A disease detection model training and detection method, device and electronic equipment

Abstract

The application discloses a disease detection model training method, a disease detection model training device, a disease detection model detection method, a disease detection model detection device and equipment, and relates to the technical field of disease detection model training and detection. The disease detection model training method comprises the following steps: obtaining training samples for representing a target disease, wherein the training samples comprise physical sign data of multiple persons and data labels indicating whether the persons are ill; training a preset multiple regression model by using the training samples, wherein the multiple regression model comprises linear parameters, the linear parameters are used for linear dimension reduction on the physical sign data of the persons, and the linear parameters are used for calculating results indicating whether the persons are ill; when a preset condition is met, outputting the multiple regression model, and constructing a disease detection model according to the multiple regression model. The technical scheme provided by the application improves the efficiency of disease identification of a target person based on physical sign data.

Description

Technical Field

[0001] This invention relates to the field of big data analysis, specifically to a disease detection model training, detection method, device, and equipment. Background Technology

[0002] With the development of big data technology, technologies for facial recognition and sentence classification using big data are emerging in various industries. Essentially, these technologies combine massive amounts of data with machine learning techniques to analyze the intrinsic differences of targets, achieving recognition and prediction capabilities that are impossible for the naked eye or other senses. In the medical field, patients typically undergo a series of examinations, including blood tests and chest X-rays, which are often inefficient and slow to produce results. This is especially true when dealing with infectious diseases, where large numbers of people need to be examined, making efficiency a crucial factor. To improve efficiency, medical staff usually conduct a preliminary assessment of individuals based on vital signs such as temperature before the actual examination. However, these assessments are often based on subjective human judgment and are therefore inaccurate. Subsequently, more and more researchers have begun to study the detection of target diseases using big data based on vital sign data. However, to improve accuracy, large-capacity neural networks are typically required for data training, meaning the detection process involves running the entire neural network on a single dataset, resulting in low efficiency. Therefore, improving the efficiency of medical staff in identifying diseases based on vital sign data while maintaining accuracy is a pressing issue. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a disease detection model training, detection method, apparatus and equipment, thereby improving the efficiency of disease identification of target personnel based on vital sign data.

[0004] According to a first aspect, the present invention provides a method for training a disease detection model, the method comprising: acquiring training samples for characterizing a target disease, the training samples including vital sign data of multiple individuals and data labels indicating whether each individual is ill; training a preset multiple regression model using the training samples, the multiple regression model including linear parameters, the linear parameters being used to linearly reduce the dimensionality of the individuals' vital sign data to calculate whether the individuals are ill; and when preset conditions are met, outputting the multiple regression model and constructing the disease detection model based on the multiple regression model.

[0005] Optionally, obtaining training samples for characterizing the target disease includes: obtaining multiple first sign data labeled with disease and using the first sign data as positive samples; obtaining multiple second sign data labeled with no disease and using the second sign data as negative samples; standardizing each positive and negative sample at the same sampling frequency; and forming a sample set from the standardized positive and negative samples to generate the training samples.

[0006] Optionally, before training the preset multiple regression model, the method further includes: using the sum of a first variable and a first noise parameter as a first linear expression, where the first variable is the product of the vital sign data and a preset first linear parameter, and the first noise parameter is used to characterize the noise of the vital sign data; using the sum of a second variable and a second noise parameter as a second linear expression, where the second variable is the product of the output of the first linear expression and a preset second linear parameter, and the second noise parameter is used to characterize the noise of the output of the first linear expression, and the output of the second linear expression is used to predict whether the person corresponding to the vital sign data is ill, wherein the preset first linear parameter and the preset second linear parameter are the linear parameters of the multiple regression model; and combining the first linear expression and the second linear expression to form the multiple regression model.

[0007] Optionally, training the preset multiple regression model using the training samples includes: transforming the first linear expression and the second linear expression into forms representing the first noise parameter and the second noise parameter, respectively; establishing an objective function based on the transformed first linear expression and the second linear expression; substituting the vital sign data and the data labels into the objective function for iterative calculation, adjusting the preset first linear parameter and the preset second linear parameter to decrease the norm of the first noise parameter and the norm of the second noise parameter; outputting at least one set of adjusted linear parameters, and generating at least one corresponding multiple regression model based on the adjusted linear parameters to complete the training.

