Prediction method of sepsis, electronic equipment and medium

Abstract

The embodiment of the invention provides a sepsis prediction method, electronic equipment and a medium, and the method comprises the steps: inputting expression data corresponding to a plurality of different target genes of a patient into a sepsis prediction model, and obtaining a sepsis prediction result of the patient; wherein the sepsis prediction model is obtained through the following steps that a plurality of first training samples are obtained, and the first training samples comprise sample expression data of all target genes; for each first training sample, inputting the first training sample into a pre-trained analytical model to obtain a training prediction result of the first training sample and contribution relationship information between each piece of sample expression data in the first training sample and the training prediction result; and adjusting the initial sepsis prediction model according to the multiple pieces of contribution relationship information corresponding to the first training samples to obtain the sepsis prediction model. According to the embodiment of the invention, the accuracy and reliability of sepsis prediction can be improved.

Description

Technical Field

[0001] The embodiments of this application relate to, but are not limited to, the field of medical technology, and in particular to a method for predicting sepsis, an electronic device, and a medium. Background Technology

[0002] Sepsis is a heterogeneous syndrome caused by infection, characterized by complex pathogenesis and high mortality, placing a huge burden on families and society.

[0003] In related technologies, sepsis is associated with multiple genes, resulting in heterogeneity in the pathogenesis of each sepsis patient. Existing diagnostic analysis models can only make rough predictions based on clinical signs and are not sensitive to the heterogeneity of pathogenesis. As a result, current models cannot capture the heterogeneity of sepsis and cannot accurately predict sepsis based on disease heterogeneity. Summary of the Invention

[0004] This application provides a method, electronic device, and medium for predicting sepsis, which can improve the accuracy and reliability of sepsis prediction.

[0005] On one hand, embodiments of this application provide a method for predicting sepsis, including: By inputting the expression data of multiple different target genes of a patient into the sepsis prediction model, the sepsis prediction results of the patient are obtained. The sepsis prediction model is obtained through the following steps: Obtain multiple first training samples, wherein the first training samples include sample expression data of each of the target genes; For each of the first training samples, the first training sample is input into the pre-trained analytical model to obtain the training prediction result of the first training sample and the contribution relationship information between the sample expression data in the first training sample and the training prediction result. The initial sepsis prediction model is adjusted based on the contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model.

[0006] In one embodiment, before adjusting the initial sepsis prediction model based on the multiple contribution relationship information corresponding to each of the first training samples, the step further includes: Obtain multiple second training samples; The untrained initial sepsis prediction model is trained using multiple second training samples to obtain a pre-trained initial sepsis prediction model. The step of adjusting the initial sepsis prediction model based on the multiple contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model includes: The pre-trained initial sepsis prediction model is adjusted based on the multiple contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model.

[0007] In one embodiment, adjusting the pre-trained initial sepsis prediction model based on the multiple contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model includes: For each target gene, the relationship information between the target gene and sepsis is determined based on the contribution relationship information corresponding to the expression data of multiple samples of the target gene; The initial sepsis prediction model is adjusted based on the relationship information corresponding to each target gene to obtain the sepsis prediction model.

[0008] In one embodiment, adjusting the initial sepsis prediction model based on the relationship information corresponding to each of the target genes to obtain the sepsis prediction model includes: Target relationship information is determined based on the relationship information of each target gene, and the target relationship information is used to represent the association between each target gene and sepsis; The initial sepsis prediction model is adjusted based on the target relationship information to obtain the sepsis prediction model.

[0009] In one embodiment, adjusting the initial sepsis prediction model based on the multiple contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model includes: The untrained initial sepsis prediction model is adjusted based on the contribution relationship information corresponding to each of the first training samples to obtain the adjusted initial sepsis prediction model. Obtain multiple second training samples; The adjusted initial sepsis prediction model is trained using multiple second training samples to obtain the sepsis prediction model.

[0010] In one embodiment, the method further includes: Gene influence map is output based on the sepsis prediction results, and the gene influence map includes the degree of influence of each expression data on the sepsis prediction results.

[0011] In one embodiment, the step of outputting a gene influence map based on the sepsis prediction result includes: Based on the sepsis prediction results, a contribution analysis is performed on each of the expression data to obtain the influence of each expression data on the sepsis prediction results; The gene influence map is output based on the sepsis prediction results and the influence of each expression data.

