Construction method of prediction model of recurrence probability of specific disease

By constructing a disease recurrence probability prediction model based on convolutional neural network and gray correlation, the subjectivity and multi-factor interaction neglect of disease recurrence prediction in the prior art are solved, and high accuracy and personalized analysis are achieved, which is suitable for Predictions of recurrence probability for different diseases.

CN118919053APending Publication Date: 2024-11-08CANCER INST & HOSPITAL CHINESE ACADEMY OF MEDICAL SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410933156.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing methods for predicting the probability of disease recurrence rely on the clinical experience of doctors and traditional statistics. There are subjective differences and the ignorance of multifactorial interactions, and it is impossible to fully consider individual differences and dynamic changes in patients.

Method used

By collecting patient medical data, using convolutional neural network to extract features, combining deviation amount and gray correlation analysis, a prediction model for recurrence probability of specific diseases is constructed, including random forest algorithm, gray symbiosis matrix and Cox proportional hazards algorithm, and optimize the model to improve the Prediction accuracy.

Benefits of technology

It improves the accuracy and accuracy of disease recurrence probability prediction, realizes personalized analysis, adapts to the needs of disease recurrence probability prediction of different standards, is universal, saves resources, and improves work efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118919053A_ABST
    Figure CN118919053A_ABST
Patent Text Reader

Abstract

The invention discloses a method for constructing a prediction model for the recurrence probability of a specific disease, and the method comprises the steps: collecting the medical data of a patient with the specific disease and a personalized treatment scheme, carrying out the preprocessing of the medical data of the patient, carrying out the feature extraction of the medical data of the patient through a contrast ratio, and obtaining a baseline feature; performing time sequence analysis on the baseline characteristics by adopting a deviation value to obtain time sequence data, calculating a personalized risk index according to the time sequence data, performing influence analysis on the personalized treatment scheme according to the time sequence data to obtain a prognosis factor, performing risk stratification on the personalized risk index to obtain a level risk, and performing risk assessment on the level risk. And constructing a specific disease recurrence probability prediction model according to the prognosis factors and the level risk, optimizing the specific disease recurrence probability prediction model, and outputting a target model. The method not only can improve the construction precision of the recurrence probability prediction model, but also has good interpretability, and can be directly applied to a disease recurrence probability prediction system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of prediction, and in particular to a method for constructing a prediction model for the recurrence probability of a specific disease. Background Art

[0002] With the continuous deepening of medical research and the rapid development of medical technology, the prediction of the recurrence probability of specific diseases has become an important research direction in the medical field. The recurrence of diseases not only brings physical pain and economic burden to patients, but also brings great challenges to the medical system. Therefore, the development of an accurate and reliable prediction model for the recurrence probability of specific diseases is of great practical significance for improving the treatment effect of diseases, reducing the risk of recurrence, and optimizing the allocation of medical resources.

[0003] Disease recurrence refers to the phenomenon that disease symptoms reappear or disease status worsens again after the patient receives treatment. Many diseases, such as cancer, cardiovascular disease, and mental illness, have a high risk of recurrence. The recurrence of these diseases is often related to a variety of factors, including the patient's genetic background, lifestyle, environmental factors, and treatment regimen. Therefore, accurately predicting the probability of disease recurrence is of great significance for developing personalized treatment plans, improving treatment outcomes, and reducing the risk of recurrence.

[0004] At present, the prediction of the probability of recurrence of a specific disease mainly relies on the doctor's clinical experience, the patient's medical history information and some traditional statistical methods. However, these methods have many limitations. First, the doctor's clinical experience is affected by subjective factors, and the prediction results of different doctors for the same patient may vary greatly. Second, traditional statistical methods can often only analyze the relationship between a single factor and disease recurrence, while ignoring the interaction between multiple factors. In addition, these methods cannot fully consider the individual differences and dynamic changes of patients. Summary of the invention

[0005] The purpose of the present invention is to provide a method for constructing a prediction model for the recurrence probability of a specific disease.

