State prediction method, device, equipment, storage medium and program product

Abstract

The present application relates to a state prediction method, device, equipment, storage medium and program product, and relates to the field of artificial intelligence technology. The method includes: obtaining gene sequence information of a target object, comparing the gene sequence information with a reference gene sequence, and obtaining abnormal sequence information in the gene sequence information; the abnormal sequence information is used to characterize the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information; obtaining the detection state information of the target object; inputting the abnormal sequence information and the detection state information into a state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtaining the target prediction state information of the target object based on the abnormal sequence features and the detection state features. The use of this method can improve the accuracy of the state prediction results.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a state prediction method, device, equipment, storage medium and program product. Background Art

[0002] In the medical field, the process of gene status is affected by many factors, such as genetic factors, environmental factors, emotional factors, etc. In this case, analyzing the process of gene status is crucial for the prediction and treatment of gene status.

[0003] In traditional technology, a large amount of gene sample information is usually collected in a statistical way to construct a table that can characterize the relationship between macro features such as regional features and emotional features and state labels. In this way, in practical applications, the state prediction results of the target object can be determined from the table based on the macro features of the target object. However, the accuracy of the state prediction results in traditional technology is low. Summary of the invention

[0004] Based on this, it is necessary to provide a state prediction method, device, equipment, storage medium and program product that can improve the accuracy of state prediction results in response to the above technical problems.

[0005] The present application provides a state prediction method. The method comprises:

[0006] Obtaining gene sequence information of a target object, comparing the gene sequence information with a reference gene sequence, and obtaining abnormal sequence information in the gene sequence information; the abnormal sequence information is used to characterize the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information;

[0007] Acquire detection status information of the target object;

[0008] The abnormal sequence information and the detection state information are input into a state prediction model to extract abnormal sequence features of the abnormal sequence information and detection state features of the detection state information, and the target prediction state information of the target object is obtained based on the abnormal sequence features and the detection state features.

[0009] The present application also provides a state prediction device. The device comprises:

[0010] A sequence acquisition module is used to acquire gene sequence information of a target object, compare the gene sequence information with a reference gene sequence, and acquire abnormal sequence information in the gene sequence information; the abnormal sequence information includes an abnormal position and abnormal type of a gene unit;

[0011] A status acquisition module, used to acquire detection status information of the target object;

[0012] The state analysis module is used to input the abnormal sequence information and the detection state information into a state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtain the target prediction state information of the target object based on the abnormal sequence features and the detection state features.

[0013] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned state prediction method when executing the computer program.

[0014] A computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above-mentioned state prediction method when executed by a processor.

[0015] A computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of the state prediction method are implemented.

[0016] The above-mentioned state prediction method, device, equipment, storage medium and program product obtain the gene sequence information of the target object, compare the gene sequence information with the reference gene sequence, and obtain the abnormal sequence information in the gene sequence information; the abnormal sequence information is used to characterize the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information; obtain the detection state information of the target object; input the abnormal sequence information and the detection state information into the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtain the target prediction state information of the target object based on the abnormal sequence features and the detection state features. In this way, the method of extracting, analyzing and predicting the features of the gene sequence information and the detection state information of the target object through the preset state prediction model can avoid the interference of subjective factors to a certain extent and ensure the objectivity of the target state prediction information. In addition, the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information are used to characterize the genetic characteristics of the target object at the microscopic level in a fine-grained manner, and the detection state information is used to characterize the statistical characteristics of the target object at the macroscopic level in a coarse-grained manner. When performing state prediction, the above information is analyzed through the state prediction model, which is equivalent to comprehensively considering the information of different modalities and different omics of the target object, so that the state prediction model can extract more comprehensive and rich multimodal biological characteristics, ensure the effectiveness of the basis for feature analysis of the state prediction model, and improve the authenticity and accuracy of the target prediction state information obtained based on multimodal biological feature analysis. In addition, the information state prediction step is executed by the state prediction model deployed on the computer device. In order to avoid the situation where the state prediction model fails to recognize the information, the gene sequence information of the target object is processed in advance before the state prediction model executes the feature extraction and analysis steps, so that the abnormal sequence information and the detection state information meet the recognition requirements of the state prediction model, ensure the normal operation of the state prediction model, and then make full use of the computing resources on the computer device. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 An application environment diagram of a state prediction method in an embodiment;

[0018] Figure 2 A schematic diagram of a first flow chart of a state prediction method in an embodiment;

[0019] Figure 3 A schematic diagram of the contents of abnormal sequence information in one embodiment;

[0020] Figure 4 A schematic diagram of a first architecture of a state prediction model in one embodiment;

[0021] Figure 5 A schematic diagram of the contents of an abnormal feature matrix in one embodiment;

[0022] Figure 6 A schematic diagram of a second architecture of a state prediction model in one embodiment;

[0023] Figure 7 A schematic diagram of a third architecture of a state prediction model in one embodiment;

[0024] Figure 8 A second flow chart of a state prediction method in one embodiment;

[0025] Fig. 9 A third flow chart of a state prediction method in one embodiment;

[0026] Fig.10 A schematic diagram of a process of a gene disease type prediction model in one embodiment;

[0027] Fig.11 is a structural block diagram of a state prediction device in one embodiment;

[0028] Fig.12 is an internal structure diagram of a computer device in one embodiment;

[0029] Fig.13 FIG. 4 is a diagram showing the internal structure of a computer device in another embodiment. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0031] The state prediction method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 104 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0032] Both the terminal and the server can be used independently to execute the state prediction method provided in the embodiments of the present application.

[0033] For example, the terminal obtains the gene sequence information of the target object, compares the gene sequence information with the reference gene sequence, and obtains the abnormal sequence information in the gene sequence information, wherein the abnormal sequence information is used to characterize the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information. The terminal obtains the detection state information of the target object, and the terminal calls the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtains the target prediction state information of the target object based on the abnormal sequence features and the detection state features.

[0034] The terminal and the server can also be used together to execute the state prediction method provided in the embodiments of the present application.

[0035] For example, the terminal will obtain the gene sequence information of the target object, compare the gene sequence information with the reference gene sequence, obtain the abnormal sequence information in the gene sequence information, and send the detection state information and abnormal sequence information of the target object to the server. The server calls the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtains the target prediction state information of the target object based on the abnormal sequence features and the detection state features. The server returns the target prediction state information to the terminal.

[0036] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0037] In one embodiment, Figure 2 As shown, a state prediction method is provided, and the method is applied to a computer device as an example. The computer device can be a terminal or a server, and the method can be executed by the terminal or the server itself alone, or through the interaction between the terminal and the server. Figure 2 , the state prediction method includes the following steps:

[0038] Step S202, obtaining gene sequence information of the target object, comparing the gene sequence information with the reference gene sequence, and obtaining abnormal sequence information in the gene sequence information; the abnormal sequence information is used to characterize the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information.

[0039] Among them, the gene sequence information is a base sequence that characterizes the genetic information of the target object, specifically a sequence generated by the arrangement and combination of adenine (A), thymine (T), cytosine (C) and guanine (G). The gene sequence information of the target object can be obtained by testing the sample tissue of the target object, such as cells, blood, hair and other samples. In the embodiment of the present application, the gene sequence information can be the gene sequence information that matches the specific gene fragment of the target object, or it can be the full gene sequence information of the target object.

[0040] In one scenario, the gene sequence information is the whole gene sequence information, the gene sequence information of the target object can be pre-stored in a data storage system, and the data storage system stores the gene sequence information of multiple objects. In this scenario, the computer device responds to a status detection request for the target object, and queries and obtains the gene sequence information of the target object from the data storage system based on the identifier of the target object.

[0041] In another scenario, the gene sequence information is the sequence of a specific gene fragment. The computer device can first obtain the full gene sequence information of the target object from the data storage system, and then extract the gene sequence information that matches the specific gene fragment from the full gene sequence.

[0042] The reference gene sequence refers to the standard gene sequence of the species corresponding to the target object, which also includes the four bases A, T, C, and G. It should be noted here that although the base types of the gene sequence information and the reference gene sequence are the same, in actual applications, the target object is affected by various uncertain factors such as the environment, emotions, and living habits, which may cause the base position, base arrangement, and gene sequence length in the target object's gene sequence information to change, that is, gene mutation, which in turn leads to certain differences between the gene sequence information and the reference gene sequence.

[0043] In the embodiment of the present application, the abnormal gene sequence in the gene sequence information is determined by comparing the gene sequence information with the reference gene sequence, and then the abnormal position and abnormal type of each abnormal gene unit in the abnormal gene sequence are sorted to obtain the abnormal sequence information, so as to characterize the gene characteristics of the target object at the microscopic level in a fine-grained manner. Among them, the abnormal position of the abnormal gene unit includes the chromosome where the abnormal gene unit is located and the absolute position of the abnormal gene unit in the reference gene sequence, and the variation type of the abnormal gene unit includes the current base type of the abnormal gene unit and the normal base type matched before the abnormal gene unit mutates.

[0044] Optionally, when obtaining gene sequence information and a reference gene sequence, two gene units at the same site of the gene sequence information and the reference gene sequence are compared in a site-by-site manner, and the abnormal gene unit that does not match the reference gene sequence, the reference gene unit corresponding to the abnormal gene unit, and the position of the abnormal gene unit are screened out from the gene sequence information; then, base type recognition is performed on the abnormal gene unit and the reference gene unit to determine the abnormal type of the abnormal gene unit, and the position of the abnormal gene unit is determined as the abnormal position; finally, the abnormal type and abnormal position of each abnormal gene unit are summarized to obtain the abnormal sequence information in the gene sequence information.

[0045] Optionally, when obtaining gene sequence information and a reference gene sequence, the gene sequence information and the reference gene sequence are compared according to gene fragments to determine abnormal gene fragments in the gene sequence. Then, the abnormal gene fragments are compared site by site with the reference gene fragments to determine the abnormal position and abnormal type of each abnormal gene unit in the abnormal gene fragment to obtain abnormal sequence information.

