comorbidity network model based on the graph neural network.

[0070] Among them, a node refers to the basic unit in the network, representing a disease in the comorbidity network; an edge refers to the line segment connecting nodes, representing the comorbidity relationship between two diseases in the comorbidity network. The graph neural network (GNN) is a deep learning model based on graph-structured data, capable of learning the features of nodes, edges, and the overall network.

[0071] Specifically, the node represents the disease diagnosis name, the edge represents the comorbidity relationship between diseases, the size of the node represents the duration of disease diagnosis, and the thickness of the connection line represents the strength of the comorbidity association between diseases.

[0072] It should be noted that by constructing a comorbidity network model based on the graph neural network, the complex associations between diseases can be clearly represented, multi-dimensional and potential comorbidity relationships can be captured, and the understanding ability of the disease network structure can be improved.

[0073] In a possible implementation manner, S5 specifically includes:

[0074] S501: Use the comorbidity association as the edge and each disease as a node to establish a preliminary comorbidity network.

[0075] S502: Use the K-means clustering algorithm to perform clustering analysis on the preliminary comorbidity network:

[0076]

[0077] Among them, J represents the result of clustering analysis, x i represents the feature vector of the i-th patient, c i represents the cluster to which the feature vector of the i-th patient belongs, represents the center point of the cluster c i and i = 1, 2,..., n, where n represents the total number of patients.

[0078] S503: According to the comorbidity association and the result of clustering analysis, construct a complete comorbidity network.

[0079] S504: Perform feature weighting on the complete comorbidity network to determine the comorbidity network model:

[0080]

[0081] Among them, S ijrepresents the correlation score between disease i and disease j, σ represents the activation function, α k represents the weight normalization coefficient of the kth feature, W k represents the feature weight of the kth feature, X i,k represents the kth feature of disease i, W y Represents the weight of attribute feature y, Y j represents the feature weight normalization coefficient of disease j, b represents the bias term, and exp represents the exponential function.

[0082] It should be noted that by modeling comorbidity associations as preliminary networks and using K-means cluster analysis, the comorbidity network can identify the comorbidity patterns and grouping rules of diseases from patient characteristics, and further optimize it by combining feature weighting. It can dynamically reflect the strength of associations between diseases and their multidimensional characteristics, thereby improving the network's expressive power.

[0083] In the present invention, when constructing a comorbidity network, each disease is first regarded as a node of the graph, and the comorbidity relationship between two diseases is regarded as an edge of the graph to form a preliminary comorbidity network. In order to improve the accuracy of network construction, a standard for defining comorbidity is set: when a certain proportion of patients in the patient population suffer from two diseases at the same time, the two diseases are considered to have a comorbidity relationship and are recorded in the graph structure. After the comorbidity network is constructed, a graph convolutional neural network (GCN) is used for feature extraction and learning. The input of the GCN is the constructed comorbidity network structure. The node attributes of the network include the clinical characteristics of the patient (such as age, gender, previous medical history, etc.), GCN The N model learns node features layer by layer through convolutional layers and calculates the feature vector of each node, ultimately achieving in-depth mining of potential relationships between diseases. In this way, the system can not only identify surface comorbidity relationships, but also mine potential comorbidity correlations. After obtaining the characteristic vector of the disease, the present invention uses a clustering algorithm to analyze the comorbidity network and divides the disease into different clustering groups to help identify comorbidity patterns with similar characteristics. This process is achieved through clustering algorithms such as K-means. The specific method is to aggregate diseases with close comorbidity relationships together according to the similarity of the disease feature vectors to form a subnetwork.

[0084] S6: Use graph convolutional neural network to learn the features of each node of the comorbidity network model and determine the deep-level features of the comorbidity network model.

[0085] Among them, the graph convolutional neural network (GCN) is a neural network model designed for graph structured data, which can learn representations in the features of nodes and their neighbors. Deep features are high-order features gradually extracted in multi-layer neural networks, reflecting complex correlation relationships.