[0008] Optionally, constructing the disease detection model based on the multiple regression model includes: if multiple multiple regression models are obtained by training a preset multiple regression model using the training samples, then the expression for calculating the average value of the output results of the multiple multiple regression models is used as the disease detection model.

[0009] According to a second aspect, the present invention provides a method for detecting a disease, the method comprising: acquiring vital sign data of a target person; substituting the vital sign data into a disease detection model generated according to any optional embodiment of the first aspect for calculation; and determining whether the target person suffers from the target disease based on the output result of the disease detection model.

[0010] According to a third aspect, the present invention provides a disease detection model training device, the device comprising: a data acquisition module for acquiring training samples for characterizing a target disease, the training samples including vital sign data of multiple individuals and data labels indicating whether each individual is diseased; a model training module for training a preset multiple regression model using the training samples, the multiple regression model including linear parameters, the linear parameters being used to linearly reduce the dimensionality of the individuals' vital sign data to calculate whether the individuals are diseased; and a model output module for outputting the multiple regression model when preset conditions are met, and constructing the disease detection model based on the multiple regression model.

[0011] According to a fourth aspect, the present invention provides a disease detection device, the device comprising: a second data acquisition module for acquiring vital sign data of a target person; a calculation module for substituting the vital sign data into a disease detection model generated according to any optional embodiment of the first aspect for calculation; and a result output module for determining whether the target person suffers from the target disease based on the output result of the disease detection model.

[0012] According to a fifth aspect, the present invention provides an electronic device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the methods described in the first aspect, the second aspect, or any optional embodiment of the first aspect and the second aspect.

[0013] According to a sixth aspect, embodiments of the present invention provide a computer-readable storage medium storing computer instructions for causing the computer to perform the methods described in the first aspect, the second aspect, or any optional embodiments of the first and second aspects.

[0014] The technical solution provided in this application has the following advantages:

[0015] The technical solution provided in this application acquires a large amount of vital sign data, including but not limited to body temperature data, blood oxygen saturation data, and heart rate data, as training samples. Then, based on the measurable data noise from the collected data, a multivariate regression model is established to reduce multidimensional data to one-dimensional recognition results through linear parameters. This aims to reduce the data noise in body temperature data, blood oxygen saturation data, and heart rate data. The linear parameters of the multivariate regression model are trained using the training samples. After the linear parameters are trained, the multivariate regression model based on simple linear calculations can be used to calculate the vital sign data of a target individual, thereby determining whether the individual is ill. In practical use, the computational load is far less than that of large-capacity neural network and other machine learning models, significantly improving diagnostic efficiency as a preliminary diagnostic model. Attached Figure Description

[0016] The features and advantages of the invention will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the invention in any way. In the drawings:

[0017] Figure 1 This diagram illustrates the steps of a disease detection model training method according to one embodiment of the present invention.

[0018] Figure 2 This diagram illustrates the steps of a disease detection method according to one embodiment of the present invention.

[0019] Figure 3 This diagram illustrates the structure of a disease detection model training device according to one embodiment of the present invention.

[0020] Figure 4 A schematic diagram of a disease detection device according to one embodiment of the present invention is shown;

[0021] Figure 5 A schematic diagram of an electronic device according to one embodiment of the present invention is shown. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1 In one embodiment, a method for training a disease detection model specifically includes the following steps:

[0024] Step S101: Obtain training samples to characterize the target disease. The training samples include vital sign data of multiple individuals and data labels indicating whether each individual is ill. Specifically, in this embodiment, at least a large amount of vital sign data of individuals, including but not limited to body temperature data, blood oxygen saturation data, heart rate data, and labels indicating whether each individual is ill (e.g., a label of "1" for those who are ill and a label of "0" for those who are not ill), are used as training samples for multiple dimensions of vital signs. This allows the training of the big data analysis model established in subsequent steps to consider the external manifestations of the target individuals' disease from multiple perspectives, thereby improving the training accuracy of the big data analysis model.