[0012] In one embodiment, the step of performing a contribution analysis on each of the expression data based on the sepsis prediction result to obtain the influence of each expression data on the sepsis prediction result includes: Obtain the patient's gender and age information; Based on the gender information, the age information, and the sepsis prediction results, a contribution analysis is performed on each of the expression data to obtain the influence of each expression data on the sepsis prediction results.

[0013] On the other hand, embodiments of this application also provide an electronic device, including: At least one processor; At least one memory for storing at least one program; The method for predicting sepsis, as described above, is implemented when at least one program is executed by at least one processor.

[0014] On the other hand, embodiments of this application also provide a computer-readable storage medium storing computer-executable instructions for performing the sepsis prediction method described above.

[0015] On the other hand, embodiments of this application also provide a computer program product, including a computer program or computer instructions, the computer program or computer instructions being stored in a computer-readable storage medium, a processor of an electronic device reading the computer program or computer instructions from the computer-readable storage medium, and the processor executing the computer program or computer instructions to cause the communication device to perform the sepsis prediction method as described above.

[0016] This application provides a method, electronic device, and medium for predicting sepsis. The method includes: inputting expression data corresponding to multiple different target genes of a patient into a sepsis prediction model to obtain a sepsis prediction result for the patient; wherein the sepsis prediction model is obtained through the following steps: acquiring multiple first training samples, each first training sample including sample expression data of each target gene; inputting each first training sample into a pre-trained analytical model to obtain a training prediction result for the first training sample and contribution relationship information between each sample expression data in the first training sample and the training prediction result; adjusting an initial sepsis prediction model according to the multiple contribution relationship information corresponding to each first training sample to obtain the sepsis prediction model. Since the contribution relationship information between the expression data of each sample and the training prediction results can reflect the intrinsic link between the expression level of each target gene and whether the patient will develop sepsis, the initial sepsis prediction model can be adjusted by using the obtained contribution relationship information, so that the obtained sepsis prediction model can be closer to the gene action mechanism of sepsis. This allows the sepsis prediction model to output more accurate and reliable sepsis prediction results, effectively improving the accuracy and reliability of sepsis prediction. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the training steps of a sepsis prediction model provided in one embodiment of this application; Figure 2 This is a schematic diagram of the training steps of a sepsis prediction model provided in another embodiment of this application; Figure 3 yes Figure 1 A flowchart illustrating an embodiment of a sub-step of step 130; Figure 4A This is a schematic diagram of the output of the analytical model provided in the embodiments of this application; Figure 4B yes Figure 4A A schematic diagram illustrating the relationship information of a target gene in a given context; Figure 5 yes Figure 3 A flowchart illustrating an embodiment of a sub-step in step 320. Figure 6 yes Figure 1 A flowchart illustrating another sub-step embodiment of step 130; Figure 7 This is a schematic diagram of the gene influence map provided in the embodiments of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] It should be noted that although the flowchart shows a logical order, in some cases, the steps shown or described may be executed in a different order than that shown in the flowchart. In the description of the embodiments of this application, "multiple" (or more than) means two or more, "greater than," "less than," and "exceeding" are understood to exclude the number itself, while "above," "below," and "within" are understood to include the number itself. If "first," "second," etc., are described, they are only used to distinguish technical features and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the order of the technical features indicated. Furthermore, in the description of the embodiments of this application, the various values ​​mentioned (such as first value, second value, etc.) can be flexibly represented as a single numeric code or an enumerated type value.

[0020] In related technologies, sepsis is associated with multiple genes, resulting in heterogeneity in the pathogenesis of each sepsis patient. Existing diagnostic analysis models can only make rough predictions based on clinical signs and are not sensitive to the heterogeneity of pathogenesis. As a result, current models cannot capture the heterogeneity of sepsis and cannot accurately predict sepsis based on disease heterogeneity.