[0046] In practical applications, in addition to abnormal gene units that have undergone gene mutations, abnormal gene sequences may also include normal gene units at adjacent sites of each abnormal gene unit. In this case, in the abnormal sequence information obtained by sorting out the gene units of the abnormal gene sequence, the abnormal position of each abnormal gene unit can be represented by the gene site of the adjacent gene unit.

[0047] Step S204: acquiring detection status information of the target object.

[0048] The detection status information is content describing the status characteristics of the target object, and the status characteristics include but are not limited to geographical information, age, image of the region of interest, etc. Accordingly, the detection status information is represented in text, and / or numerical form, and / or image form.

[0049] Optionally, in response to a status detection request for a target object, the computer device queries and obtains detection status information of the target object from a data storage system based on an identifier of the target object.

[0050] Optionally, the state characteristics of the target object in different omics are collected, and then the state characteristics of different omics are uniformly converted into a format to obtain the detection state information of the target object, and coarse-grainedly characterize the statistical characteristics of the target object at the macro level.

[0051] Taking the state characteristics of the target object including geographical information, age information and image of the region of interest as an example, the numerical conversion model can be called to convert the geographical information, age information and image information of the region of interest into state information in data format respectively, and then the converted state information is spliced ​​to obtain the detection state information of the target object.

[0052] Taking the state characteristics of the target object including geographical information, age information and image of the region of interest as an example, the text conversion model can be called to convert the geographical information, age information and image information of the region of interest into state information in text format respectively, and then the converted state information is spliced ​​to obtain the detection state information of the target object.

[0053] Step S206, inputting the abnormal sequence information and the detection state information into the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtaining the target prediction state information of the target object based on the abnormal sequence features and the detection state features.

[0054] The state prediction model is a deep learning model, such as a transformer model, which supports parallel calculation of different types of input information, and integrates the characteristics of various types of input information to output target prediction state information that matches the target object. The number of target prediction state information can be one or more, and the content of the target prediction state includes each target prediction state and the probability value of each target prediction state.

[0055] In the embodiment of the present application, the state prediction model characterizes the complex association between multiple different types of state factors of the object and the actual state information. The input of the state prediction model is: the abnormal sequence information and detection state information of the object; the internal processing logic of the state prediction model is: by combining the individual analysis results of each type of input information and the correlation analysis results between different input information, determine the degree of matching between the object and multiple candidate state information; the output of the state prediction model is: one or more target predicted state information matching the object and the probability value of each target predicted state information.

[0056] In one scenario, the state prediction model includes a coding model and a decoding model, the coding model is used for feature extraction to obtain abnormal sequence features and detection state features, and the decoding model is used to decode the extracted features to obtain target prediction state information. Specifically, the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information are extracted in parallel by the coding model, and the abnormal sequence features and the detection state features are analyzed by the decoding model to obtain the target prediction state information of the target object.

[0057] In another scenario, the state prediction model includes a coding model and a decoding model, the coding model is used for feature extraction and feature fusion to obtain multimodal biometric features, and the decoding model is used to decode the fused multimodal biometric features to obtain target prediction state information. Specifically, the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information are extracted in parallel by the coding model, the extracted abnormal sequence features and the detection state features are fused to obtain the multimodal biometric features of the target object, and the multimodal biometric features are analyzed by the decoding model to obtain the target prediction state information of the target object.

[0058] It should be emphasized that the abnormal sequence information includes the abnormal position and abnormal type, so the abnormal sequence features extracted based on the abnormal sequence information also include position features and type features. Furthermore, the multimodal biometric features obtained by fusing the abnormal sequence features and the detection state features include position features, type features and detection state features, so that the state prediction model can learn the impact of features under each modality on the state information, the impact of the correlation of various features under different modalities on the state information, and then accurately predict the state information of the target object.

[0059] In the above embodiment, the gene sequence information of the target object is obtained, the gene sequence information is compared with the reference gene sequence, and the abnormal sequence information in the gene sequence information is obtained; the abnormal sequence information is used to characterize the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information; the detection state information of the target object is obtained; the abnormal sequence information and the detection state information are input into the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and the target prediction state information of the target object is obtained based on the abnormal sequence features and the detection state features. In this way, the gene sequence information and the detection state information of the target object are extracted, analyzed and predicted by the preset state prediction model, which avoids the interference of subjective factors to a certain extent and ensures the objectivity of the target state prediction information. In addition, the abnormal position and abnormal type of the abnormal gene unit in the gene sequence information are used to characterize the genetic characteristics of the target object at the microscopic level in a fine-grained manner, and the detection state information is used to characterize the statistical characteristics of the target object at the macroscopic level in a coarse-grained manner. When performing state prediction, the above information is analyzed through the state prediction model, which is equivalent to comprehensively considering the information of different modalities and different omics of the target object, so that the state prediction model can extract more comprehensive and rich multimodal biological characteristics, ensure the effectiveness of the basis for feature analysis of the state prediction model, and thus improve the authenticity and accuracy of the target prediction state information obtained based on multimodal biological feature analysis. In addition, the information state prediction step is executed by the state prediction model deployed on the computer device. In order to avoid the situation where the state prediction model fails to recognize the information, the gene sequence information of the target object is processed in advance before the state prediction model executes the feature extraction and analysis steps, so that the abnormal sequence information and the detection state information meet the recognition requirements of the state prediction model, ensure the normal operation of the state prediction model, and then make full use of the computing resources on the computer device.

[0060] In one embodiment, obtaining gene sequence information of a target object includes:

[0061] The initial gene sequence of the target object is obtained, and the initial gene sequence includes gene units of multiple sites; the initial gene sequence is quality controlled to obtain gene sequence information.

[0062] The initial gene sequence may be a base sequence obtained by DNA sequencing, or may be collected by sequencing equipment. The initial gene sequence includes multiple gene units, and each gene unit refers to a single base in the base sequence.

[0063] After obtaining the initial gene sequence, a quality control processing step is performed to update the initial gene sequence to obtain gene sequence information. Quality control processing of the initial gene sequence refers to the operation of identifying and deleting low-quality gene data in the initial gene sequence and retaining high-quality gene units.

[0064] Optionally, the quality control process includes removing erroneous data, which refers to data introduced by the testing device due to hardware problems or testing methods during the process of the testing device testing the initial gene sequence. Specifically, according to a preset evaluation method, the quality of each gene unit in the initial gene sequence is quantitatively scored, and the gene units with quantitative score values ​​lower than the preset quality value are deleted from the initial gene sequence to obtain gene sequence information.

[0065] Optionally, the quality control process includes a duplicate data removal operation, where duplicate data refers to a gene sequence that is identical or similar to the original gene sequence and appears multiple times. Specifically, the original gene sequence can be divided into a plurality of gene subsequences, and the similarity between the gene subsequences is calculated. Among the two gene subsequences whose similarity is greater than a preset similarity threshold, any one of the two gene subsequences is deleted to obtain the gene sequence information.

[0066] In the above embodiment, after obtaining the initial gene sequence of the gene unit including multiple sites, the initial gene sequence is quality controlled to obtain gene sequence information, which is equivalent to taking into account multiple factors including the test error introduced in the process of obtaining the initial gene sequence and the redundancy of the initial gene sequence, eliminating invalid data of the initial gene sequence, and screening out valid data, thereby improving the reliability and effectiveness of the gene sequence information while reducing the computational burden of the subsequent state prediction model.

[0067] In one embodiment, comparing the gene sequence information with the reference gene sequence to obtain abnormal sequence information in the gene sequence information includes:

[0068] The gene sequence information and the reference gene sequence are compared to obtain the target gene sequence in which the gene sequence information is in the candidate site set; each gene unit on the target gene sequence is compared with the corresponding gene unit on the reference gene sequence to determine the normal gene unit and the abnormal gene unit in the target gene sequence; based on the reference gene sequence, the normal gene unit and the abnormal gene unit, the abnormal type sequence and the abnormal position sequence of the target object are constructed; according to the absolute position of the gene unit in the reference gene sequence, the abnormal type sequence and the abnormal position sequence are compared to obtain the abnormal sequence information.

[0069] Sequence alignment refers to aligning two gene sequences according to the absolute position of the gene unit in the reference gene sequence. The candidate site set includes multiple sites, which can be sites of multiple gene units preset based on empirical values. The gene units at each site in the candidate site set are queried in the gene sequence information, and the queried multiple gene units are sorted according to the absolute position of the gene unit to obtain the target gene sequence.

[0070] Each gene unit in the target gene sequence is compared with the gene unit at the corresponding site on the reference gene sequence. For any site on the target gene sequence, if the gene unit of the target gene sequence at the site is the same as the gene unit of the reference gene sequence at the site, the gene unit at the site is determined to be a normal gene unit; if the gene unit of the target gene sequence at the site is different from the gene unit type of the reference gene sequence at the site, or the target gene sequence has a gene unit at the site, while the reference gene sequence does not have a gene unit at the site, it is determined that a gene mutation has occurred at the site, and the gene unit of the target gene sequence at the site is an abnormal gene unit; if the target gene sequence does not have a gene unit at the site, it is determined that a gene mutation has occurred at the site, and the gene unit at the site on the reference gene sequence is an abnormal gene unit.

[0071] On the basis of determining the normal gene units and the abnormal gene units, first, for each abnormal gene unit, the gene unit before and after the abnormal gene unit is mutated are taken as a group of mutated gene combinations to obtain a two-dimensional vector corresponding to the abnormal gene unit; for each normal gene unit, the normal gene unit is copied twice as a group of non-mutated gene combinations to obtain a two-dimensional vector corresponding to the normal gene unit.

[0072] Next, the absolute position of each abnormal gene unit is determined on the reference gene sequence, and based on the chromosomal position and absolute position of each abnormal gene unit, the two-dimensional vector corresponding to each abnormal gene unit is expanded to obtain a four-dimensional vector corresponding to each abnormal gene unit; the absolute position of each normal gene unit is determined on the reference gene sequence, and based on the chromosomal position and absolute position of each normal gene unit, the two-dimensional vector corresponding to each normal gene unit is expanded to obtain a four-dimensional vector corresponding to each normal gene unit.