[0006] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0007] The present invention comprises the following steps:

[0008] Collecting medical data and personalized treatment plans of patients with specific diseases, preprocessing the medical data of the patients, and extracting features of the medical data of the patients through contrast to obtain baseline features;

[0009] Performing time series analysis on the baseline characteristics using the deviation amount to obtain time series data, and calculating a personalized risk index based on the time series data;

[0010] Performing an impact analysis on the personalized treatment plan according to the time series data to obtain prognostic factors, and performing risk stratification on the personalized risk index to obtain graded risk;

[0011] A specific disease recurrence probability prediction model is constructed according to the prognostic factors and the grade risk, the specific disease recurrence probability prediction model is optimized, and a target model is output.

[0012] Furthermore, the method of extracting features from the patient medical data by contrast to obtain baseline features includes:

[0013] The patient medical data is input into the feature extraction model, and the convolutional neural network is used to extract the initial features of the patient medical data. Clustering is used to reduce the dimension of the initial features. The expression is:

[0014]

[0015] The center value of the i-th dimension of the y-th cluster is b y (i) The value of the i-th dimension of the a-th initial feature is g a (i), the ath initial feature is g a , the yth cluster is l y , initial feature g a Belongs to cluster l y The distance is l y (g a ), the dimension reduction feature of the i-th cluster is

[0016] Improved distance, expressed as:

[0017]

[0018] The improved distance is Transposed to T, the bias of the y-th cluster is h y , the weight of the y-th cluster is u y , the error constant is σ;

[0019] Update the feature, the expression is:

[0020]

[0021] The updated dimension reduction feature is The network structure is used to extract shallow and deep features of the dimensionality reduction features and calculate the descriptor distance:

[0022] ε=μ1ε υ +μ2ε s

[0023] The shallow feature distance is ε υ, the deep feature distance is ε s , the shallow layer weight is μ1, the deep layer weight is μ2, and the descriptor distance is ε;

[0024] Calculate the contrast between feature points:

[0025]

[0026] The feature point on the dth image is g d , the feature point on the pth image is g p , the xth feature point g d The region descriptor on the image is The xth feature point g p The region descriptor on the image is The mth feature point g d The region descriptor on the image is The mth feature point g p The region descriptor on the image is The xth feature point g p The local descriptor on the image is The xth feature point g d The local descriptor on the image is The L2 norm function is ‖·‖2, and the feature point g d and feature point g p The comparison degree is εi(g d ,g p ), feature point g d and feature point g p The similarity is φ(g d ,g p ), feature point g d and feature point g p The suppression factor is ρ d,p , the number of feature points is n;

[0027] Feature points with contrast less than 0.357 are output as baseline features.

[0028] Furthermore, the method of using the deviation to perform time series analysis on the baseline characteristics to obtain time series data includes:

[0029] The convolutional neural network is used to extract the temporal pattern of baseline features, which is expressed as:

[0030]

[0031] The length of the sliding window for extracting information is The baseline feature K time pattern vector of the f-th convolution kernel in the j-th dimension is The upper limit of time λ is M, and the jth dimension is The temporal pattern vector of a convolution kernel is The jth dimension The baseline characteristics of the convolution kernel are

[0032] Compute the correlation between the time pattern and the time state:

[0033]

[0034] The weight matrix is ​​ξ ι , the hidden state at time λ is Transposed to T, the temporal pattern vector of the xth baseline feature is Time Mode With hidden state The correlation is

[0035] Calculate the deviation:

[0036]

[0037] The activation function is sigmoid(·), and the number of baseline features is The hidden state at time λ-1 is Time Mode With hidden state The correlation is The deviation in the temporal pattern of the xth baseline characteristic is R λ ;

[0038] Output the time patterns with deviations greater than 0.279 as time series data.