[0021] To improve the accuracy and reliability of sepsis prediction, embodiments of this application provide a sepsis prediction method, electronic device, computer-readable storage medium, and computer program product. The method includes: inputting expression data corresponding to multiple different target genes of a patient into a sepsis prediction model to obtain a sepsis prediction result for the patient; wherein, the sepsis prediction model is obtained through the following steps: acquiring multiple first training samples, the first training samples including sample expression data of each target gene; inputting the first training samples into a pre-trained analytical model for each first training sample to obtain a training prediction result of the first training sample and contribution relationship information between the expression data of each sample in the first training sample and the training prediction result; adjusting the initial sepsis prediction model according to the multiple contribution relationship information corresponding to each first training sample to obtain a sepsis prediction model. Since the contribution relationship information between the expression data of each sample and the training prediction results can reflect the intrinsic link between the expression level of each target gene and whether the patient will develop sepsis, the initial sepsis prediction model can be adjusted by using the obtained contribution relationship information, so that the obtained sepsis prediction model can be closer to the gene action mechanism of sepsis. This allows the sepsis prediction model to output more accurate and reliable sepsis prediction results, effectively improving the accuracy and reliability of sepsis prediction.

[0022] The sepsis prediction method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the sepsis prediction method, but is not limited to the above forms.

[0023] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0024] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the normal operation of embodiments of this application obtained.

[0025] In the embodiments of this application, when performing the sepsis prediction method, the expression data corresponding to multiple different target genes of the patient can be input into the sepsis prediction model to obtain the sepsis prediction result of the patient.

[0026] In one embodiment, the target gene refers to a biomarker that can express proteins in a patient, leading to sepsis. Multiple different target genes may include IRAK3, LIN7A, MME, MS4A3, CEACAM8, SESN3, EPHB1, ARG2, DEFA4, and CRISP3, etc., and are not specifically limited here.

[0027] In one embodiment, the expression data of the target gene refers to data used to characterize the expression level of the target gene in the patient's body. The expression data can be the abundance of the target gene, etc., and is not specifically limited here.

[0028] In one embodiment, the sepsis prediction model refers to a neural network model that can predict whether a patient will develop sepsis based on the current expression levels of multiple target genes in the patient's body. The sepsis prediction model can be a deep neural network (DNN), a large language model (LLM), etc., and is not specifically limited here.

[0029] See Figure 1 , Figure 1 The training steps for a sepsis prediction model are illustrated. In one embodiment, the sepsis prediction model can be obtained through the following steps.

[0030] Step 110: Obtain multiple first training samples, which include the expression data of each target gene; Step 120: For each first training sample, input the first training sample into the pre-trained analytical model to obtain the training prediction result of the first training sample and the contribution relationship information between the expression data of each sample in the first training sample and the training prediction result. Step 130: Adjust the initial sepsis prediction model based on the multiple contribution relationship information corresponding to each first training sample to obtain the sepsis prediction model.

[0031] In one embodiment, the sample expression data of the target gene refers to data on the expression level of the target gene in the body of a patient serving as a sample. The patient serving as a sample can be a patient with sepsis or a patient without sepsis; no specific limitation is made here.

[0032] In one embodiment, obtaining multiple first training samples refers to retrieving data from one or more databases to obtain multiple first training samples. The sources of the first training samples are broad, including samples shared in the National Inpatient Samples (NIS) database and samples from the Global Burden of Disease (GBD) database. It should be noted that the first training samples may also include sample labels for comparison with the training prediction results.

[0033] In one embodiment, during the acquisition of multiple first training samples, the retrieved samples can be filtered according to a preset screening sample to obtain multiple first training samples. By filtering the retrieved samples, samples with low quality or missing key data can be removed from the multiple samples, thereby obtaining higher quality first training samples, which is beneficial to improving the training quality of the sepsis prediction model.

[0034] In one embodiment, the pre-trained analytical model refers to a neural network model that possesses actual reasoning capabilities and, during the reasoning process, can correctly predict whether a patient, as a sample, suffers from sepsis based on the sample expression data of each target gene, and analyze the contribution relationship information between each target gene and the training prediction result. The analytical model can be a Kolmogorov-Arnold Network (KAN). Furthermore, the contribution relationship information refers to information used to represent the degree of influence of the sample expression data of the target gene on the training prediction result decided by the analytical model. This contribution relationship information can be specifically represented through relationship curve functions, data containing multiple relationship parameters, etc., without specific limitations here.