[0025] Step S102: Train the preset multiple regression model using training samples. The multiple regression model includes linear parameters, which are used to linearly reduce the dimensionality of the person's vital signs data to calculate whether the person is ill.

[0026] Step S103: When the preset conditions are met, output the multiple regression model and construct the disease detection model based on the multiple regression model.

[0027] Specifically, considering the capacity (number of layers and neurons) of existing neural network models, which are typically set to be quite large to meet certain accuracy requirements, the actual model calculation process during detection can essentially be regarded as a training process without adjusting model parameters, resulting in long computation time. This severely impacts detection efficiency when a large number of people need to undergo vital sign detection. Therefore, in this embodiment of the invention, based on the goal of transforming multidimensional vital sign data into a one-dimensional result of whether or not a person is ill, a multivariate regression model is established. Linear multivariate regression (multidimensional parameters in matrix form) is achieved by setting linear parameters for dimensionality reduction. Essentially, through linear calculation, it realizes the function of obtaining a one-dimensional result from multidimensional data. To ensure appropriate selection of linear parameters, this embodiment optimizes the model parameters by incorporating easily detectable data noise in various types of vital sign data. For example, an optimization objective function is established based on the goal of minimizing noise and maximizing the amount of reasonable data in each type of data. Thus, through the above steps, a machine learning model based on the multivariate regression algorithm is established, making subsequent methods for detecting human diseases based on vital sign data essentially simple linear calculations, which can significantly improve disease detection efficiency. In this embodiment, the preset conditions that the training process can satisfy include, but are not limited to: ending training when the number of training iterations reaches a preset number; and ending training when the linear parameters of the multiple regression model tend to stabilize and their change is less than a preset threshold. In this embodiment, the trained multiple regression model is used as a disease detection model, thus enabling the detection of the target disease based on the multiple regression model.

[0028] Specifically, in one embodiment, step S101 above includes the following steps:

[0029] Step 1: Obtain multiple primary vital sign data labeled with disease status, and use the primary vital sign data as positive samples.

[0030] Step 2: Obtain multiple secondary vital sign data labeled as "not diseased" and use the secondary vital sign data as negative samples.

[0031] Step 3: Standardize each positive and negative sample using the same sampling frequency.

[0032] Step 4: Combine the standardized positive and negative samples into a sample set to generate training samples.

[0033] Specifically, the collection of various vital sign data may result in different data sparsity due to varying sampling frequencies. Without processing, data with different sparsities will not produce accurate training results. Therefore, in this embodiment, the sampling frequency is standardized for multi-dimensional sampling data. The standardized positive and negative samples are then used to train the multivariate regression model in both positive and negative directions, resulting in higher accuracy. For example, vital sign data includes the following three dimensions: 1. Minute-level body temperature measured by a thermometer based on a wearable device x t 2. Based on the ten-minute heart rate measured by a smart bracelet h 3. Hourly blood oxygen saturation measured by a smart bracelet b Since the three dimensions of vital signs data are sampled at different frequencies, an averaging method is used for data standardization. We use hourly data for a unified data collection standard. Therefore, hourly body temperature is the average body temperature over the past sixty minutes, denoted as x. T Hourly heart rate is the average heart rate over the past sixty minutes, denoted as x. H After standardizing the sampled data, we have vital sign data samples, using a specific hour as the data unit.

[0034] x = [x T ;x H ;x b ]

[0035] In other words, the vital signs data samples are three-dimensional vectors collected every hour. Expanding the vital signs data according to a time series, we obtain the vital signs data sample set, D = {x1, x2, ...}.

[0036] Specifically, in one embodiment, before step S102 above, the following step is also included:

[0037] Step 5: Use the sum of the first variable and the first noise parameter as the first linear expression. The first variable is the product of the vital signs data and the preset first linear parameter. The first noise parameter is used to characterize the noise of the vital signs data.