[0039] Furthermore, the method for calculating the personalized risk index according to the time series data includes:

[0040]

[0041] Among them The patient's personalized risk index is The observation time is No. patient no. The forward heart rate difference at time No. patient no. The deviation at time is No. The heritability of the patients is No. patient no. The adjustment factor for time is

[0042] Furthermore, a method for obtaining prognostic factors by performing an impact analysis on the personalized treatment plan according to the time series data includes:

[0043] The improvement degree of the preset disease is obtained according to the personalized treatment plan, and the correlation degree between the time series data and the improvement degree is calculated using grey correlation analysis;

[0044] The grey correlation degree is calculated according to the correlation degree, and the time series data with a grey correlation degree greater than 0.74 are output as individual prognostic factors;

[0045] The individual prognostic factors of multiple patients are statistically analyzed, and the individual prognostic factors with a probability greater than 73% are output as prognostic factors.

[0046] Furthermore, a method for performing risk stratification on the personalized risk index to obtain a graded risk includes:

[0047] A polygenic risk score is used to obtain an individual's susceptibility to a preset disease. A support vector machine performs risk stratification based on the boundary between the personalized risk index and susceptibility, and outputs a graded risk.

[0048] Furthermore, a method for constructing a prediction model for the probability of recurrence of a specific disease based on the prognostic factors and the grade risk includes:

[0049] The prediction models for the probability of recurrence of a specific disease include the random forest algorithm, gray-level co-occurrence matrix, template matching algorithm, and Cox proportional hazards algorithm;

[0050] The random forest algorithm divides the input data into training and test sets in a ratio of 4:1;

[0051] The gray-level co-occurrence matrix extracts the change features by analyzing the gray relationship between pixels and neighboring pixels in the training set;

[0052] The template matching algorithm screens out key features through the correlation between the change characteristics and prognostic factors;

[0053] The Cox proportional hazards algorithm obtains the probability of occurrence through key features and evaluates the patient's risk ratio based on the probability of occurrence;

[0054] According to the position of the risk ratio in the hierarchical risk, the recurrence probability of the recurrence impact is output, and the recurrence probability prediction model of the specific disease is tested by a testing machine. The parameters of the recurrence probability prediction model of the specific disease are adjusted according to the variance of the predicted recurrence probability and the actual recurrence probability.

[0055] Furthermore, the method for optimizing the specific disease recurrence probability prediction model comprises:

[0056] Calculate the fitness of the searcher:

[0057]

[0058] The fitness function of the w-th searcher is The penalty factor is θ, The number of treatment responses is The regularization value of the w-th searcher is H(w), and the The performance of the w-th searcher for treatment response is

[0059] When there is an obstruction in front, update the search subposition, the expression is:

[0060]

[0061] The position of the w-th searcher at iteration t+1 is The position of the w-th searcher at iteration t is The position of the w-th searcher at iteration t-1 is The global worst position is The natural coefficient is θ, and the deflection coefficient is The regulatory constant is δ, the heritability coefficient is ρ, and the natural constant is e;

[0062] When there is no obstruction ahead, update the search subposition, the expression is:

[0063]

[0064] The deflection angle is Update the position of the cross searcher, the expression is:

[0065]

[0066] The upper limit of the cross search is L2, the lower limit of the cross search is L1, and the maximum number of iterations is t max , the position of the w-th cross searcher at the t+1th iteration is The position of the w-th cross searcher at the t-th iteration is The first cross weight is v1, and the second cross weight is v2;

[0067] Update the position of the young searcher, the expression is:

[0068]

[0069] The position of the w-th young searcher at the t+1th iteration is c w (t+1), the position of the w-th young searcher at the t-th iteration is c w (t), the global optimal position is The random numbers from 0 to 1 are r1 and r2 respectively;

[0070] Update the position of the intrusion searcher, the expression is:

[0071]

[0072] The position of the w-th intruder in the t+1th iteration is The position of the w-th intruder at the t-th iteration is The best position for the intrusion searcher is The normal random variable is

[0073] Calculate the covariance of the search sub-positions and construct a covariance matrix based on the covariance;

[0074] When the covariance matrix is ​​less than the minimum error, the global optimal position and the fitness of the searcher are updated until the maximum number of iterations is reached, otherwise the covariance of the searcher position is recalculated.