[0035] In one embodiment, adjusting the initial sepsis prediction model based on multiple contribution relationship information corresponding to each first training sample to obtain a sepsis prediction model refers to adjusting the hyperparameters in the initial sepsis prediction model based on all obtained contribution relationship information, and then using the hyperparameter-tuned initial sepsis prediction model to complete subsequent training steps and obtain the sepsis prediction model.

[0036] It is important to note that the different properties of the initial sepsis prediction models will lead to substantial differences in the training steps after parameter tuning. Specifically, there are two ways to adjust the initial sepsis prediction model using multiple contribution relationship information to obtain the sepsis prediction model: one is to adjust the pre-trained initial sepsis prediction model, and the other is to adjust the untrained initial sepsis prediction model and then train the adjusted initial sepsis prediction model.

[0037] Let's first explain the first scenario.

[0038] See Figure 2 In one embodiment, prior to step 130, the training step further includes the following steps.

[0039] Step 210: Obtain multiple second training samples; Step 220: Train the untrained initial sepsis prediction model based on multiple second training samples to obtain a pre-trained initial sepsis prediction model; Step 130 may also include the following sub-steps.

[0040] Step 230: Adjust the pre-trained initial sepsis prediction model according to the multiple contribution relationship information corresponding to each first training sample to obtain the sepsis prediction model.

[0041] In one embodiment, obtaining multiple second training samples refers to the operation of retrieving data from one or more databases to obtain multiple second training samples. The sources of these second training samples are broad, including samples shared in the National Inpatient Samples (NIS) database and samples from the Global Burden of Disease (GBD) database.

[0042] In one embodiment, during the acquisition of multiple second training samples, the retrieved samples can be filtered according to a preset screening sample to obtain multiple second training samples. By filtering the retrieved samples, samples with low quality or missing key data can be removed from the multiple samples, thereby obtaining higher quality first training samples, which is beneficial to improving the training quality of the sepsis prediction model.

[0043] In one embodiment, the second training sample includes sample labels and sample expression data of each target gene. Furthermore, the multiple second training samples can be derived from multiple first training samples, or they can be other training samples besides the multiple first training samples, etc., and no specific limitations are specified here.

[0044] In one embodiment, in the process of training an untrained initial sepsis prediction model based on multiple second training samples to obtain a pre-trained initial sepsis prediction model, specifically, the untrained initial sepsis prediction model can first be trained using the second training samples to obtain the training result corresponding to each second training sample. Then, the target loss function is determined based on the training result and sample label corresponding to each second training sample. Finally, the parameters of the initial sepsis prediction model are tuned based on the target loss function to obtain the pre-trained initial sepsis prediction model.

[0045] In one embodiment, adjusting the pre-trained initial sepsis prediction model based on multiple contribution relationship information corresponding to each first training sample to obtain a sepsis prediction model refers to adjusting one or more hyperparameters in the pre-trained initial sepsis prediction model based on all obtained contribution relationship information, then verifying the hyperparameter-adjusted initial sepsis prediction model through a preset validation set, and determining the successfully verified initial sepsis prediction model as the sepsis prediction model.

[0046] For a pre-trained initial sepsis prediction model, adjusting the model using a loss function only ensures that the model passes validation. However, the loss function cannot express the intrinsic relationship between the expression of each target gene and sepsis. In the medical field, such a model, lacking guidance from this intrinsic relationship, is prone to errors, leading to incorrect medical decisions by healthcare professionals and even patient deaths. Therefore, a model trained in this way is unreliable for sepsis prediction. Multiple contribution relationship information, on the other hand, reflects the intrinsic relationship between the expression of each target gene and sepsis. Adjusting the pre-trained initial sepsis prediction model using multiple contribution relationship information allows the adjusted model to predict sepsis based on this intrinsic relationship, thereby improving the reliability of the sepsis prediction results.

[0047] See Figure 3 In one embodiment, the process of adjusting the pre-trained initial sepsis prediction model based on multiple contribution relationship information corresponding to each first training sample to obtain the sepsis prediction model includes the following steps.

[0048] Step 310: For each target gene, determine the relationship between the target gene and sepsis based on the contribution relationship information corresponding to the expression data of multiple samples of the target gene; Step 320: Adjust the initial sepsis prediction model based on the relationship information corresponding to each target gene to obtain the sepsis prediction model.