[0086] It should be noted that by using a graph convolutional neural network to perform feature learning on the nodes of the comorbidity network, the attributes of disease nodes and the associations between neighbor nodes can be fully utilized to extract deep features reflecting complex comorbidity relationships, capture the high-order dependencies between nodes, and improve the understanding of potential comorbidity patterns and the accuracy of disease risk prediction.

[0087] In one possible implementation, S6 is specifically as follows:

[0088] Determine the deep features of the comorbidity network model according to the following formula:

[0089]

[0090] where represents the feature representation of node v of the comorbidity network model in the k-th layer of the convolutional neural network, σ represents the activation function, W (k) represents the learnable weight matrix of the k-th layer, AGG represents the aggregation function of neighbor node features, represents the feature representation of neighbor node u in the (k - 1)-th layer, N(v) represents the set of neighbor nodes of node v, h t represents the final embedding representation of node t, softmax represents the activation function, q t represents the query vector of node t in the multi-head attention, v k represents the value vector of neighbor node k in the multi-head attention, k k represents the key vector of neighbor node k in the multi-head attention, represents the normalization term of the feature vector, k = 1, 2,..., T, and T represents the total number of heads of the multi-head attention.

[0091] S7: Use a risk assessment algorithm to perform comorbidity risk prediction on the deep features.

[0092] In one possible implementation, S7 is specifically as follows:

[0093] Perform comorbidity risk prediction on the deep features through the evaluation formula based on the risk assessment algorithm.

[0094] The evaluation formula is specifically as follows:

[0095]

[0096] where min represents minimization, represents the error between the actual efficacy and the predicted efficacy, y represents the actual efficacy, represents the predicted efficacy, μ represents the weight for balancing efficacy and side effects, represents the error between the actual side effects and the predicted side effects, s represents the actual side effects, represents the predicted side effects.

[0097] It should be noted that by using a risk assessment algorithm to predict the comorbidity risk of deep features, the possibility and severity of a patient developing other comorbidities can be quantified, significantly improving the prediction accuracy and clinical practicability, and providing scientific risk grading and intervention suggestions for doctors.

[0098] In the present invention, in order to improve the accuracy of comorbidity prediction, imaging data and clinical data are combined to achieve multimodal data fusion. After the imaging data is preprocessed and feature extracted, it is jointly input into a deep learning model with the clinical feature data. The system processes these data through a multimodal neural network to extract comprehensive features, thereby achieving a comprehensive assessment of the comorbidity risk, generating a personalized comorbidity risk assessment report according to the characteristics of different patients, which will not only show the comorbidities that the patient is most likely to suffer from, but also provide targeted prevention and treatment suggestions based on the comorbidity risk distribution.

[0099] In a possible implementation manner, after S7, it further includes:

[0100] S8: Determine a treatment plan based on the comorbidity risk prediction result.

[0101] S9: Treat the lung cancer patient according to the treatment plan.

[0102] It should be noted that formulating a treatment plan based on the comorbidity risk prediction result and implementing personalized treatment can optimize the treatment effect of lung cancer, control the development of comorbidities at the same time, and significantly improve the prognosis and quality of life of patients.

[0103] In the present invention, based on the patient's comorbidity pattern and clinical characteristics, personalized treatment plan recommendations are provided for doctors. The analysis results of the comorbidity network model are combined with a recommendation algorithm to automatically generate a personalized treatment plan that complies with the treatment guidelines according to the specific condition of the patient. The system provides an intuitive comorbidity network diagram to show the comorbidity structure of the patient. The color and size of each disease node can be adjusted in real time according to the comorbidity risk. Doctors can click on the node to view the detailed information of the disease. This visual display facilitates doctors to quickly identify the patient's comorbidity pattern and improve the diagnosis efficiency. The system dynamically adjusts the treatment plan and comorbidity risk prediction according to the patient's treatment effect and comorbidity progress. For example, after the patient has used a certain treatment plan for a period of time, the system can automatically evaluate its effect according to real-time data. If it is found that the curative effect is not good, the system will prompt the doctor to adjust the plan. This dynamic feedback mechanism ensures that the patient always obtains the best treatment plan at different stages of the disease course.