[0075] In a second aspect, an embodiment of the present application further provides an electronic device, including:

[0076] A processor; and a memory arranged to store computer executable instructions, which when executed cause the processor to perform the method steps described in the first aspect.

[0077] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple applications, the electronic device executes the method steps described in the first aspect.

[0078] The beneficial effects of the present invention are:

[0079] The present invention is a method for constructing a prediction model for the recurrence probability of a specific disease. Compared with the prior art, the present invention has the following technical effects:

[0080] The present invention can improve the accuracy of prediction of the recurrence probability of a specific disease through preprocessing, feature extraction, time series analysis, calculation of personalized risk index, impact analysis, risk stratification, model construction and model optimization steps, thereby improving the accuracy of prediction of the recurrence probability of a specific disease, and optimizing the construction of a prediction model for the recurrence probability of a specific disease, which can greatly save resources and improve work efficiency, and can realize automatic prediction of the recurrence probability of a specific disease, and perform personalized analysis on the construction of a prediction model for the recurrence probability of a specific disease in real time, which is of great significance for the prediction of the recurrence probability of a specific disease, and can adapt to the construction of prediction models for the recurrence probability of specific diseases of different standards and the prediction needs of the recurrence probability of different specific diseases, and has a certain universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 A flowchart of the steps of the method for constructing a prediction model for the probability of recurrence of a specific disease of the present invention;

[0082] Figure 2 It is a schematic diagram of the structure of an electronic device in an embodiment of this specification. DETAILED DESCRIPTION

[0083] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0084] The method for constructing a prediction model for the recurrence probability of a specific disease of the present invention comprises the following steps:

[0085] like Figure 1 As shown, in this embodiment, the following steps are included:

[0086] Collecting medical data and personalized treatment plans of patients with specific diseases, preprocessing the medical data of the patients, and extracting features of the medical data of the patients through contrast to obtain baseline features;

[0087] In the actual evaluation, patient medical data and personalized treatment plans of 2148 non-metastatic NPC patients were collected, among which high-risk stage III patients were recruited and randomly assigned to receive capecitabine combined with AC;

[0088] Baseline characteristics were: longitudinal cfEBV-DNA and tumor regression during pretreatment, post-IC, mid-RT, and post-RT;

[0089] Performing time series analysis on the baseline characteristics using the deviation amount to obtain time series data, and calculating a personalized risk index based on the time series data;

[0090] In the actual evaluation, the time series data were cfEBV-DNA, central lymph node necrosis status, T stage, N stage, tumor treatment regression, survival, age, and lactate dehydrogenase;

[0091] Performing an impact analysis on the personalized treatment plan according to the time series data to obtain prognostic factors, and performing risk stratification on the personalized risk index to obtain graded risk;

[0092] In the actual evaluation, the prognostic factors were cfEBV-DNA, central lymph node necrosis status, T stage, N stage, tumor treatment regression, and survival period; when the personalized risk index was less than 85%, it was low risk, and when the personalized risk index was greater than 85%, it was high risk. 20.8% of the selected patients showed clinical low-risk stage I-II, and 47.1% of the patients showed clinical high-risk stage IVA;

[0093] A specific disease recurrence probability prediction model is constructed according to the prognostic factors and the grade risk, the specific disease recurrence probability prediction model is optimized, and a target model is output.