[0049] In one embodiment, the relational information refers to information used to represent the pattern of action of the target gene on the patient's sepsis. This relational information can be specifically represented through relational curve functions, data containing multiple relational parameters, etc., without any specific limitation here.

[0050] For example, see Figure 4A and Figure 4B ,like Figure 4A As shown, inputting a training sample into the analytical model enables the contribution relationship information corresponding to each target gene. Then, for each target gene, the relationship information of each target gene is formed based on the contribution relationship information of the target gene in each training sample. For example... Figure 4B The relationship information of the target gene EPHB1 is shown. The relationship information of the target gene EPHB1 is represented by the relationship curve function obtained by fitting each contribution relationship information. The target gene EPHB1 has a negative impact on the sepsis prediction results. The higher the expression level of the target gene EPHB1 in the patient's body, the more likely the patient is to develop sepsis.

[0051] Since relational information can reflect the effect of the corresponding target genes on sepsis, adjusting the pre-trained initial sepsis prediction model by using the relational information of each target gene can make the initial sepsis prediction model closer to the pathogenesis of sepsis, thereby improving the reliability of sepsis prediction results.

[0052] See Figure 5 In one embodiment, the process of adjusting the initial sepsis prediction model based on the relationship information corresponding to each target gene to obtain the sepsis prediction model may include the following steps.

[0053] Step 410: Determine the target relationship information based on the relationship information of each target gene. The target relationship information is used to represent the association between each target gene and sepsis. Step 420: Adjust the initial sepsis prediction model based on the target relationship information to obtain the sepsis prediction model.

[0054] In one embodiment, the association between each target gene and sepsis refers to the relationship between the amount of protein expressed by each target gene and the risk of developing sepsis.

[0055] In one embodiment, the target genes are IRAK3, LIN7A, MME, MS4A3, CEACAM8, SESN3, EPHB1, ARG2, DEFA4 and CRISP3 mentioned above, and the target relationship information can be represented by the following formula (1).

[0056] ; Q can be used to indicate the risk of developing sepsis; X1 can be used to represent expression data of the target gene DEFA4; X2 can be used to represent expression data of the target gene CRISP3; X3 can be used to represent expression data of the target gene EPHB1; X4 can be used to represent expression data of the target gene ARG2; X5 can be used to represent expression data of the target gene SESN3; X6 can be used to represent expression data of the target gene IRAK3; X7 can be used to represent expression data of the target gene CEACAM8; X8 can be used to represent expression data of target gene MME; X9 can be used to represent expression data of the target gene LIN7A; X 10 It can be used to represent expression data of the target gene MS4A3.

[0057] Let's now explain the second scenario.

[0058] See Figure 6 In one embodiment, the process of adjusting the initial sepsis prediction model based on multiple contribution relationship information corresponding to each first training sample to obtain a sepsis prediction model may include the following steps.

[0059] Step 510: Adjust the untrained initial sepsis prediction model based on the multiple contribution relationship information corresponding to each first training sample to obtain the adjusted initial sepsis prediction model. Step 520: Obtain multiple second training samples; Step 530: Train the adjusted initial sepsis prediction model based on multiple second training samples to obtain the sepsis prediction model.

[0060] In one embodiment, adjusting the untrained initial sepsis prediction model based on multiple contribution relationship information corresponding to each first training sample to obtain an adjusted initial sepsis prediction model refers to adjusting one or more hyperparameters in the untrained initial sepsis prediction model based on all obtained contribution relationship information to obtain the sepsis prediction model to be trained. It should be noted that the adjustment process in step 510 can refer to steps 310 to 320 and steps 410 to 420 described above, and will not be repeated here.

[0061] It should be noted that the principle of step 520 is the same as that of step 210, and will not be repeated here.

[0062] In one embodiment, in the process of training an untrained and adjusted initial sepsis prediction model based on multiple second training samples to obtain a sepsis prediction model, specifically, the untrained initial sepsis prediction model can first be trained using the second training samples to obtain the training result corresponding to each second training sample. Then, the target loss function is determined based on the training result and sample label corresponding to each second training sample. Finally, the initial sepsis prediction model is tuned based on the target loss function to obtain the sepsis prediction model.