[0073] Then, while keeping the vector length unchanged, the four-dimensional vectors corresponding to each abnormal gene unit and the four-dimensional vectors corresponding to each normal gene unit are stacked to construct a fourth-order matrix of the target gene sequence. Furthermore, the two vectors representing each gene unit before and after mutation are used as the abnormal type sequence, and the vectors representing the chromosome position and absolute position of each gene unit are used as the abnormal position sequence.

[0074] Taking the four-dimensional vector as a column vector as an example, if the target gene sequence includes M variant gene sequences and N non-variant gene sequences, the size of the fourth-order matrix of the target gene sequence is four rows and (M+N) columns. Each row represents one dimension of information of each gene unit in the target gene sequence, and each column represents four dimensions of information of a gene unit in terms of gene type before mutation, gene type after mutation, chromosome position of mutation, and absolute position of mutation. At this time, the first and second rows of the fourth-order matrix are determined as abnormal type sequences, and the third and fourth rows are determined as abnormal position sequences.

[0075] Taking a four-dimensional vector as a row vector as an example, if the target gene sequence includes M variant gene sequences and N non-variant gene sequences, the size of the fourth-order matrix of the target gene sequence is four columns and (M+N) rows. Each column represents one dimension of information of each gene unit in the target gene sequence, and each row represents the information of a gene unit in four dimensions: gene type before mutation, gene type after mutation, chromosome position of mutation, and absolute position of mutation. At this time, the first and second columns of the fourth-order matrix are determined as abnormal type sequences, and the third and fourth columns are determined as abnormal position sequences.

[0076] Finally, the abnormal type sequences and abnormal position sequences are sorted in order from small to large (or from large to small) according to the absolute position of the gene unit in the reference gene sequence to obtain the aligned abnormal type sequence and abnormal position sequence as the abnormal sequence information.

[0077] In the above embodiment, the gene sequence information and the reference gene sequence are compared to facilitate the analysis of each gene unit in the gene sequence information; then, the target gene sequence of the candidate site set is screened out from the gene sequence information, which further reduces the analysis scope for the subsequent gene unit variation analysis, improves the gene abnormality analysis speed, and thus quickly and accurately determines the normal gene units and abnormal gene units in the target gene sequence; then, based on the two dimensions of abnormal type and abnormal position, the variation of each gene unit in the target gene sequence is finely characterized to obtain the abnormal type sequence and the abnormal position sequence, thereby enriching the abnormal sequence information content.

[0078] In one embodiment, the abnormal gene unit includes at least one of a newly added gene unit, a deleted gene unit and a replaced gene unit; the abnormal type sequence includes a gene reference sequence and a gene variation sequence; and the abnormal position sequence includes a chromosome position sequence and a gene absolute position sequence.

[0079] Based on the reference gene sequence, normal gene unit and abnormal gene unit, the abnormal type sequence and abnormal position sequence of the target object are constructed, including:

[0080] According to the absolute gene position of the gene unit in the reference gene sequence, the following steps are performed: the type identifier of each gene unit in the reference gene sequence corresponding to the target gene sequence and the gene addition identifier of each newly added gene unit are sorted and combined to obtain a gene reference sequence; the type identifier of the normal gene unit, the type identifier of the replacement gene unit, the type identifier of the newly added gene unit and the gene deletion identifier of each deleted gene unit are sorted and combined to obtain a gene variation sequence; the chromosome position corresponding to each gene unit in the gene variation sequence and the chromosome position of each deleted gene unit are sorted and combined to obtain a chromosome position sequence; and the absolute gene position of each gene unit in the gene variation sequence and the absolute gene position of each deleted gene unit are sorted and combined to obtain a gene absolute position sequence.

[0081] Among them, the newly added gene unit refers to the gene unit of the target gene sequence at the same site in the scenario where the target gene sequence has a gene unit but the reference gene sequence does not have a gene unit. The replaced gene unit refers to the gene unit of the target gene sequence at the same site in the scenario where both the target gene sequence and the reference gene sequence have a gene unit but the gene unit types are different. The deleted gene unit refers to the gene unit of the reference gene sequence at the same site in the scenario where the target gene sequence does not have a gene unit but the reference gene sequence has a gene unit.

[0082] For example, the abnormal gene unit includes a mutated abnormal gene unit and an adjacent normal gene unit of the abnormal gene unit. Figure 3 , Figure 3 is a schematic diagram of abnormal sequence information. Figure 3 The first two rows in the table are abnormal type sequences, with the first and second rows being gene reference sequences and gene variant sequences, respectively; Figure 3 The last two rows in the figure are the abnormal position sequences, and the third and fourth rows are the chromosome position and gene absolute position sequences, respectively.

[0083] Figure 3 In the , the gene reference sequence includes five identifiers: "A", "T", "C", "G" and "+". Among them, "A", "T", "C", "G" represent the base type of the gene unit, that is, the type identifier of each gene unit in the reference gene sequence corresponding to the target gene sequence, "+" represents the gene addition identifier of the newly added gene unit; the gene variation sequence includes five identifiers: "A", "T", "C", "G" and "-". Among them, "A", "T", "C", "G" represent the base type of the gene unit, and "-" represents the gene deletion identifier; the numerical value of the chromosome position represents the chromosome number where the base is located; the numerical value of the gene absolute position sequence represents the absolute position of the base on the reference genome.

[0084] from Figure 3 It can be seen from the abnormal sequence information shown that at the site with an absolute gene position of 20, the gene unit of the target object mutates from G to C, the gene unit of the target object at the site with an absolute gene position of 21-23 is deleted, and the target object has a new gene unit at the site with an absolute gene position of 81-84, and the new gene units are "C", "A", "A", and "A" bases respectively.

[0085] In the above embodiments, by identifying the gene addition marker in the gene reference sequence, the situation of the newly added gene unit of the target object is determined; by identifying the gene deletion marker in the gene variation series, the situation of the deleted gene unit of the target object is determined; by comparing the gene reference sequence and the gene variation series, the base type of the gene unit of the target object before and after the variation is determined; the positions of the newly added gene unit, the deleted gene unit and the abnormal gene unit in the reference genome sequence can be determined through the chromosome position sequence and the gene absolute position sequence.

[0086] In one embodiment, obtaining detection status information of a target object includes:

[0087] Acquire the text status information, numerical status information and image status information of the target object; perform data mapping on the text status information according to the preset text status conversion relationship to obtain the text feature information of the target object; perform vectorization processing on the image status information to obtain the image data information of the image status information, and summarize the numerical status information and the image data information to obtain the numerical feature information of the target object; obtain the detection status information based on the text feature information and the numerical feature information.

[0088] The text status information includes the category information of the target object, such as the gender of the target object, whether a certain behavioral feature exists, etc.; the numerical status information includes the life numerical information of the target object, such as the age and MSI value of the target object; the image status information includes the image of the target object's area of ​​interest, such as Computed Tomography (CT) images and Magnetic Resonance Imaging (MRI) images.

[0089] For text state information, the text state conversion relationship can be a mapping relationship between text state and numerical value, and the text state information is converted into a matching numerical value according to the text state conversion relationship as text feature information. For example, using onehot encoding, male gender is converted to 1, and female gender is converted to 0; using onehot encoding, the presence of a certain behavior feature is converted to 1, and the absence of a certain behavior feature is converted to 0.

[0090] For the numerical state information and image state information, the image features of the image state information are first extracted, and the image features are vectorized to obtain the characteristic values ​​of the image state information as the image data information. Then, the numerical state information and the image data information are spliced ​​in a preset order to obtain the numerical feature information of the target object.

[0091] In the order that the text feature information comes before the numerical feature information, the text feature information and the numerical feature information are summarized, that is, the text feature information, the numerical status information and the image data information are spliced ​​to obtain the detection status information of the target object.

[0092] It should be noted that, whether it is text feature information or numerical feature information, the format is unified and is represented in numerical form. The purpose is to retain the multi-dimensional feature information of the target object while facilitating the state prediction model to identify and process it, thereby improving the prediction speed of the state prediction model.

[0093] In the above embodiment, data mapping is performed on the text status information according to the preset text status conversion relationship to obtain text feature information of the target object; the image status information is vectorized to obtain image data information of the image status information, and then the text feature information, numerical status information and image data information are summarized to obtain detection status information, which is equivalent to unifying the form of text status information and image status information into a numerical representation, and comprehensively representing the status information of the target object in multiple different dimensions in the form of objective data, so as to improve the comprehensiveness and accuracy of the detection status information.

[0094] In one embodiment, Figure 4 As shown, the state prediction model includes an encoding network, a conversion network and a decoding network. The abnormal sequence information and the detection state information are input into the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and the target prediction state information of the target object is obtained based on the abnormal sequence features and the detection state features, including:

[0095] Based on the encoding network, the abnormal sequence features of the abnormal sequence information are extracted, and the detection state features of the detection state information are extracted; based on the conversion network, the abnormal sequence features and the detection state features are spliced, and the spliced ​​features are processed with dimension increase to obtain multimodal biological features; based on the decoding network, the multimodal biological features are processed with dimension reduction to output the target prediction state information of the target object.

[0096] In the state prediction model, the state prediction process is as follows: the abnormal sequence information and the detection state information are input into the encoding network to perform the feature extraction steps of the above two types of information in parallel, specifically extracting the abnormal sequence features of the abnormal sequence information and extracting the detection state features of the detection state information. Then, the extracted abnormal sequence features and detection state features are input into the conversion network to splice the abnormal sequence features and the detection state features to obtain fusion features, and further extract high-dimensional features from the fusion features to obtain the multimodal biometric features of the target object. Finally, the multimodal biometric features are input into the decoding network, and the high-dimensional multimodal biometric features are reduced by the decoding network to obtain the probability of the target object and multiple preset candidate state information. Then, according to the probability of the target object and multiple preset candidate state information, the target prediction state information of the target object is determined.

[0097] Optionally, multiple candidate state information and the probabilities of the candidate state information are summarized as the target predicted state information of the target object, or the candidate state information corresponding to the maximum probability is used as the target predicted state information of the target object.