[0104] For example, for a 65-year-old male lung cancer patient with a past medical history including comorbidities such as hypertension and diabetes, the system first obtains the patient's medical record data from the hospital's HIS and EMR, including information such as imaging examinations, blood tests, and past medical history. These information are cleaned, standardized, and feature-extracted. Comorbidity information such as diabetes and hypertension is incorporated into the analysis in the form of ICD codes. Subsequently, the patient's disease information is added to the comorbidity network model, and the comorbidity relationship is feature-extracted through a graph neural network. According to the patient's characteristics such as age and gender, the possible future comorbidities are quantitatively evaluated, and an individualized comorbidity risk report is generated. Finally, combining the patient's comorbidity pattern and real-time health data, a personalized treatment plan is recommended for the patient, and the diagnosis and treatment recommendations are dynamically adjusted according to the changes in the condition during the treatment process. If the patient's blood pressure rises during the treatment process, the system will prompt the doctor to adjust the medication in real time to ensure good management of the patient's comorbidities.

[0105] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0106] In the embodiment of the present invention, through comorbidity network modeling, lung cancer and its related comorbidities are presented in a networked form, which can deeply explore the complex association patterns among multiple diseases, effectively reflect the multi-dimensional and dynamic relationships, enhance the understanding of the comorbidity status of lung cancer patients, improve the accuracy and flexibility of comorbidity prediction, use a graph convolutional neural network for deep feature learning, extract the potential associations between diseases, improve the accuracy of comorbidity risk prediction, and find the optimal balance point between efficacy and side effects through a risk assessment algorithm, significantly improving the accuracy and efficiency of comorbidity prediction, and can recommend personalized treatment plans for patients, improving the treatment effect and reducing the risk of side effects.

[0107] Refer to the appended Figure 2 drawings, which show a schematic structural diagram of a lung cancer multiple comorbidities prediction system based on a comorbidity network model provided by the present invention.

[0108] The present invention also provides a lung cancer multiple comorbidities prediction system 20 based on a comorbidity network model, which is applied to the above-mentioned lung cancer multiple comorbidities prediction method based on a comorbidity network model, and includes:

[0109] A processor 201.

[0110] A memory 202, on which computer-readable instructions are stored. When the computer-readable instructions are executed by the processor 201, the lung cancer multiple comorbidities prediction method based on a comorbidity network model as in the method embodiment is implemented.

[0111] The lung cancer comorbidity prediction system 20 based on the comorbidity network model provided by the present invention can execute the above-mentioned lung cancer comorbidity prediction method based on the comorbidity network model and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate further.

[0112] The beneficial effects brought by the technical solution provided by the embodiments of the present invention at least include:

[0113] In the embodiments of the present invention, through comorbidity network modeling, lung cancer and its related comorbidities are presented in a networked form, which can deeply explore the complex association patterns among multiple diseases, effectively reflect multi-dimensional and dynamic relationships, enhance the understanding of the comorbidity status of lung cancer patients, improve the accuracy and flexibility of comorbidity prediction, use a graph convolutional neural network for in-depth feature learning to extract potential associations between diseases, improve the accuracy of comorbidity risk prediction, and find the optimal balance point between efficacy and side effects through a risk assessment algorithm, significantly improving the accuracy and efficiency of comorbidity prediction, enabling personalized treatment plans to be recommended for patients, improving the treatment effect, and reducing the risk of side effects.

[0114] It should be understood that the processor in the embodiments of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0115] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0116] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any arbitrary combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions according to the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0117] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. Additionally, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0118] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or plural.