[0038] Step Six: The sum of the second variable and the second noise parameter is used as the second linear expression. The second variable is the product of the output of the first linear expression and the preset second linear parameter. The second noise parameter is used to characterize the noise of the output of the first linear expression. The output of the second linear expression is used to predict whether the person corresponding to the vital signs data is sick. The preset first linear parameter and the preset second linear parameter are the linear parameters of the multiple regression model.

[0039] Step 7: Combine the first linear expression and the second linear expression to form a multiple regression model.

[0040] Specifically, the multiple regression model expressions for steps five through seven above are shown below:

[0041] Y = AX + ∈ Y

[0042] y=YB+∈ y

[0043] Where Y∈R 1×3 Let X ∈ R be the latent variable (i.e., the output of the first linear expression). W×3 Let A ∈ R be the input training sample (i.e., vital sign data). 1×W For the first linear parameter, ∈ Y ∈R 1×3 The noise parameters for body temperature, blood oxygen saturation, and heart rate data (i.e., the first noise parameter). y∈R, representing the output result indicating whether a person is ill (i.e., the output result of the second linear expression), B∈R. 3×1 For the second linear parameter, ∈ y ∈R represents the noise parameter (i.e., the second noise parameter) used to represent the output after the first linear calculation, AX is the first variable, and YB is the second variable.

[0044] Specifically, dimensionality reduction of multidimensional vital sign data X in matrix form typically employs linear processing by multiplying it by the corresponding dimensionality reduction matrix (i.e., the first linear parameter A and the second linear parameter B), i.e., y = AXB, thereby transforming the high-dimensional vital sign data into a one-dimensional output y. The actual collected vital sign data X contains data noise, which can be calculated using, but is not limited to, the variance formula. If we assume the existence of linear parameters A and B that can extract useful components from the vital sign data X, ensuring that the dimensionality-reduced one-dimensional data contains the most useful components and the fewest noise components, then after two linear processing steps, the noise components at input and output can be expressed as: ∈ Y =Y-AX and ∈ y =y-YB, which is the component remaining after multiplying the training sample by the linear parameter to extract the useful component. By minimizing noise, the linear parameters A and B can be optimized, thereby enabling machine learning training of the multivariate regression model to find the most suitable linear parameters. By inputting multi-sign data, a binary classification result of whether the target person is sick can be calculated.

[0045] Specifically, in one embodiment, based on steps five to seven above, step S102 specifically includes the following steps:

[0046] Step 8: Transform the first linear expression and the second linear expression into forms representing the first noise parameter and the second noise parameter, respectively;

[0047] Step 9: Establish the objective function based on the transformed first and second linear expressions;

[0048] Step 10: Substitute the vital signs data and data labels into the objective function for iterative calculation, and adjust the preset first linear parameter and the preset second linear parameter to reduce the norm of the first noise parameter and the norm of the second noise parameter.

[0049] Step 11: Output at least one set of adjusted linear parameters, and generate at least one corresponding multiple regression model based on the adjusted linear parameters to complete the training.

[0050] Specifically, the objective function expressions for steps eight and nine above are as follows:

[0051]

[0052] Where λ∈(0,1) is an artificially adjustable hyperparameter, N is the number of negative samples, M is the number of positive samples, the output y of a positive sample is 1, indicating disease, the output y of a negative sample is 0, indicating no disease, ||Y-AX||2 is the norm of the first noise parameter, and ||y-YB||2 is the norm of the second noise parameter.