[0098] In the above embodiment, the abnormal sequence features of the abnormal sequence information are extracted based on the encoding network, and the detection state features of the detection state information are extracted; the abnormal sequence features and the detection state features are spliced ​​based on the conversion network, and the spliced ​​features are processed by dimension increase to obtain multimodal biometric features, and the multimodal biometric features are processed by dimension reduction based on the decoding network to output the target prediction state information of the target object. In this way, feature extraction by parallel encoding can improve the feature extraction efficiency of the model, and then considering the correlation between features of different modes, the extracted features are spliced, which further improves the effectiveness between multimodal biometric features, and the spliced ​​features are dimensionally increased and dimensionally reduced to obtain accurate target prediction state information.

[0099] In one embodiment, the coding network includes a sequence anomaly coding layer; the abnormal sequence information includes a gene reference sequence, a gene variation sequence, a chromosome position sequence, and a gene absolute position sequence; and the abnormal sequence features of the abnormal sequence information are extracted based on the coding network, including:

[0100] The gene reference sequence and gene variation sequence are identified through the sequence anomaly coding layer to determine the abnormal type of each gene unit in the abnormal sequence information. Based on the abnormal type of each gene unit, the abnormal type vector of each gene unit is generated, and the abnormal type vectors of each gene unit are combined to obtain the type matrix of the abnormal sequence information; the chromosome position sequence is identified through the sequence anomaly coding layer to determine the chromosome position of each gene unit in the abnormal sequence information. Based on the chromosome position of each gene unit, the position vector of each gene unit is generated, and the position vectors of each gene unit are combined to obtain the chromosome position matrix of the abnormal sequence information; the type matrix, the chromosome position matrix and the gene absolute position sequence are spliced ​​to obtain the abnormal feature matrix, and the abnormal feature matrix is ​​determined as the abnormal sequence feature.

[0101] The input of the sequence anomaly coding layer is the gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence, and the output is the anomaly feature matrix. The sequence anomaly coding layer can be a linear layer (linear), and the coding logic is: encode and convert multiple input sequences, and use a unified representation form to describe the variation of different gene units.

[0102] The gene reference sequence and the gene variation sequence are identified through the sequence anomaly coding layer to determine whether each gene unit in the abnormal sequence information is mutated. After determining the abnormal gene unit, the abnormal type vector of each abnormal gene unit is determined from the preset abnormal type vector according to the variation type of the abnormal gene unit, and the normal type vector of each normal gene unit is determined according to the gene type of the normal gene unit. Among them, the dimensions of the abnormal type vector and the normal type vector are consistent. Then, the abnormal type vector of each abnormal gene unit and the normal type vector of each normal gene unit are combined to obtain the type matrix of the abnormal sequence information.

[0103] The chromosome position sequence is identified through the sequence anomaly coding layer to determine the chromosome position of each gene unit in the abnormal sequence information. The chromosome position of each gene unit is onehot encoded to obtain the position vector of each gene unit. The position vectors of each gene unit are combined to obtain the chromosome position matrix of the abnormal sequence information.

[0104] The type matrix and chromosome position matrix of the abnormal sequence information are spliced ​​through the sequence anomaly coding layer, and then the absolute gene position of each gene unit in the gene absolute position sequence is extracted, and the absolute gene position of each gene unit is spliced ​​at the adjacent position of the chromosome position vector of the corresponding gene unit to obtain the abnormal feature matrix of the abnormal sequence information, that is, the abnormal sequence feature.

[0105] In the embodiment of the present application, the gene reference sequence and the gene variation sequence are identified, processed and converted by the sequence anomaly coding layer to obtain the type matrix of the abnormal sequence information; the chromosome position sequence is identified and converted by the sequence anomaly coding layer to obtain the chromosome position matrix of the abnormal sequence information; the type matrix, the chromosome position matrix and the gene absolute position sequence are spliced ​​to obtain the abnormal feature matrix, and the abnormal feature matrix is ​​determined as the abnormal sequence feature. In this way, in the abnormal sequence feature, the same format is used to characterize the abnormal type and abnormal position of each gene unit to normalize the expression of the variation of each gene unit, facilitate the subsequent feature fusion of the variation feature and the detection state feature of the gene unit, improve the fusion speed of the conversion network in the state detection model, and thus shorten the state prediction time of the state detection model.

[0106] In one embodiment, based on the abnormal type of each gene unit, an abnormal type vector of each gene unit is generated, including:

[0107] For any gene unit, an initial type vector of the gene unit is constructed; the initial type vector includes a newly added gene identifier, a deleted gene identifier, multiple types of reference gene identifiers and multiple types of variant gene identifiers; if the gene unit is a newly added gene unit, the value of the newly added gene identifier in the initial type vector and the value of the variant gene identifier that matches the type of the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value, to generate an abnormal type vector of the gene unit; if the gene unit is a deleted gene unit, the value of the deleted gene identifier in the initial type vector and the value of the reference gene identifier that matches the type of the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value, to generate an abnormal type vector of the gene unit; if the gene unit is a replaced gene unit, the historical gene type of the gene unit is determined, the value of the reference gene identifier that matches the historical gene type and the value of the variant gene identifier that matches the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value, to generate an abnormal type vector of the gene unit.

[0108] The length of the initial type vector of each gene unit and the content of the vector value are the same, including one newly added gene identifier, one deleted gene identifier, four types of reference gene identifiers (A, G, C, T) and four types of variant gene identifiers (A, G, C, T). In addition, the length of the abnormal type vector of each gene unit is the same, and the difference between the abnormal type vectors of different gene units lies in the specific values ​​in the vector.

[0109] In the case where the gene unit is a newly added gene unit, the addition of the gene unit is characterized by setting the value of the newly added gene identification bit of the initial type vector and the value of the identification bit of the four types of variant gene identification bits (A, G, C, T) that matches the type of the newly added gene unit. In this embodiment, the value of the newly added gene identification bit and the value of the variant gene identification bit that matches the type of the newly added gene unit are set to a first preset value, such as 1, and the values ​​of other identification bits in the initial type vector are set to a second preset value, such as 0, to achieve numerical encoding of the initial type vector and obtain an abnormal type vector of the gene unit.

[0110] In the case where the gene unit is a deleted gene unit, the deletion of the gene unit is characterized by setting the value of the deleted gene identification bit in the initial type vector and the value of the identification bit that matches the type of the deleted gene unit in the four types of reference gene identification bits (A, G, C, T). In this embodiment, the value of the deleted gene identification bit and the value of the reference gene identification bit that matches the type of the deleted gene unit are set to a first preset value, such as 1, and the values ​​of other identification bits in the initial type vector are set to a second preset value, such as 0, to achieve numerical encoding of the initial type vector and obtain an abnormal type vector of the gene unit.

[0111] In the case where the gene unit is a replacement gene unit, the historical gene type of the gene unit, that is, the type of the gene unit before replacement, is determined, and the modification of the gene unit is characterized by setting the numerical value of the identification bit that matches the historical gene type in the four types of reference gene identification bits (A, G, C, T) in the initial type vector, and the numerical value of the identification bit that matches the replacement gene type in the four types of variant gene identification bits (A, G, C, T). In this embodiment, the numerical value of the reference gene identification bit that matches the historical gene type and the numerical value of the variant gene identification bit that matches the gene unit are set to a first preset value, such as 1, and the numerical values ​​of other identification bits in the initial type vector are set to a second preset value, such as 0, to achieve numerical encoding of the initial type vector and obtain an abnormal type vector of the gene unit.

[0112] When the gene unit has not mutated, the values ​​of the newly added gene identifier of the initial type vector and the values ​​of the identifiers of the four types of mutated gene identifiers (A, G, C, T) that match the type of the newly added gene unit are all set to 0 to indicate that the gene unit has not mutated.

[0113] by Figure 3 Taking the abnormal sequence information shown in the figure as an example, the first preset value is 1 and the second preset value is 0. Figure 5 Schematic diagram of the abnormal feature matrix output by the encoder. Figure 5 The abnormal feature matrix in consists of 11 rows, each row represents a Figure 3A gene unit in Figure 5 The abnormal feature matrix in the data includes 35 columns. Columns 1-10 are the variation of the characteristic gene units, which are four types of reference gene identifiers (A, G, C, T), newly added gene unit identifiers, four types of variant gene identifiers (A, G, C, T), and deleted gene unit identifiers; columns 11-34 are the chromosomal positions of the characteristic gene units, which are 22 autosomes and two sex chromosomes; column 35 is the absolute gene position of the characteristic gene units.

[0114] Figure 5 The gene unit represented by the first row is Figure 3 The gene unit represented by the first column in is the gene unit that has not mutated. Then in the first 10 columns of the first row, the value corresponding to each identification bit is 0. Figure 5 The gene units in the sixth and eleventh rows are Figure 3 The gene units represented by the sixth and eleventh columns in , that is, the gene units that have not mutated, have specific data setting methods as above and will not be repeated here.

[0115] Figure 5 The gene unit represented by the second row is Figure 3 The gene unit represented by the second column in the figure is the gene unit that has undergone gene mutation, from G base to C base. Then in the first 10 columns of the second row, the values ​​corresponding to the reference gene identifier corresponding to the G base (the second column) and the variant gene identifier corresponding to the C base (the eighth column) are 1, and the values ​​of other columns are 0.

[0116] Figure 5 The gene unit represented by the third row is Figure 3 The gene unit represented by the third column in the , that is, the gene unit where gene deletion occurs, then in the first 10 columns of the third row, the values ​​corresponding to the reference gene identifier (third column) and the deleted gene identifier (tenth column) corresponding to the deleted gene unit (C base) are 1, and the values ​​of other columns are 0; Figure 5 The gene units in the fourth and fifth rows are Figure 3 The gene units represented by the fourth and fifth columns in the , that is, the gene units where gene deletion occurs, are set in the same way as above and will not be repeated here.

[0117] Figure 5 The gene unit represented by the seventh row is Figure 3 The gene unit represented by the seventh column in the is the newly added gene unit. Then, in the first 10 columns of the third row, the value corresponding to the value of the reference gene identifier (eighth column) corresponding to the newly added gene unit (C base) is 1, and the values ​​of other columns are 0; Figure 5The gene units in the eighth to tenth rows are Figure 3 The gene units represented by the eighth to tenth columns in , that is, the newly added gene units, have the same data setting method as above and will not be repeated here.