[0119] It should be understood that in various embodiments of the present invention, the magnitudes of the serial numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0120] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0121] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0122] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in an electrical, mechanical, or other form.

[0123] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0124] In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0125] If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0126] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the prediction method for co-morbidities of lung cancer based on a co-morbidity network model as in the method embodiment.

[0127] The computer-readable storage medium provided by the present invention can implement the steps and effects of the prediction method for co-morbidities of lung cancer based on a co-morbidity network model in the above method embodiment. To avoid repetition, the present invention will not elaborate further.

[0128] The beneficial effects brought by the technical solution provided by the embodiments of the present invention at least include:

[0129] In the embodiments of the present invention, through co-morbidity network modeling, lung cancer and its related co-morbidities are presented in a networked form, which can deeply explore the complex association patterns among multiple diseases, effectively reflect multi-dimensional and dynamic relationships, enhance the understanding of the co-morbidity status of lung cancer patients, improve the accuracy and flexibility of co-morbidity prediction, use a graph convolutional neural network for deep feature learning, extract potential associations between diseases, improve the accuracy of co-morbidity risk prediction, and find an optimal balance point between efficacy and side effects through a risk assessment algorithm, significantly improving the accuracy and efficiency of co-morbidity prediction, and can recommend personalized treatment plans for patients, improving the treatment effect and reducing the risk of side effects.

[0130] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0131] The following points need to be explained:

[0132] (1) The accompanying drawings of the embodiments of the present invention only relate to the structures involved in the embodiments of the present invention, and other structures may refer to the general designs.

[0133] (2) For clarity, in the drawings used to describe the embodiments of the present invention, the thickness of layers or regions is enlarged or reduced, that is, these drawings are not drawn to actual scale. It can be understood that when an element such as a layer, film, region, or substrate is referred to as being "on" or "under" another element, the element can be "directly" on or under the other element or there can be intervening elements.

[0134] (3) Without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0135] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for predicting multiple co-occurrence of lung cancer based on a co-occurrence network model, characterized in that: include: S1: Obtain clinical diagnosis and treatment data of lung cancer patients; S2: preprocessing the clinical diagnosis and treatment data; S3: Extract and fuse features of the preprocessed clinical diagnosis and treatment data to determine the fusion features; S4: Determine comorbidity associations based on the fusion signatures; S5: Taking the comorbidity association as an edge and each disease as a node, a comorbidity network model based on a graph neural network is established; S6: using a graph convolutional neural network to perform feature learning on each node of the comorbidity network model to determine the deep-level features of the comorbidity network model; S7: Utilize the risk assessment algorithm to predict the risk of comorbidity based on the deep-level features.

2. The method for predicting multiple co-occurrence of lung cancer based on a co-occurrence network model according to claim 1, characterized in that: The clinical diagnosis and treatment data specifically includes: doctor's advice information, treatment information, medical records, DICOM images, surgery information and laboratory results; The S1 is specifically: The clinical diagnosis and treatment data of lung cancer patients are extracted from the hospital information system and electronic medical records.

3. The method for predicting multiple co-occurrence of lung cancer based on a co-occurrence network model according to claim 1, characterized in that: The S2 is specifically: Performing preprocessing on the clinical diagnosis and treatment data including data cleaning and data standardization; The data cleaning specifically includes: Interpolation algorithms or mean substitution methods were used to fill in missing clinical diagnosis and treatment data; The filtering method was used to eliminate abnormal clinical diagnosis and treatment data; The data standardization is specifically as follows: Among them, x i ′ represents the standardized clinical diagnosis and treatment data, x i represents clinical diagnosis and treatment data, μ represents the mean of clinical diagnosis and treatment data, and σ represents the standard deviation of clinical diagnosis and treatment data.