[0053] In this embodiment, the optimization of the first linear parameter and the second linear parameter is achieved through the above-mentioned objective function. The specific calculation method of the above formula is as follows: First, the latent variable Y and the second linear parameter B are randomly assigned values. Since the actual output y of the training sample is known (i.e., the data label), values ​​can be assigned to ∈ y =y-YB is calculated, and then the value of ∈ is obtained using the randomly assigned latent variable and the randomly assigned first linear parameter A. Y =Y-AX, where the vital signs data X is known, and since the actual noise corresponding to the vital signs data ∈ Y Since the input noise components are actually measurable, the error between the actually measured input noise components and the calculated input noise components is used to determine whether the randomly selected first linear parameter and latent variable are suitable. If the error is large (e.g., the error is higher than a preset threshold), the first linear parameter and latent variable are not suitable. If the error is small (e.g., the error is lower than a preset threshold), the current first linear parameter, latent variable, and second linear parameter are considered acceptable. At this time, the current first linear parameter and second linear parameter are recorded, along with the current two data noises. Multiple iterative calculations are then performed alternately, recording the data noise each time. Finally, the two data noises with the smallest noise levels among all data noise cases are selected, thus obtaining the first linear parameter and second linear parameter corresponding to the current data noise, ultimately optimizing the linear parameters. Then, the optimized linear parameters are substituted into the noise-ignoring model y = AXB to achieve the function of obtaining a one-dimensional diagnostic result based on high-dimensional vital sign data. Furthermore, the current first linear parameter and second linear parameter can extract the useful components from the vital sign data to the maximum extent while ignoring noise components to the maximum extent. For example, the target person's vital signs data are input into a trained multiple regression model y = AXB, and the output is y = 0.7. In this embodiment, if the output is greater than a preset threshold of 0.5, the target person is considered to have the target disease; if the output is less than the preset threshold of 0.5, the target person is considered not to have the target disease. Through the above steps, the goal of judging the target person's disease based on vital signs data can be quickly calculated and achieved, greatly improving the efficiency of medical staff in the initial detection of whether a target person is ill.

[0054] Specifically, although the objective function in steps eight to nine can obtain a set of linear parameters that meet the objective function conditions at the end of one round of training through random assignment and cross-iteration, the linear parameters in the objective function are randomly assigned. Even for the same training samples, multiple rounds of training may yield multiple sets of different linear parameters that meet the constraints of the objective function. Essentially, the different linear parameters have different emphases. For example, the linear parameters in the first training may emphasize body temperature data more, while the linear parameters in the second training may emphasize blood oxygen saturation concentration data more. Therefore, in this embodiment, steps ten and eleven can generate multiple sets of linear parameters through multiple rounds of training to further improve the prediction accuracy of the multiple regression model. Then, the obtained multiple sets of linear parameters are substituted into the multiple regression model to obtain multiple trained models with different emphases. After that, the vital sign data are input into each model to obtain multiple prediction results. Then, the prediction results are averaged and combined to obtain the final prediction result. This allows the disease prediction based on vital sign data to consider the influence of data from various dimensions to a greater extent, making the prediction probability of whether the target person is sick more accurate.

[0055] Specifically, in one embodiment, based on steps eight to eleven above, step S103 specifically includes the following steps:

[0056] Step 12: If multiple multiple regression models are obtained by training the preset multiple regression model using training samples, the expression for calculating the average of the output results of the multiple multiple regression models is used as the disease detection model. Specifically, in this embodiment, the disease detection model does not directly use a single multiple regression model, but rather calculates the average of the output values ​​of multiple multiple regression models. A detailed description of the principle behind this step can be found in the descriptions of steps 10 and 11 above, and will not be repeated here.

[0057] Please see Figure 2 In one embodiment, a disease detection method specifically includes the following steps:

[0058] Step S201: Obtain the vital signs data of the target personnel.

[0059] Step S202: Substitute the vital signs data into the disease detection model generated by the above training method for calculation.

[0060] Step S203: Determine whether the target person has the target disease based on the output of the disease detection model.

[0061] Specifically, the detailed principles of identifying target diseases based on vital sign data can be found in the relevant descriptions of the above-mentioned disease detection model training method embodiments, and will not be repeated here.

[0062] Through the above steps, the technical solution provided in this application acquires a large amount of vital sign data, including but not limited to body temperature data, blood oxygen saturation data, and heart rate data, as training samples. Then, based on the measurable data noise from the collected data, a multivariate regression model is established to reduce multidimensional data to one-dimensional recognition results through linear parameters. This aims to reduce the data noise in body temperature data, blood oxygen saturation data, and heart rate data. The linear parameters of the multivariate regression model are trained using the training samples. After the linear parameters are trained, the multivariate regression model based on simple linear calculations can be used to calculate the vital sign data of the target person, thereby determining whether the target person is ill. In practical use, the computational load is far less than that of a large-capacity neural network, and as a diagnostic model for preliminary diagnosis, it can greatly improve diagnostic efficiency.