[0094] In this embodiment, the method for extracting features from the patient medical data by contrast to obtain baseline features includes:

[0095] The patient medical data is input into the feature extraction model, and the convolutional neural network is used to extract the initial features of the patient medical data. Clustering is used to reduce the dimension of the initial features. The expression is:

[0096]

[0097] The center value of the i-th dimension of the y-th cluster is b y (i) The value of the i-th dimension of the a-th initial feature is g a (i), the ath initial feature is g a , the yth cluster is l y , initial feature g a Belongs to cluster l y The distance is l y (g a ), the dimension reduction feature of the i-th cluster is

[0098] Improved distance, expressed as:

[0099]

[0100] The improved distance is Transposed to T, the bias of the y-th cluster is h y , the weight of the y-th cluster is u y , the error constant is σ;

[0101] Update the feature, the expression is:

[0102]

[0103] The updated dimension reduction feature is The network structure is used to extract shallow and deep features of the dimensionality reduction features and calculate the descriptor distance:

[0104] ε=μ1ε υ +μ2ε s

[0105] The shallow feature distance is ε υ , the deep feature distance is ε s , the shallow layer weight is μ1, the deep layer weight is μ2, and the descriptor distance is ε;

[0106] Calculate the contrast between feature points:

[0107]

[0108] The feature point on the dth image is g d , the feature point on the pth image is g p , the xth feature point g d The region descriptor on the image is The xth feature point g p The region descriptor on the image is The mth feature point g d The region descriptor on the image is The mth feature point g p The region descriptor on the image is The xth feature point g p The local descriptor on the image is The xth feature point g d The local descriptor on the image is The L2 norm function is ‖·‖2, and the feature point g d and feature point g p The comparison degree is εi(g d ,g p ), feature point g d and feature point g p The similarity is φ(g d ,g p ), feature point g d and feature point g p The suppression factor is ρ d,p , the number of feature points is n;

[0109] Feature points with contrast less than 0.357 are output as baseline features.

[0110] In this embodiment, the method of using the deviation to perform time series analysis on the baseline feature to obtain time series data includes:

[0111] The convolutional neural network is used to extract the temporal pattern of baseline features, which is expressed as:

[0112]

[0113] The length of the sliding window for extracting information is The baseline feature K time pattern vector of the f-th convolution kernel in the j-th dimension is The upper limit of time λ is M, and the jth dimension is The temporal pattern vector of a convolution kernel is The jth dimension The baseline characteristics of the convolution kernel are

[0114] Compute the correlation between the time pattern and the time state:

[0115]

[0116] The weight matrix is ​​ξ ι , the hidden state at time λ is Transposed to T, the temporal pattern vector of the xth baseline feature is Time Mode With hidden state The correlation is

[0117] Calculate the deviation:

[0118]

[0119] The activation function is sigmoid(·), and the number of baseline features is The hidden state at time λ-1 is Time Mode With hidden state The correlation is The deviation in the temporal pattern of the xth baseline characteristic is R λ ;

[0120] Output the time patterns with deviations greater than 0.279 as time series data.

[0121] In this embodiment, the method for calculating the personalized risk index according to the time series data includes:

[0122]

[0123] Among them The patient's personalized risk index is The observation time is No. patient no. The forward heart rate difference at time No. patient no. The deviation at time is No. The heritability of the patients is No. patient no. The adjustment factor for time is

[0124] In this embodiment, the method for obtaining prognostic factors by performing an impact analysis on the personalized treatment plan according to the time series data includes:

[0125] The improvement degree of the preset disease is obtained according to the personalized treatment plan, and the correlation degree between the time series data and the improvement degree is calculated using grey correlation analysis;

[0126] The grey correlation degree is calculated according to the correlation degree, and the time series data with a grey correlation degree greater than 0.74 are output as individual prognostic factors;

[0127] The individual prognostic factors of multiple patients are statistically analyzed, and the individual prognostic factors with a probability greater than 73% are output as prognostic factors.

[0128] In this embodiment, the method of performing risk stratification on the personalized risk index to obtain a graded risk includes:

[0129] A polygenic risk score is used to obtain an individual's susceptibility to a preset disease. A support vector machine performs risk stratification based on the boundary between the personalized risk index and susceptibility, and outputs a graded risk.