[0118] In the above embodiment, for any gene unit, an initial type vector of the gene unit is constructed, and the numerical values ​​of the newly added gene identification bit, the deleted gene identification bit, multiple types of reference gene identification bits and multiple types of mutant gene identification bits in the initial type vector are set to represent the unchanged state of the gene unit, as well as the gene mutation state such as addition, deletion and replacement of the gene unit, so as to standardize the mutation state of each gene unit.

[0119] In one embodiment, based on the chromosome position of each gene unit, generating a position vector of each gene unit includes:

[0120] For any gene unit, an initial position vector of the gene unit is constructed; the initial position vector includes multiple candidate chromosome identification positions; among the multiple candidate chromosome identification positions, a target chromosome identification position corresponding to the chromosome position of the gene unit is determined, and the numerical value of the target chromosome identification position is set to a first preset value, and the numerical values ​​of other chromosome identification positions in the initial position vector are set to a second preset value, so as to obtain a position vector of the gene unit.

[0121] It should be noted that regardless of whether the gene unit mutates, there is a corresponding chromosome position, and the number of chromosomes of the target object is fixed, which is 24 chromosomes.

[0122] Continue with Figure 3 The abnormal sequence information shown is shown in the figure. The first preset value is 1 and the second preset value is 0. Figure 5 The abnormal feature matrix shown, Figure 5 The chromosomal locations of the 11-34 characterization gene units are 22 autosomes and two sex chromosomes.

[0123] Figure 5 The gene unit represented by the first row is Figure 3 The gene unit represented by the first column in is the gene unit with chromosome position 0. Then in columns 11-34 of the first row, the candidate chromosome identification bit in column 11 is determined as the target chromosome identification bit, and the value of column 11 is set to 1, and the values ​​of columns 12-34 are set to 0.

[0124] Figure 5 The gene unit represented by the second row is Figure 3The gene unit represented by the second column in is the gene unit with chromosome position 1. Then in columns 11-34 of the first row, the candidate chromosome identification bit in column 12 is determined as the target chromosome identification bit, and the value of column 12 is set to 1, and the values ​​of columns 11 and 13-34 are set to 0. Figure 5 The gene units in the third to fifth rows are Figure 3 The specific data setting method for the gene units represented by the third to fifth columns in is the same as above and will not be repeated here.

[0125] Figure 5 The gene unit represented by the sixth row in Figure 3 The gene unit represented by the sixth column in is the gene unit at chromosome position 2. Then in columns 11-34 of the first row, the candidate chromosome identification bit in column 13 is determined as the target chromosome identification bit, and the value of column 13 is set to 1, and the values ​​of columns 11-12 and 14-34 are set to 0. Figure 5 The gene units in the seventh to eleventh rows are Figure 3 The specific data setting method for the gene units represented by the seventh to eleventh columns in is the same as above and will not be repeated here.

[0126] In the embodiment of the present application, for any gene unit, an initial position vector of the gene unit is constructed, and the chromosome position of the gene unit is represented by setting the numerical values ​​of multiple candidate chromosome identification bits in the initial position vector to standardize the representation of the chromosome position of each gene unit.

[0127] In one embodiment, Figure 6 As shown, the encoding network includes a text encoding layer and a numerical encoding layer, and extracts the detection state features of the detection state information, including:

[0128] The text features in the detection state information are extracted based on the text coding layer, and the numerical features in the detection state information are extracted based on the numerical coding layer; the detection state features are determined according to the text features and the numerical features.

[0129] Figure 6 In the embodiment, the encoding network includes a text encoding layer, a numerical encoding layer and a sequence anomaly encoding layer arranged in parallel. Among them, the text encoding layer includes a linear layer for extracting text features. In this embodiment, the input of the text encoding layer is the text feature information in the detection state information, and the output is the text features in the detection state information. The numerical encoding layer includes a linear layer for extracting numerical features. In this embodiment, the input of the numerical encoding layer is the numerical state information in the detection state information, and the output is the numerical features in the detection state information. The input of the sequence anomaly encoding layer is the abnormal sequence information, and the output is the abnormal feature matrix. The content of extracting features using onehot encoding can be found in the aforementioned embodiment.

[0130] It should be noted that the text coding layer, the numerical coding layer and the sequence anomaly coding layer extract features in parallel. In this embodiment, the text features extracted by the text coding layer and the numerical features extracted by the numerical coding layer are summarized to obtain the detection state features of the detection state information, and the abnormal feature matrix extracted by the sequence anomaly coding layer is used as the abnormal sequence features.

[0131] In the above embodiment, a text coding layer is used to extract text features in the detection state information, and a data coding layer is used to extract numerical features in the detection state information. This is equivalent to extracting different types of features in the detection state information in parallel through different coding layers, so as to improve the feature extraction speed of the detection state information, thereby improving the efficiency of the state prediction model in predicting the state.

[0132] In one embodiment, Figure 7 As shown, the decoding network includes a pooling layer and a classification layer; based on the decoding network, the multimodal biometric features are reduced in dimension and the target prediction state information of the target object is output, including:

[0133] Based on the pooling layer, the multimodal biological features are reduced in dimension to obtain the one-dimensional abnormal sequence features of the target object; based on the classification layer, the one-dimensional abnormal sequence features are reduced in dimension to predict the matching probability between the target object and each candidate state information in the candidate state information set, and the candidate state information whose matching probability is in a preset probability interval is determined as the target predicted state information of the target object.

[0134] Figure 7 The decoding network in includes a pooling layer and a classification layer connected in series. The pooling layer is connected to the conversion network. The pooling layer can be an average pooling layer, which is used to reduce the dimension of the multimodal biometric features output by the conversion network to obtain a one-dimensional abnormal sequence feature. The classification layer can include a linear layer, an activation layer, and a linear layer, which are used to further reduce the dimension of the one-dimensional abnormal sequence feature, predict the matching probability of the target object and each candidate state information according to the multimodal features of the target object, and use the candidate state information corresponding to one or more matching probabilities in a preset probability interval as the target predicted state information.

[0135] In a feasible scenario, the candidate state information corresponding to the maximum matching probability and the matching probability of the candidate state information can be used as the target prediction state information of the target object, or the candidate state information and the matching probability of each candidate state information can be directly summarized as the target prediction state information.

[0136] In the above embodiment, the multimodal biometric features are reduced in dimension based on the pooling layer to fully integrate the multimodal biometric features, and then the one-dimensional abnormal sequence features obtained by the dimensionality reduction processing are reduced in dimension again based on the classification layer to accurately predict the matching probability between the target object and each candidate state information in the candidate state information set, and the candidate state information whose matching probability is in a preset probability interval is determined as the target predicted state information of the target object, thereby providing a true and comprehensive judgment basis for the state prediction result of the target object and improving the accuracy of the state prediction result.

[0137] In one embodiment, a method for training a state prediction model includes:

[0138] A state information data set is obtained; the state information data set includes historical abnormal sequence information, historical detection state information, and historical state information labels of multiple objects; the historical abnormal sequence information and the historical detection state information are predicted by an initial state prediction model to obtain historical predicted state information, and the initial state prediction model is trained according to the loss value of the historical predicted state information and the historical state information label to obtain a state prediction model.

[0139] The content of the status information dataset is constructed based on historical data, including actual abnormal sequence information of multiple objects, actual detection status information, and actual status information labels.

[0140] The initial state prediction model includes an initial encoding network, an initial conversion network and an initial decoding network. The initial encoding network includes an initial sequence anomaly encoding layer, an initial text encoding layer and an initial numerical encoding layer. The initial decoding network includes a pooling layer and an initial classification layer.

[0141] The state information data set is input into the initial state prediction model, and the initial state prediction model is used to learn the logical relationship between the historical abnormal sequence information and the historical detection state information of each object and the historical state information label of the object to obtain the state prediction model.

[0142] The training process of the initial state prediction model refers to: the initial state prediction model predicts the historical prediction state information based on the historical abnormal sequence information and the historical detection state information, inputs the historical prediction state information and the historical state information label into the loss function to obtain the model prediction accuracy, and optimizes the parameters of the initial state prediction model according to the model prediction accuracy until the loss value is less than the preset threshold to obtain the state prediction model.

[0143] In the above embodiment, the state information data set constructed by using the historical abnormal sequence information, historical detection state information and historical state information labels of multiple objects is used to train the initial state prediction model to obtain the state prediction model. This end-to-end training method reduces the difficulty of obtaining the state information data set and can provide a more diverse data set for model training, thereby improving the generalization ability of the state prediction model. In addition, the trained state prediction model can be flexibly deployed on various mobile terminals according to actual application needs, further improving the application scenarios of the state prediction model, thereby improving the scenario applicability of the state prediction method.

[0144] In one embodiment, the state prediction model is a gene disease prediction model; the method further comprises:

[0145] Obtain the gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence of the target object; obtain the clinical information and imaging information of the target object; input the gene reference sequence, gene variation sequence, chromosome position sequence, gene absolute position sequence, clinical information and imaging information into the gene disease prediction model to extract the multimodal biological characteristics of the target object, and analyze the multimodal biological characteristics to predict the type of genetic disease of the target object.

[0146] Based on the biological tissue sample of the target object, a gene sequencing tool is used to sequence the biological tissue sample to obtain the initial gene sequence of the target object. The initial gene sequence is then compared with the reference gene sequence to construct the gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence of the target object.

[0147] The original clinical omics information of the target object, such as gender, smoking status, age, etc., is collected by questionnaire survey, and the original clinical omics information of gender and smoking status is digitized, and the digitized clinical omics information and age are used as the clinical information of the target object; the original imaging omics information of the target object, such as MRI images, is collected by medical scanning, and the original imaging omics information is feature extracted and vectorized to obtain imaging information.

[0148] The clinical information, imaging information, gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence are input into the gene disease prediction model in turn, the state features in the clinical information and imaging information are extracted through the gene disease prediction model, the sequence features of the gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence are extracted, and then the state features and sequence features are fused to obtain the multimodal features of the target object. Finally, the multimodal biological features are decoded and classified to obtain the probability of the target object and each gene disease type, and the gene disease type corresponding to the maximum probability is output as the gene disease type of the target object.