[0063] like Figure 3 As shown, this embodiment also provides a disease detection model training device, the device comprising:

[0064] The data acquisition module 101 is used to acquire training samples to characterize the target disease. The training samples include vital sign data of multiple individuals and data labels indicating whether each individual has the disease. For details, please refer to the relevant description of step S101 in the above method embodiment, which will not be repeated here.

[0065] The model training module 102 is used to train a preset multiple regression model using training samples. The multiple regression model includes linear parameters, which are used to linearly reduce the dimensionality of the person's vital signs data to calculate whether the person is ill. For details, please refer to the relevant description of step S102 in the above method embodiment, which will not be repeated here.

[0066] The model output module 103 is used to output a multiple regression model when preset conditions are met, and to construct a disease detection model based on the multiple regression model. For details, please refer to the relevant description of step S103 in the above method embodiment, which will not be repeated here.

[0067] The present invention provides a disease detection model training device for executing a disease detection model training method provided in the above embodiments. Its implementation method and principle are the same. For details, please refer to the relevant description of the above method embodiments, which will not be repeated here.

[0068] like Figure 4 As shown, this embodiment also provides a disease detection device, which includes:

[0069] The second data acquisition module 201 acquires the vital signs data of the target personnel. For details, please refer to the relevant description of step S201 in the above method embodiment, which will not be repeated here.

[0070] The calculation module 202 is used to input the vital sign data into the disease detection model generated by the above-described disease detection model training method for calculation. For details, please refer to the relevant description of step S202 in the above method embodiment, which will not be repeated here.

[0071] The result output module 203 is used to determine whether the target person suffers from the target disease based on the output results of the disease detection model. For details, please refer to the relevant description of step S203 in the above method embodiment, which will not be repeated here.

[0072] The present invention provides a disease detection device for performing a disease detection method provided in the above embodiments. Its implementation method and principle are the same. For details, please refer to the relevant description of the above method embodiments, which will not be repeated here.

[0073] Through the collaborative efforts of the aforementioned components, the technical solution provided in this application acquires a large amount of vital sign data, including but not limited to body temperature data, blood oxygen saturation data, and heart rate data, as training samples. Then, based on the measurable noise in the collected data, a multivariate regression model is established to reduce multidimensional data to a one-dimensional identification result using linear parameters. This aims to reduce the noise in body temperature, blood oxygen saturation, and heart rate data. The linear parameters of the multivariate regression model are trained using the training samples. After the linear parameters are trained, the multivariate regression model based on simple linear calculations can be used to calculate the vital sign data of a target individual, thereby determining whether the individual is ill. In practical use, the computational load is far less than that of a large-capacity neural network, significantly improving diagnostic efficiency as a preliminary diagnostic model.

[0074] Figure 5 An electronic device according to an embodiment of the present invention is shown. The device includes a processor 901 and a memory 902, which can be connected via a bus or other means. Figure 5 Taking the bus connection between China and Israel as an example.

[0075] Processor 901 can be a Central Processing Unit (CPU). Processor 901 can 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, or combinations of the above types of chips.

[0076] The memory 902, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the methods in the above method embodiments. The processor 901 executes various functional applications and data processing of the processor by running the non-transitory software programs, instructions, and modules stored in the memory 902, thereby implementing the methods in the above method embodiments.

[0077] The memory 902 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor 901, etc. Furthermore, the memory 902 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory 902 may optionally include memory remotely located relative to the processor 901, and these remote memories may be connected to the processor 901 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0078] One or more modules are stored in memory 902, and when executed by processor 901, they perform the methods described in the above method embodiments.

[0079] The specific details of the aforementioned electronic device can be understood by referring to the relevant descriptions and effects in the above method embodiments, and will not be repeated here.

[0080] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The implemented program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium can also include combinations of the above types of memory.