4. The method for predicting multiple co-occurrence of lung cancer based on a co-occurrence network model according to claim 1, characterized in that: The S3 specifically includes: S301: extracting features from the preprocessed clinical diagnosis and treatment data to determine data features, wherein the data features include text features, image features, and numerical features; S302: Fusing the text feature, the image feature and the numerical feature to obtain the data feature: X=concat([f text (T),f image (I),f num (N)]) Among them, X represents the fusion feature, concat represents the feature concatenation function, and f text (T) represents the text feature, f image (I) represents the image feature, f num (N) indicates a numerical feature.

5. The method for predicting multiple co-occurrence of lung cancer based on co-occurrence network model according to claim 1, characterized in that: The S4 specifically includes: S401: Determine the comorbidity association based on the fusion feature; S402: Quantify the comorbidity association using a support formula and a confidence formula: Among them, Support represents the support function, Confidence represents the confidence function, Count represents the calculation function, A and B represent the disease combination, and Total represents the total number of patients.

6. The method for predicting multiple co-occurrence of lung cancer based on a co-occurrence network model according to claim 1, characterized in that: The S5 specifically includes: S501: Using the comorbidity association as an edge and each disease as a node, a preliminary comorbidity network is established; S502: Perform cluster analysis on the preliminary comorbidity network using a K-means clustering algorithm: Among them, J represents the cluster analysis result, x i represents the feature vector of the ith patient, c i represents the cluster to which the feature vector of the i-th patient belongs, Represents cluster c i The center point of , i = 1, 2, ..., n, n represents the total number of patients; S503: constructing a complete comorbidity network according to the comorbidity association and cluster analysis results; S504: Perform feature weighting on the complete comorbidity network to determine the comorbidity network model: Among them, S ij represents the correlation score between disease i and disease j, σ represents the activation function, α k represents the weight normalization coefficient of the kth feature, W k represents the feature weight of the kth feature, X i,k represents the kth feature of disease i, W y Represents the weight of attribute feature y, Y j represents the feature weight normalization coefficient of disease j, b represents the bias term, and exp represents the exponential function.

7. The method for predicting multiple co-occurrence of lung cancer based on co-occurrence network model according to claim 1, characterized in that: The S6 is specifically: The deep-level features of the comorbidity network model are determined according to the following formula: in, represents the feature representation of node v in the kth layer of the convolutional neural network of the comorbidity network model, σ represents the activation function, and W (k) represents the learnable weight matrix of the kth layer, AGG represents the aggregation function of neighbor node features, represents the feature representation of neighbor node u in the k-1th layer, N(v) represents the set of neighbor nodes of node v, and h t represents the final embedding representation of node t, softmax represents the activation function, and q t represents the query vector of node t in multi-head attention, v k Represents the value vector of neighbor node k in multi-head attention, k k represents the key vector of neighbor node k in multi-head attention, represents the normalization term of the feature vector, k=1,2,...,T, where T represents the total number of multi-head attention heads.

8. The method for predicting multiple co-occurrence of lung cancer based on co-occurrence network model according to claim 1, characterized in that: The S7 is specifically: Predicting the risk of comorbidity for the deep-level features by using an evaluation formula based on the risk evaluation algorithm; The evaluation formula is specifically: Among them, min means minimization, represents the error between the actual efficacy and the predicted efficacy, y represents the actual efficacy, represents the predicted efficacy, μ represents the weight of balancing efficacy and side effects, represents the error between the actual side effect and the predicted side effect, s represents the actual side effect, Indicates predicted side effects.

9. The method for predicting multiple co-occurrence of lung cancer based on a co-occurrence network model according to claim 1, characterized in that: The S7 also includes: S8: Determine the treatment plan based on the comorbidity risk prediction results; S9: Treating the lung cancer patient according to the treatment plan.

10. A lung cancer multi-disease co-occurrence prediction system based on a co-disease network model, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for predicting multiple co-occurrence of lung cancer based on a co-morbidity network model according to any one of claims 1 to 9 is implemented.

Images