[0130] In this embodiment, the method for constructing a prediction model for the probability of recurrence of a specific disease based on the prognostic factors and the grade risk includes:

[0131] The prediction models for the probability of recurrence of a specific disease include the random forest algorithm, gray-level co-occurrence matrix, template matching algorithm, and Cox proportional hazards algorithm;

[0132] The random forest algorithm divides the input data into training and test sets in a ratio of 4:1;

[0133] The gray-level co-occurrence matrix extracts the change features by analyzing the gray relationship between pixels and neighboring pixels in the training set;

[0134] The template matching algorithm screens out key features through the correlation between the change characteristics and prognostic factors;

[0135] The Cox proportional hazards algorithm obtains the probability of occurrence through key features and evaluates the patient's risk ratio based on the probability of occurrence;

[0136] According to the position of the risk ratio in the hierarchical risk, the recurrence probability of the recurrence impact is output, and the recurrence probability prediction model of the specific disease is tested by a testing machine. The parameters of the recurrence probability prediction model of the specific disease are adjusted according to the variance of the predicted recurrence probability and the actual recurrence probability.

[0137] In this embodiment, the method for optimizing the specific disease recurrence probability prediction model includes:

[0138] Calculate the fitness of the searcher:

[0139]

[0140] The fitness function of the w-th searcher is The penalty factor is θ, The number of treatment responses is The regularization value of the w-th searcher is H(w), and the The performance of the w-th searcher for treatment response is

[0141] When there is an obstruction in front, update the search subposition, the expression is:

[0142]

[0143] The position of the w-th searcher at iteration t+1 is The position of the w-th searcher at iteration t is The position of the w-th searcher at iteration t-1 is The global worst position is The natural coefficient is θ, and the deflection coefficient is The regulatory constant is δ, the heritability coefficient is ρ, and the natural constant is e;

[0144] When there is no obstruction ahead, update the search subposition, the expression is:

[0145]

[0146] The deflection angle is Update the position of the cross searcher, the expression is:

[0147]

[0148] The upper limit of the cross search is L2, the lower limit of the cross search is L1, and the maximum number of iterations is t max , the position of the w-th cross searcher at the t+1th iteration is The position of the w-th cross searcher at the t-th iteration is The first cross weight is v1, and the second cross weight is v2;

[0149] Update the position of the young searcher, the expression is:

[0150]

[0151] The position of the w-th young searcher at the t+1th iteration is c w (t+1), the position of the w-th young searcher at the t-th iteration is c w (t), the global optimal position is The random numbers from 0 to 1 are r1 and r2 respectively;

[0152] Update the position of the intrusion searcher, the expression is:

[0153]

[0154] The position of the w-th intruder in the t+1th iteration is The position of the w-th intruder at the t-th iteration is The best position for the intrusion searcher is The normal random variable is

[0155] Calculate the covariance of the search sub-positions and construct a covariance matrix based on the covariance;

[0156] When the covariance matrix is ​​less than the minimum error, the global optimal position and the fitness of the searcher are updated until the maximum number of iterations is reached, otherwise the covariance of the searcher position is recalculated.

[0157] Figure 2 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 2 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.

[0158] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0159] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0160] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a device for constructing a prediction model of the recurrence probability of a specific disease at a logical level. The processor executes the program stored in the memory and is specifically used to execute the method for constructing a prediction model of the recurrence probability of any specific disease mentioned above.

[0161] The above application Figure 1The method for constructing a prediction model of the probability of recurrence of a specific disease disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0162] The electronic device may also perform Figure 1 A method for constructing a prediction model for the recurrence probability of a specific disease in Figure 1 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0163] An embodiment of the present application also proposes a computer-readable storage medium, which stores one or more programs, and the one or more programs include instructions. When the instructions are executed by an electronic device including multiple applications, they execute the method for constructing a prediction model for the recurrence probability of any specific disease mentioned above.