[0063] By adjusting the pre-trained initial sepsis prediction model using multiple contribution relationship information, the adjusted initial sepsis prediction model can make sepsis predictions based on intrinsic connections that are close to the pathogenesis of sepsis. Since the sepsis prediction is close to the pathogenesis of sepsis, an initial sepsis prediction model with high accuracy can be obtained with fewer training rounds, effectively shortening the training time of the initial sepsis prediction model. Thus, while ensuring the reliability of the initial sepsis prediction model, the training efficiency of the initial sepsis prediction model can be effectively improved.

[0064] In one embodiment, the sepsis prediction method may further include the following steps: outputting a gene influence map based on the sepsis prediction results, wherein the gene influence map includes the degree of influence of each expression data on the sepsis prediction results.

[0065] In one embodiment, a gene influence map refers to the impact of the current expression data of each target gene on the sepsis prediction model's output of the sepsis prediction result. The gene influence map can be represented in various forms, such as... Figure 7 The waterfall chart shown, illustrating the influence of each data point, can also be presented as follows: Figure 7 The force control diagram shown illustrates the influence of each expressed data point, etc., but specific details are not limited here.

[0066] Gene influence maps can reveal the heterogeneity of the influence of each target gene, allowing doctors to make targeted medical decisions based on the expression data of the target genes. This can improve the treatment effect for patients already suffering from sepsis and the prevention effect for patients who have not yet developed sepsis.

[0067] In one embodiment, in the process of outputting the gene influence map based on the sepsis prediction results, the contribution analysis of each expression data can be performed first based on the sepsis prediction results to obtain the influence of each expression data on the sepsis prediction results, and then the gene influence map can be output based on the sepsis prediction results and the influence of each expression data.

[0068] In one embodiment, during the contribution analysis of each expression data based on the sepsis prediction results, SHAP calculation can be performed based on each expression data and the sepsis prediction results to obtain the degree of influence of each expression data on the sepsis prediction results.

[0069] In one embodiment, during the contribution analysis of each expression data based on the sepsis prediction result, the distance between the feature information corresponding to each expression data and the sepsis prediction result in the feature space can be calculated, and the distance between the feature information corresponding to each expression data and the sepsis prediction result is determined as the influence of each expression data on the sepsis prediction result.

[0070] In one embodiment, in the process of performing contribution analysis on each expression data based on the sepsis prediction results to obtain the influence of each expression data on the sepsis prediction results, the patient's gender and age information can be obtained first, and then contribution analysis can be performed on each expression data based on the gender information, age information and sepsis prediction results to obtain the influence of each expression data on the sepsis prediction results.

[0071] In one embodiment, the contribution analysis of each expression data based on gender information, age information, and sepsis prediction results refers to determining the influence weighting value corresponding to each target gene based on gender information and age information, and then performing a contribution analysis of each expression data based on the influence weighting value corresponding to each target gene and the sepsis prediction results to obtain the influence of each expression data on the sepsis prediction results.

[0072] Factors such as gender and age are actually associated with the expression of target genes. Therefore, by using gender and age information to perform contribution analysis on various expression data, the contribution analysis process can focus on the impact of gender and age on the expression of target genes. This allows the obtained influence to be closer to the pathogenesis of sepsis, which can help doctors make targeted medical decisions, thereby improving the treatment effect for patients with sepsis and the prevention effect for patients who have not yet developed sepsis.

[0073] In addition to the embodiments described above, one embodiment of this application also provides an electronic device. See also Figure 8 , Figure 8 This is a schematic diagram of the structure of a printer provided in one embodiment of this application. Figure 8 As shown, the electronic device includes a memory 1100 and a processor 1200. The number of memories 1100 and processors 1200 can be one or more. Figure 8 Taking a memory 1100 and a processor 1200 as an example; Figure 8 The memory 1100 and processor 1200 can be connected via a bus or other means. Figure 8 Taking the example of a connection between China and Israel via a bus.

[0074] The memory 1100, as a computer-readable storage medium, can be used to store one or more software programs, computer-executable programs, and modules, such as the programs, instructions, or modules corresponding to the information processing methods provided in any embodiment of this application. The processor 1200 implements the sepsis prediction method provided in any embodiment of this application by executing one or more computer programs, instructions, and modules stored in the memory 1100.