[0149] In the above embodiment, the gene reference sequence, gene variation sequence, chromosome position sequence, gene absolute position sequence, clinical information and imaging information of the target object are input into the gene disease prediction model to instruct the gene disease prediction model to predict the gene disease type of the target object based on the multimodal biological characteristics of the target object. In this way, the gene disease type predicted by comprehensively considering multiple factors of imaging gene diseases is more accurate.

[0150] In one embodiment, the state prediction method can be applied to the scenario of genetic disease prediction, such as Figure 8 As shown, the following steps are included:

[0151] Step S801, obtaining sequencing data of a target object.

[0152] When an initial gene sequence is obtained through second-generation or third-generation gene sequencing, the initial gene sequence is preprocessed, including but not limited to quality control, deduplication, etc., to obtain sequencing data of the target object.

[0153] Step S802: sequence alignment.

[0154] Align the sequencing data to the reference gene sequence.

[0155] Step S803, performing variation detection on the target gene sequence in the candidate site set to obtain abnormal sequence information.

[0156] Optionally, a gene variation detection tool is used to detect the two groups of gene sequences in sequence alignment, output the variation detection results, and then construct a gene reference sequence, a gene variation sequence, a chromosome position sequence and a gene absolute position sequence based on the variation detection results as abnormal sequence information.

[0157] Step S804, obtaining clinical omics information of the target object.

[0158] Clinical omics information includes categorical features and numerical features. Categorical features include gender and smoking status, and numerical features include age and microsatellite instability score (MSI). In practical applications, onehot mutation processing is used to digitize the categorical features in clinical omics features, and the digitized categorical features replace the categorical features in clinical omics information to obtain clinical omics information in numerical form.

[0159] Step S805: obtaining image omics information of the target object.

[0160] On the basis of obtaining the region of interest image of the target object, the feature value of the region of interest image is further obtained as the imageomics information of the target object.

[0161] Step S806, inputting abnormal sequence information, clinical omics information and image omics information into a genetic disease prediction model, and outputting the probability of the genetic disease type of the target object.

[0162] Among them, the genetic disease prediction model can be a transformer model, specifically a multimodal genetic disease prediction model obtained by training with historical data sets of a large number of objects before the genetic disease prediction model is actually put into use.

[0163] In the above embodiment, the genetic disease prediction model is used to predict the genetic disease of the target object based on the clinical information, genetic variation, imaging genomics and other information of the target object. This is equivalent to combining the macro-level tabular features of the target object with the micro-level genetic fine-grained features, so that the multimodal genetic disease discrimination model learns the degree of influence between various complex factors and genetic diseases, so as to make more accurate genetic disease prediction results.

[0164] In one embodiment, Fig. 9 As shown, the state prediction method includes the following steps:

[0165] Step S901, obtaining text status information, numerical status information and image status information of a target object.

[0166] Step S902: performing data mapping on the text state information according to the preset text state conversion relationship to obtain text feature information of the target object.

[0167] Step S903, vectorizing the image state information to obtain image data information of the image state information, and summarizing the numerical state information and the image data information to obtain numerical feature information of the target object.

[0168] Step S904, obtaining the initial gene sequence of the target object, performing quality control processing on the initial gene sequence, and obtaining gene sequence information.

[0169] Step S905 , comparing the target gene sequence in the candidate site set in the gene sequence information with the reference gene sequence, and constructing a gene reference sequence, a gene variation sequence, a chromosome position sequence and a gene absolute position sequence.

[0170] Step S906: extracting text features of the text feature information based on the text coding layer.

[0171] Step S907: extracting numerical features of the numerical feature information based on the numerical coding layer.

[0172] Step S908: extracting abnormal sequence features based on the sequence abnormality coding layer.

[0173] Step S909, based on the conversion network, the text features, numerical features and abnormal sequence features are spliced, and the spliced ​​features are subjected to dimensionality upgrading processing to obtain multimodal biometric features.

[0174] Step S910, reducing the dimension of the multimodal biometric features based on the pooling layer.

[0175] Step S911, predicting the multimodal biometric features after dimensionality reduction based on the classification layer, and outputting target prediction state information of the target object.

[0176] It should be noted that the above steps S901-S903 are steps for obtaining detection status information, and the above steps S904-S905 are steps for obtaining gene sequence information. In actual applications, the acquisition process of detection status information and gene sequence information can be executed in parallel or in sequence. The above steps S906-S911 are the state prediction flow chart of the preset state prediction model.

[0177] In the embodiment of the present application, the method of extracting, analyzing and predicting the gene sequence information and detection state information of the target object through each network layer in the preset state prediction model can avoid the interference of subjective factors to a certain extent and ensure the objectivity of the prediction results of the state prediction model. In addition, when performing state prediction, the macro-level table features of the target object and the micro-level genetic fine-grained features are comprehensively considered to comprehensively and accurately characterize the actual abnormal state of the target object, thereby improving the authenticity and accuracy of the target prediction state information obtained by multimodal biometric analysis.

[0178] In a specific embodiment, the state prediction method can be applied to a pan-gene disease identification scenario, in which the state prediction model is a gene type prediction model for multiple types of diseases, and the target prediction state information is a single gene disease type.

[0179] First, the computer equipment obtains multimodal information of the target object, namely, clinical omics information, image omics information corresponding to multiple parts of the target object, and gene sequences corresponding to multiple parts of the target object, and processes the acquired information to obtain various types of model input information to facilitate identification and processing by the disease type prediction model.

[0180] Among them, clinical omics information includes gender information, information on whether there is smoking behavior, age information and microsatellite instability score information. The clinical omics information is processed in the following ways: gender is numerically processed, specifically using 1 to represent male gender and 0 to represent female gender; the information on whether the target object has smoking behavior is numerically processed, specifically using 1 to represent the presence of smoking behavior and 0 to represent the absence of smoking behavior.

[0181] Imageomics information includes, but is not limited to, lung CT scan image information, brain CT scan image information, etc. of the target object. The imageomics information is processed, specifically, image features of the lung CT scan image are extracted and vectorized to obtain a first feature vector, and image features of the brain CT scan image are extracted and vectorized to obtain a second feature vector.

[0182] The gene sequences corresponding to multiple parts of the target object are processed, specifically, from the whole gene sequence of the target object, a variation detection tool is used to extract the target gene sequences corresponding to multiple parts, and the target gene sequence is subjected to variation detection to obtain the variation detection result; then, the variation detection result is parsed according to the preset conversion conditions to construct a gene reference sequence, a gene variation sequence, a chromosome position sequence, and a gene absolute position sequence. It should be noted that the gene reference sequence, the gene variation sequence, the chromosome position sequence, and the gene absolute position sequence have the same sequence length. In the embodiment of the present application, the above sequences can be summarized into a fourth-order gene sequence matrix.

[0183] Then, the gender information, the judgment information of whether there is smoking behavior, the age information, the microsatellite instability score information, the first eigenvector, the second eigenvector and the gene sequence matrix after numerical processing are input into the gene type prediction model to obtain the probability of the target object and multiple gene disease types output by the gene type prediction model. Finally, the gene disease type corresponding to the maximum probability is used as the gene disease type of the target object.

[0184] It should be noted that clinical omics information and image omics information can be selected alternatively. For example, using clinical omics information and gene sequences to predict genetic diseases, we continue to take male gender (mapped value is 1), no smoking behavior (mapped value is 0), age is 60, and MSI score is 0.5 as an example. Fig.10 , input the gene sequence matrix of "1", "0", "60", "0.5" and the fourth order into the preset disease stage prediction model. Fig.10 In the schematic diagram of the architecture of the gene disease type prediction model shown, the gene disease type prediction model performs the following encoding operations in parallel: extracting features corresponding to gender and whether there is smoking behavior based on the text encoding layer, extracting features of age and MSI score based on the numerical encoding layer, and extracting features of the gene sequence matrix based on the sequence abnormality encoding layer. Then, the features extracted from the above three encoding layers are fused through the core network (that is, the conversion network in the aforementioned embodiment), and the fused multimodal features are extracted. The core network includes multiple transformer layers. Then, the multimodal features are reduced in dimension through the average pooling layer in the decoding network, and the matching probability of the reduced multimodal features and multiple gene disease types is predicted through the classification layer, and the gene disease type corresponding to the maximum matching probability is output as the target gene disease type.

[0185] In the above-mentioned pan-gene disease identification scenario, the state prediction method provided in the embodiment of the present application is adopted, which comprehensively considers the clinical genomics information of the target object, the image genomics information corresponding to multiple parts of the target object, and the multimodal biological characteristics including the gene sequence, and accurately locates the potential genetic disease information corresponding to the current state of the target object, providing a reliable reference basis for subsequent disease prevention.

[0186] In a specific embodiment, the state prediction method can be applied to a specific biological tissue, such as a genetic disease identification scenario of the lungs. In this scenario, the state prediction model is a prediction model for the genetic disease type of the lungs, and the target prediction state information is a certain lung genetic disease. The computer device obtains the target object's gender, smoking behavior, age, symptom duration, lung image, and target gene sequence matching the lung disease. Next, the gender is mapped, using 1 to represent male gender and 0 to represent female gender; the smoking behavior is mapped, using 1 to represent the presence of smoking behavior and 0 to represent the absence of smoking behavior; the unit of symptom duration is unified into days; the lung image feature vector is extracted; and based on the target gene sequence, a lung gene reference sequence, a lung gene variation sequence, a lung gene chromosome position sequence, and a lung gene absolute position sequence are constructed. Then, the mapped gender data, mapped smoking behavior characteristic data, age, symptom duration, lung image feature vector, gene reference sequence, lung gene variation sequence, lung gene chromosome position sequence and lung gene absolute position sequence are input into the prediction model of genetic disease type, and the gender data and smoking behavior characteristic data are encoded through the category coding layer of the model to obtain category features, and the age, symptom duration and lung image feature vector are encoded through the numerical coding layer of the model to obtain numerical features, and the gene reference sequence, lung gene variation sequence, lung gene chromosome position sequence and lung gene absolute position sequence are encoded through the abnormal gene coding layer of the model to obtain numerical features. The chromosome position sequence and the lung gene absolute position sequence are encoded to obtain the variation features. The category features, numerical features and variation features are spliced ​​through the core network of the model to further extract the multimodal features of the target object for the lung, and the multimodal features are reduced in dimension through the average pooling layer of the model. The reduced features are correlated with the features of multiple lung gene diseases through the classification module of the model to obtain the matching degree between the target and each lung gene disease, and the lung gene disease with the highest matching degree is determined as the target lung gene disease of the target object, providing a reliable reference for the subsequent formulation of the target object disease treatment plan.