[0164] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0165] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0168] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0169] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0170] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0171] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0172] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0173] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a prediction model for the probability of recurrence of a specific disease, characterized in that: The following steps are involved: Collecting medical data and personalized treatment plans of patients with specific diseases, preprocessing the medical data of the patients, and extracting features of the medical data of the patients through contrast to obtain baseline features; Performing time series analysis on the baseline characteristics using the deviation amount to obtain time series data, and calculating a personalized risk index based on the time series data; Performing an impact analysis on the personalized treatment plan according to the time series data to obtain prognostic factors, and performing risk stratification on the personalized risk index to obtain graded risk; A specific disease recurrence probability prediction model is constructed according to the prognostic factors and the grade risk, the specific disease recurrence probability prediction model is optimized, and a target model is output.

2. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method for extracting features from the patient medical data by contrast to obtain baseline features includes: The patient medical data is input into the feature extraction model, and the convolutional neural network is used to extract the initial features of the patient medical data. Clustering is used to reduce the dimension of the initial features. The expression is: The center value of the i-th dimension of the y-th cluster is b y (i) The value of the i-th dimension of the a-th initial feature is g a (i), the ath initial feature is g a , the yth cluster is l y , initial feature g a Belongs to cluster l y The distance is l y (g a ), the dimension reduction feature of the i-th cluster is Q iy ; Improved distance, expressed as: The improved distance is Transposed to T, the bias of the y-th cluster is h y , the weight of the y-th cluster is u y , the error constant is σ; Update the feature, the expression is: The updated dimension reduction feature is The network structure is used to extract shallow and deep features of the dimensionality reduction features and calculate the descriptor distance: e=μ1e υ +μ2ε s The shallow feature distance is ε υ , the deep feature distance is ε s , the shallow layer weight is μ1, the deep layer weight is μ2, and the descriptor distance is ε; Calculate the contrast between feature points: The feature point on the dth image is g d , the feature point on the pth image is g p , the xth feature point g d The region descriptor on the image is The xth feature point g p The region descriptor on the image is The mth feature point g d The region descriptor on the image is The mth feature point g p The region descriptor on the image is The xth feature point g p The local descriptor on the image is The xth feature point g d The local descriptor on the image is The L2 norm function is ‖·‖2, and the feature point g d and feature point g p The comparison degree is εi(g d ,g p ), feature point g d and feature point g p The similarity is φ(g d ,g p ), feature point g d and feature point g p The suppression factor is ρ d,p , the number of feature points is n; Feature points with contrast less than 0.357 are output as baseline features.

3. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method of using the deviation to perform time series analysis on the baseline characteristics to obtain time series data includes: The convolutional neural network is used to extract the temporal pattern of baseline features, which is expressed as: The length of the sliding window for extracting information is The baseline feature K time pattern vector of the f-th convolution kernel in the j-th dimension is The upper limit of time λ is M, and the jth dimension is The temporal pattern vector of a convolution kernel is The jth dimension The baseline characteristics of the convolution kernel are Compute the correlation between the time pattern and the time state: The weight matrix is ​​ξ ι , the hidden state at time λ is Transposed to T, the temporal pattern vector of the xth baseline feature is Time Mode With hidden state The correlation is Calculate the deviation: The activation function is sigmoid(·), the number of baseline features is a, and the hidden state at time λ-1 is Time Mode With hidden state The correlation is The deviation in the temporal pattern of the xth baseline characteristic is R λ ; Output the time patterns with deviations greater than 0.279 as time series data.

4. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method for calculating the personalized risk index according to the time series data includes: Among them The patient's personalized risk index is The observation time is d, patient no. The forward heart rate difference at time No. patient no. The deviation at time is No. The heritability of the patients is No. patient no. The adjustment factor for time is 5. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method for obtaining prognostic factors by performing an impact analysis on the personalized treatment plan according to the time series data comprises: The improvement degree of the preset disease is obtained according to the personalized treatment plan, and the correlation degree between the time series data and the improvement degree is calculated using grey correlation analysis; The grey correlation degree is calculated according to the correlation degree, and the time series data with a grey correlation degree greater than 0.74 are output as individual prognostic factors; The individual prognostic factors of multiple patients are statistically analyzed, and the individual prognostic factors with a probability greater than 73% are output as prognostic factors.

6. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method of performing risk stratification on the personalized risk index to obtain a graded risk includes: A polygenic risk score is used to obtain an individual's susceptibility to a preset disease. A support vector machine performs risk stratification based on the boundary between the personalized risk index and susceptibility, and outputs a graded risk.

7. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method for constructing a prediction model for the probability of recurrence of a specific disease according to the prognostic factors and the grade risk comprises: The prediction models for the probability of recurrence of a specific disease include the random forest algorithm, gray-level co-occurrence matrix, template matching algorithm, and Cox proportional hazards algorithm; The random forest algorithm divides the input data into training and test sets in a ratio of 4:1; The gray-level co-occurrence matrix extracts the change features by analyzing the gray relationship between pixels and neighboring pixels in the training set; The template matching algorithm screens out key features through the correlation between the change characteristics and prognostic factors; The Cox proportional hazards algorithm obtains the probability of occurrence through key features and evaluates the patient's risk ratio based on the probability of occurrence; According to the position of the risk ratio in the hierarchical risk, the recurrence probability of the recurrence impact is output, and the recurrence probability prediction model of the specific disease is tested by a testing machine. The parameters of the recurrence probability prediction model of the specific disease are adjusted according to the variance of the predicted recurrence probability and the actual recurrence probability.

8. The method for constructing a prediction model for the recurrence probability of a specific disease according to claim 1, characterized in that: The method for optimizing the specific disease recurrence probability prediction model comprises: Calculate the fitness of the searcher: The fitness function of the w-th searcher is The penalty factor is θ, and the number of treatment responses of the ith type is The regularization value of the w-th searcher is H(w), and the performance of the w-th searcher for the ith treatment response is F i (w); When there is an obstruction in front, update the search subposition, the expression is: The position of the w-th searcher at iteration t+1 is The position of the w-th searcher at iteration t is The position of the w-th searcher at iteration t-1 is The global worst position is The natural coefficient is θ, and the deflection coefficient is The regulatory constant is δ, the heritability coefficient is ρ, and the natural constant is e; When there is no obstruction ahead, update the search subposition, the expression is: The deflection angle is Update the position of the cross searcher, the expression is: The upper limit of the cross search is L2, the lower limit of the cross search is L1, and the maximum number of iterations is t max , the position of the w-th cross searcher at the t+1th iteration is The position of the w-th cross searcher at the t-th iteration is The first cross weight is v1, and the second cross weight is v2; Update the position of the young searcher, the expression is: The position of the w-th young searcher at the t+1th iteration is c w (t+1), the position of the w-th young searcher at the t-th iteration is c w (t), the global optimal position is The random numbers from 0 to 1 are r1 and r2 respectively; Update the position of the intrusion searcher, the expression is: The position of the w-th intruder in the t+1th iteration is The position of the w-th intruder at the t-th iteration is The best position for the intrusion searcher is The normal random variable is Calculate the covariance of the search sub-positions and construct a covariance matrix based on the covariance; When the covariance matrix is ​​less than the minimum error, the global optimal position and the fitness of the searcher are updated until the maximum number of iterations is reached, otherwise the covariance of the searcher position is recalculated.

9. An electronic device, comprising: processor; as well as A memory arranged to store computer executable instructions, which when executed cause the processor to perform the method according to any one of claims 1 to 8.

10. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, enables the electronic device to execute the method according to any one of claims 1 to 8.