[0081] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for training a disease detection model, characterized in that, The method includes: Obtain training samples to characterize the target disease, the training samples including vital sign data of multiple individuals and data labels indicating whether each individual has the disease; The sum of the first variable and the first noise parameter is used as the first linear expression. The first variable is the product of the vital signs data and the preset first linear parameter. The first noise parameter is used to characterize the noise of the vital signs data. The sum of the second variable and the second noise parameter is used as the second linear expression. The second variable is the product of the output of the first linear expression and the preset second linear parameter. The second noise parameter is used to characterize the noise of the output of the first linear expression. The output of the second linear expression is used to predict whether the person corresponding to the vital signs data is sick. The preset first linear parameter and the preset second linear parameter are the linear parameters of the multiple regression model. The linear parameters are used to perform linear dimensionality reduction on the person's vital signs data to calculate whether the person is sick. The first linear expression and the second linear expression are combined to form a multiple regression model; The preset multivariate regression model is trained using the training samples; When the preset conditions are met, the multivariate regression model is output, and the disease detection model is constructed based on the multivariate regression model.

2. The method according to claim 1, characterized in that, The acquisition of training samples for characterizing the target disease includes: Acquire multiple first sign data labeled with disease status, and use the first sign data as a positive sample; Acquire multiple second sign data labeled as "not diseased" and use the second sign data as negative samples; Standardize each positive and negative sample using the same sampling frequency; The standardized positive and negative samples are combined to form a sample set, which is used to generate the training samples.

3. The method according to claim 1, characterized in that, The step of training a preset multivariate regression model using the training samples includes: The first linear expression and the second linear expression are respectively transformed into forms representing the first noise parameter and the second noise parameter; The objective function is established based on the transformed first and second linear expressions. The vital signs data and the data labels are substituted into the objective function for iterative calculation. The preset first linear parameter and the preset second linear parameter are adjusted so that the norm of the first noise parameter and the norm of the second noise parameter decrease. Output at least one set of adjusted linear parameters, and generate at least one corresponding multiple regression model based on the adjusted linear parameters to complete the training.

4. The method according to claim 3, characterized in that, The construction of the disease detection model based on the multivariate regression model includes: If the training samples are used to train a preset multiple regression model to obtain multiple multiple regression models, then the expression for calculating the average value of the output results of the multiple multiple multiple regression models is used as the disease detection model.

5. A method for detecting a disease, characterized in that, The method includes: Obtain vital sign data of the target personnel; The vital sign data are substituted into the disease detection model generated by the method according to any one of claims 1-4 for calculation; Based on the output of the disease detection model, it is determined whether the target person suffers from the target disease.

6. A disease detection model training device, characterized in that, The device includes: The data acquisition module is used to acquire training samples for characterizing the target disease. The training samples include the vital signs data of multiple people and data labels on whether each person is ill. The model training module is used to: use the sum of a first variable and a first noise parameter as a first linear expression, where the first variable is the product of the vital sign data and a preset first linear parameter, and the first noise parameter is used to characterize the noise of the vital sign data; use the sum of a second variable and a second noise parameter as a second linear expression, where the second variable is the product of the output of the first linear expression and a preset second linear parameter, and the second noise parameter is used to characterize the noise of the output of the first linear expression; the output of the second linear expression is used to predict whether the person corresponding to the vital sign data is ill; the preset first linear parameter and the preset second linear parameter are linear parameters of a multiple regression model, and the linear parameters are used to perform linear dimensionality reduction on the person's vital sign data to calculate whether the person is ill; combine the first linear expression and the second linear expression to form a multiple regression model; and train the preset multiple regression model using the training samples, where the multiple regression model includes linear parameters, and the linear parameters are used to perform linear dimensionality reduction on the person's vital sign data to calculate whether the person is ill. The model output module is used to output the multivariate regression model when preset conditions are met, and to construct the disease detection model based on the multivariate regression model.

7. A disease detection device, characterized in that, The device includes: The second data acquisition module acquires the vital signs data of the target personnel; The calculation module is used to substitute the vital sign data into the disease detection model generated by the method according to any one of claims 1-4 for calculation; The result output module is used to determine whether the target person suffers from the target disease based on the output results of the disease detection model.

8. An electronic device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method as described in any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the method as described in any one of claims 1-5.

Images