[0187] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0188] Based on the same inventive concept, the embodiment of the present application also provides a state prediction device for implementing the state prediction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more state prediction device embodiments provided below can refer to the limitations of the state prediction method above, and will not be repeated here.

[0189] In one embodiment, Fig.11 As shown, a state prediction device is provided, including: a sequence acquisition module 1101, a state acquisition module 1102 and a state analysis module 1103, wherein:

[0190] The sequence acquisition module 1101 is used to acquire the gene sequence information of the target object, compare the gene sequence information with the reference gene sequence, and acquire abnormal sequence information in the gene sequence information; the abnormal sequence information includes the abnormal position and abnormal type of the gene unit;

[0191] A state acquisition module 1102 is used to acquire detection state information of a target object;

[0192] The state analysis module 1103 is used to input the abnormal sequence information and the detection state information into the state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtain the target prediction state information of the target object based on the abnormal sequence features and the detection state features.

[0193] In one embodiment, the sequence acquisition module 1101 includes:

[0194] An initial sequence acquisition unit, used to acquire an initial gene sequence of a target object, the initial gene sequence including gene units of multiple sites;

[0195] The sequence quality control processing unit is used to perform quality control processing on the initial gene sequence to obtain gene sequence information.

[0196] In one embodiment, the sequence acquisition module 1101 includes:

[0197] The first sequence alignment unit is used to align the gene sequence information with the reference gene sequence to obtain the target gene sequence in which the gene sequence information is in the candidate site set;

[0198] A gene unit comparison unit, used to compare each gene unit on the target gene sequence with the corresponding gene unit on the reference gene sequence, and determine normal gene units and abnormal gene units in the target gene sequence;

[0199] An abnormal sequence construction unit, used to construct an abnormal type sequence and an abnormal position sequence of a target object according to a reference gene sequence, a normal gene unit and an abnormal gene unit;

[0200] The second sequence comparison unit is used to compare the abnormal type sequence and the abnormal position sequence according to the absolute gene position of the gene unit in the reference gene sequence to obtain abnormal sequence information.

[0201] In one embodiment, the abnormal gene unit includes at least one of a newly added gene unit, a deleted gene unit and a replaced gene unit; the abnormal type sequence includes a gene reference sequence and a gene variation sequence; the abnormal position sequence includes a chromosome position sequence and a gene absolute position sequence; the abnormal sequence construction unit is used to perform the following steps according to the gene absolute position of the gene unit in the reference gene sequence: sorting and combining the type identifier of each gene unit in the reference gene sequence corresponding to the target gene sequence and the gene addition identifier of each newly added gene unit to obtain a gene reference sequence; sorting and combining the type identifier of the normal gene unit, the type identifier of the replaced gene unit, the type identifier of the newly added gene unit and the gene deletion identifier of each deleted gene unit to obtain a gene variation sequence; sorting and combining the chromosome position corresponding to each gene unit in the gene variation sequence and the chromosome position of each deleted gene unit to obtain a chromosome position sequence; and sorting and combining the gene absolute position of each gene unit in the gene variation sequence and the gene absolute position of each deleted gene unit to obtain a gene absolute position sequence.

[0202] In one embodiment, the status acquisition module 1102 includes:

[0203] An information acquisition unit, used to acquire text state information, numerical state information and image state information of a target object;

[0204] An information mapping unit, used to perform data mapping on the text state information according to a preset text state conversion relationship to obtain text feature information of the target object;

[0205] A vectorization processing unit is used to perform vectorization processing on the image state information to obtain image data information of the image state information, and to summarize the numerical state information and the image data information to obtain numerical feature information of the target object;

[0206] The information summarizing unit is used to obtain the detection status information based on the text feature information and the numerical feature information.

[0207] In one embodiment, the state prediction model includes an encoding network, a conversion network, and a decoding network; the state analysis module 1103 includes:

[0208] A feature encoding unit, used for extracting abnormal sequence features of abnormal sequence information based on the encoding network, and extracting detection state features of detection state information;

[0209] A feature concatenation unit is used to concatenate abnormal sequence features and detection status features based on a conversion network, and to perform dimensionality-upgrading processing on the concatenated features to obtain multimodal biological features;

[0210] The feature decoding unit is used to perform dimensionality reduction processing on the multimodal biometric features based on the decoding network and output the target prediction state information of the target object.

[0211] In one embodiment, the coding network includes a sequence anomaly coding layer; the abnormal sequence information includes a gene reference sequence, a gene variation sequence, a chromosome position sequence and a gene absolute position sequence; the feature coding unit is also used to identify the gene reference sequence and the gene variation sequence through the sequence anomaly coding layer, determine the abnormal type of each gene unit in the abnormal sequence information, generate an abnormal type vector of each gene unit based on the abnormal type of each gene unit, and combine the abnormal type vectors of each gene unit to obtain a type matrix of the abnormal sequence information; identify the chromosome position sequence through the sequence anomaly coding layer, determine the chromosome position of each gene unit in the abnormal sequence information, generate a position vector of each gene unit based on the chromosome position of each gene unit, and combine the position vectors of each gene unit to obtain a chromosome position matrix of the abnormal sequence information; splice the type matrix, the chromosome position matrix and the gene absolute position sequence to obtain an abnormal feature matrix, and determine the abnormal feature matrix as the abnormal sequence feature.

[0212] In one embodiment, the feature encoding unit is also used to construct an initial type vector of a gene unit for any gene unit; the initial type vector includes a newly added gene identifier, a deleted gene identifier, multiple types of reference gene identifiers, and multiple types of variant gene identifiers; if the gene unit is a newly added gene unit, the value of the newly added gene identifier in the initial type vector and the value of the variant gene identifier that matches the type of the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value to generate an abnormal type vector of the gene unit; if the gene unit is a deleted gene unit, the value of the deleted gene identifier in the initial type vector and the value of the reference gene identifier that matches the type of the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value to generate an abnormal type vector of the gene unit; if the gene unit is a replaced gene unit, the historical gene type of the gene unit is determined, the value of the reference gene identifier that matches the historical gene type and the value of the variant gene identifier that matches the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value to generate an abnormal type vector of the gene unit.

[0213] In one embodiment, the feature encoding unit is also used to construct an initial position vector of a gene unit for any gene unit; the initial position vector includes multiple candidate chromosome identification positions; among the multiple candidate chromosome identification positions, the target chromosome identification position corresponding to the chromosome position of the gene unit is determined, and the numerical value of the target chromosome identification position is set to a first preset value, and the numerical values ​​of other chromosome identification positions in the initial position vector are set to a second preset value, so as to obtain the position vector of the gene unit.

[0214] In one embodiment, the encoding network includes a text encoding layer and a numerical encoding layer; the feature encoding unit is also used to extract text features in the detection state information based on the text encoding layer, and to extract numerical features in the detection state information based on the numerical encoding layer; the detection state features are determined based on the text features and the numerical features.

[0215] In one embodiment, the decoding network includes a pooling layer and a classification layer; a feature decoding unit is used to perform dimensionality reduction processing on the multimodal biometric features based on the pooling layer to obtain a one-dimensional abnormal sequence feature of the target object; based on the classification layer, the one-dimensional abnormal sequence feature is reduced in dimension to predict the matching probability between the target object and each candidate state information in the candidate state information set, and the candidate state information whose matching probability is in a preset probability interval is determined as the target predicted state information of the target object.

[0216] In one embodiment, the state prediction device comprises:

[0217] A data set acquisition module is used to acquire a state information data set; the state information data set includes historical abnormal sequence information, historical detection state information and historical state information labels of multiple objects;

[0218] The model training module is used to predict the historical abnormal sequence information and the historical detection state information through the initial state prediction model to obtain the historical prediction state information, and train the initial state prediction model according to the loss value of the historical prediction state information and the historical state information label to obtain the state prediction model.

[0219] In one embodiment, the state prediction device comprises:

[0220] A genomics information acquisition module is used to obtain the gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence of the target object;

[0221] Radiomics information acquisition module, used to obtain clinical information and imaging information of the target object;

[0222] The disease type prediction module is used to input gene reference sequences, gene variation sequences, chromosome position sequences, gene absolute position sequences, clinical information and imaging information into the gene disease prediction model to extract the multimodal biological characteristics of the target object, and analyze the multimodal biological characteristics to predict the gene disease type of the target object.

[0223] The above-mentioned state prediction device can avoid the interference of subjective factors to a certain extent and ensure the objectivity of the target state prediction information by extracting, analyzing and predicting the characteristics of the gene sequence information and detection state information of the target object through a preset state prediction model. In addition, when performing state prediction, the detection state information of the target object, the abnormal position of the abnormal gene unit in the gene sequence, and the multimodal information of the abnormal type are comprehensively considered to comprehensively and accurately characterize the actual abnormal state characteristics of the target object, so that the multimodal biological characteristics extracted by the state prediction model are more in line with the actual abnormal state of the target object, thereby improving the authenticity and accuracy of the target prediction state information obtained by the state prediction model from the analysis of the multimodal biological characteristics.

[0224] Each module in the above-mentioned state prediction device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0225] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.12As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store state prediction data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a state prediction method is implemented.

[0226] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.13 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a state prediction method is implemented. The display unit of the computer device is used to form a visually visible image, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse, etc.