[0075] The memory 1100 may primarily include a program storage area and a data storage area, wherein the program storage area may store the operating system and computer programs required for at least one function. Furthermore, the memory 1100 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 1100 may further include memory remotely located relative to the processor 1200, and these remote memories may be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0076] In addition to the embodiments described above, one embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions for performing the sepsis prediction method as described in any of the preceding embodiments.

[0077] Furthermore, one embodiment of this application also provides a computer program product, including a computer program or computer instructions stored in a computer-readable storage medium, wherein a processor of an electronic device reads the computer program or computer instructions from the computer-readable storage medium, and the processor executes the computer program or computer instructions to cause a communication device to perform the sepsis prediction method as described in any of the preceding embodiments.

[0078] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0079] The above is a detailed description of the preferred embodiments of this application. However, this application is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method of predicting sepsis, characterized by, The method comprises: inputting expression data corresponding to a plurality of different target genes of a patient into a sepsis prediction model to obtain a sepsis prediction result of the patient; wherein the sepsis prediction model is obtained by the following steps: obtaining a plurality of first training samples, wherein each of the first training samples comprises sample expression data of each of the target genes; for each of the first training samples, inputting the first training sample into a pre-trained analysis model to obtain a training prediction result of the first training sample and contribution relationship information between each of the sample expression data in the first training sample and the training prediction result; adjusting an initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model.

2. The method of claim 1, wherein, Before adjusting the initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples, the method further comprises: obtaining a plurality of second training samples; training the untrained initial sepsis prediction model according to a plurality of the second training samples to obtain the pre-trained initial sepsis prediction model; the adjusting the initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model comprises: adjusting the pre-trained initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model.

3. The method of claim 2, wherein, the adjusting the pre-trained initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model comprises: for each of the target genes, determining relationship information between the target gene and sepsis according to the contribution relationship information corresponding to a plurality of the sample expression data of the target gene; adjusting the initial sepsis prediction model according to the relationship information corresponding to each of the target genes to obtain the sepsis prediction model.

4. The method of claim 3, wherein, the adjusting the initial sepsis prediction model according to the relationship information corresponding to each of the target genes to obtain the sepsis prediction model comprises: determining target relationship information according to the relationship information of each of the target genes, wherein the target relationship information is used to represent the association relationship between each of the target genes and sepsis; adjusting the initial sepsis prediction model according to the target relationship information to obtain the sepsis prediction model.

5. The method of claim 1, wherein, the adjusting the initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples to obtain the sepsis prediction model comprises: adjusting the untrained initial sepsis prediction model according to a plurality of the contribution relationship information corresponding to each of the first training samples to obtain an adjusted initial sepsis prediction model; obtaining a plurality of second training samples; training the adjusted initial sepsis prediction model according to a plurality of the second training samples to obtain the sepsis prediction model.

6. The method of claim 1, wherein, the method further comprises: output a gene influence graph according to the sepsis prediction result, the gene influence graph comprising an influence degree of each expression data on the sepsis prediction result.

7. The method of claim 6, wherein, The gene influence graph comprises: perform contribution analysis on each expression data according to the sepsis prediction result, to obtain the influence degree of each expression data on the sepsis prediction result; output the gene influence graph according to the sepsis prediction result and the influence degree of each expression data.

8. The method of claim 7, wherein, The contribution analysis on each expression data according to the sepsis prediction result, to obtain the influence degree of each expression data on the sepsis prediction result, comprises: obtain gender information and age information of the patient; perform contribution analysis on each expression data according to the gender information, the age information and the sepsis prediction result, to obtain the influence degree of each expression data on the sepsis prediction result.

9. An electronic device, comprising: comprise: at least one processor; at least one memory for storing at least one program; when at least one of the programs is executed by at least one of the processors, the sepsis prediction method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium storing computer-executable instructions, the computer-executable instructions comprising: The computer executable instructions are used to execute the sepsis prediction method according to any one of claims 1 to 8. The computer executable instructions are used to execute the sepsis prediction method according to any one of claims 1 to 8.

Images