[0227] Those skilled in the art will understand that Fig.12 and Fig.13The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0228] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0229] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0230] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0231] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0232] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0233] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0234] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A state prediction method, characterized in that: The method comprises: Obtaining gene sequence information of a target object, performing sequence comparison between the gene sequence information and a reference gene sequence, and obtaining a target gene sequence in which the gene sequence information is in a set of candidate sites; Comparing each gene unit on the target gene sequence with the corresponding gene unit on the reference gene sequence to determine normal gene units and abnormal gene units in the target gene sequence; the abnormal gene unit includes at least one of a newly added gene unit, a deleted gene unit and a replaced gene unit; Constructing an abnormal type sequence and an abnormal position sequence of the target object according to the reference gene sequence, the normal gene unit and the abnormal gene unit; According to the absolute gene position of the gene unit in the reference gene sequence, the abnormal type sequence and the abnormal position sequence are compared to obtain abnormal sequence information; the abnormal type sequence includes a gene reference sequence and a gene variation sequence; the gene reference sequence is obtained by sorting and combining the type identifier of each gene unit in the reference gene sequence corresponding to the target gene sequence and the gene addition identifier of each newly added gene unit; the gene variation sequence is obtained by sorting and combining the type identifier of the normal gene unit in the target gene sequence, the type identifier of the replacement gene unit, the type identifier of the newly added gene unit and the gene deletion identifier of each deleted gene unit; Acquire detection status information of the target object; The abnormal sequence information and the detection state information are input into a state prediction model to extract abnormal sequence features of the abnormal sequence information and detection state features of the detection state information, and the target prediction state information of the target object is obtained based on the abnormal sequence features and the detection state features.

2. The method according to claim 1, characterized in that The step of obtaining the gene sequence information of the target object includes: Acquire an initial gene sequence of the target object, wherein the initial gene sequence includes gene units of multiple sites; The initial gene sequence is subjected to quality control processing to obtain the gene sequence information.

3. The method according to claim 1, characterized in that The abnormal position sequence includes a chromosome position sequence and a gene absolute position sequence; the abnormal position sequence is obtained by: According to the absolute gene position of the gene unit in the reference gene sequence, the chromosome position corresponding to each gene unit in the gene variation sequence and the chromosome position of each deleted gene unit are sorted and combined to obtain the chromosome position sequence; and the absolute gene position of each gene unit in the gene variation sequence and the absolute gene position of each deleted gene unit are sorted and combined to obtain the absolute gene position sequence.

4. The method according to claim 1, characterized in that: The acquiring the detection status information of the target object includes: Acquire text status information, numerical status information and image status information of the target object; According to a preset text state conversion relationship, data mapping is performed on the text state information to obtain text feature information of the target object; Performing vectorization processing on the image state information to obtain image data information of the image state information, and summarizing the numerical state information and the image data information to obtain numerical feature information of the target object; The detection status information is obtained based on the text feature information and the numerical feature information.

5. The method according to any one of claims 1 to 4, characterized in that: The state prediction model includes an encoding network, a conversion network, and a decoding network; the abnormal sequence information and the detection state information are input into the state prediction model to extract abnormal sequence features of the abnormal sequence information and detection state features of the detection state information, and the target prediction state information of the target object is obtained based on the abnormal sequence features and the detection state features, including: Extracting abnormal sequence features of the abnormal sequence information based on the encoding network, and extracting detection state features of the detection state information; Based on the conversion network, the abnormal sequence feature and the detection state feature are spliced, and the spliced ​​features are subjected to dimensionality-increasing processing to obtain a multimodal biological feature; Based on the decoding network, the multimodal biometric feature is subjected to dimensionality reduction processing, and target prediction state information of the target object is output.

6. The method according to claim 5, characterized in that The coding network includes a sequence anomaly coding layer; the abnormal sequence information includes a gene reference sequence, a gene variation sequence, a chromosome position sequence and a gene absolute position sequence; the abnormal sequence features of the abnormal sequence information extracted based on the coding network include: The gene reference sequence and the gene variation sequence are identified by the sequence abnormality coding layer, the abnormal type of each gene unit in the abnormal sequence information is determined, based on the abnormal type of each gene unit, an abnormal type vector of each gene unit is generated, and the abnormal type vectors of each gene unit are combined to obtain a type matrix of the abnormal sequence information; The chromosome position sequence is identified by the sequence abnormality encoding layer, the chromosome position of each gene unit in the abnormal sequence information is determined, based on the chromosome position of each gene unit, a position vector of each gene unit is generated, and the position vectors of each gene unit are combined to obtain a chromosome position matrix of the abnormal sequence information; The type matrix, the chromosome position matrix and the gene absolute position sequence are spliced ​​to obtain an abnormal feature matrix, and the abnormal feature matrix is ​​determined as the abnormal sequence feature.

7. The method according to claim 6, characterized in that The generating of the abnormal type vector of each gene unit based on the abnormal type of each gene unit comprises: For any gene unit, construct an initial type vector of the gene unit; the initial type vector includes a newly added gene identifier, a deleted gene identifier, multiple types of reference gene identifiers, and multiple types of variant gene identifiers; If the gene unit is a newly added gene unit, the value of the newly added gene identification bit in the initial type vector and the value of the variant gene identification bit matching the type of the gene unit are set to a first preset value, and the values ​​of other identification bits in the initial type vector are set to a second preset value to generate an abnormal type vector of the gene unit; If the gene unit is a deleted gene unit, the value of the deleted gene identification bit in the initial type vector and the value of the reference gene identification bit matching the type of the gene unit are set to the first preset value, and the values ​​of other identification bits in the initial type vector are set to the second preset value to generate an abnormal type vector of the gene unit; If the gene unit is a replacement gene unit, the historical gene type of the gene unit is determined, the numerical value of the reference gene identification bit that matches the historical gene type and the numerical value of the variant gene identification bit that matches the gene unit are set to the first preset value, and the numerical values ​​of other identification bits in the initial type vector are set to the second preset value to generate an abnormal type vector of the gene unit.

8. The method according to claim 6, characterized in that The step of generating a position vector of each gene unit based on the chromosome position of each gene unit comprises: For any gene unit, construct an initial position vector of the gene unit; the initial position vector includes a plurality of candidate chromosome identification positions; Among the multiple candidate chromosome identifiers, determine the target chromosome identifier corresponding to the chromosome position of the gene unit, set the value of the target chromosome identifier to a first preset value, and set the values ​​of other chromosome identifiers in the initial position vector to a second preset value to obtain the position vector of the gene unit.

9. The method according to claim 5, characterized in that The encoding network includes a text encoding layer and a numerical encoding layer; the extraction of the detection state feature of the detection state information includes: Extracting text features in the detection status information based on the text encoding layer, and extracting numerical features in the detection status information based on the numerical encoding layer; The detection status feature is determined according to the text feature and the numerical feature.

10. The method according to claim 5, characterized in that The decoding network includes a pooling layer and a classification layer; the dimensionality reduction processing of the multimodal biometric features based on the decoding network and outputting the target prediction state information of the target object includes: Based on the pooling layer, the multimodal biometric features are subjected to dimensionality reduction processing to obtain a one-dimensional abnormal sequence feature of the target object; Based on the classification layer, the one-dimensional abnormal sequence features are reduced in dimension, the matching probability between the target object and each candidate state information in the candidate state information set is predicted, and the candidate state information whose matching probability is in a preset probability interval is determined as the target predicted state information of the target object.

11. The method according to any one of claims 1 to 4, characterized in that: The training method of the state prediction model includes: Acquire a state information data set; the state information data set includes historical abnormal sequence information, historical detection state information and historical state information labels of multiple objects; The historical abnormal sequence information and the historical detection state information are predicted by a state prediction model to obtain historical predicted state information, and the state prediction model is trained according to the loss value of the historical predicted state information and the historical state information label to obtain the state prediction model.

12. The method according to any one of claims 1 to 4, characterized in that: The state prediction model is a gene disease prediction model; the method further comprises: Obtaining a gene reference sequence, a gene variation sequence, a chromosome position sequence, and a gene absolute position sequence of the target object; Acquiring clinical information and imaging information of the target object; The gene reference sequence, the gene variation sequence, the chromosome position sequence, the gene absolute position sequence, the clinical information and the imaging information are input into the gene disease prediction model to extract the multimodal biological characteristics of the target object, and the multimodal biological characteristics are analyzed to predict the gene disease type of the target object.

13. A state prediction device, characterized in that: The device comprises: A sequence acquisition module is used to acquire gene sequence information of a target object, compare the gene sequence information with a reference gene sequence, and acquire the target gene sequence in which the gene sequence information is in a candidate site set; compare each gene unit on the target gene sequence with the corresponding gene unit on the reference gene sequence, and determine normal gene units and abnormal gene units in the target gene sequence; the abnormal gene unit includes at least one of a newly added gene unit, a deleted gene unit, and a replaced gene unit; construct an abnormal type sequence and an abnormal position sequence of the target object according to the reference gene sequence, the normal gene unit, and the abnormal gene unit; and The abnormal type sequence and the abnormal position sequence are compared at the absolute position of the gene in the reference gene sequence to obtain abnormal sequence information; the abnormal type sequence includes a gene reference sequence and a gene variation sequence; the gene reference sequence is obtained by sorting and combining the type identifier of each gene unit in the reference gene sequence corresponding to the target gene sequence and the gene addition identifier of each newly added gene unit; the gene variation sequence is obtained by sorting and combining the type identifier of the normal gene unit in the target gene sequence, the type identifier of the replacement gene unit, the type identifier of the newly added gene unit and the gene deletion identifier of each deleted gene unit; A status acquisition module, used to acquire detection status information of the target object; The state analysis module is used to input the abnormal sequence information and the detection state information into a state prediction model to extract the abnormal sequence features of the abnormal sequence information and the detection state features of the detection state information, and obtain the target prediction state information of the target object based on the abnormal sequence features and the detection state features.

14. The device according to claim 13, characterized in that The sequence acquisition module comprises: An initial sequence acquisition unit, used to acquire an initial gene sequence of the target object, wherein the initial gene sequence includes gene units of multiple sites; A sequence quality control processing unit is used to perform quality control processing on the initial gene sequence to obtain the gene sequence information.

15. The device according to claim 13, characterized in that The abnormal position sequence includes a chromosome position sequence and a gene absolute position sequence; The sequence acquisition module is used to sort and combine the chromosome position corresponding to each gene unit in the gene variation sequence and the chromosome position of each deleted gene unit according to the absolute gene position of the gene unit in the reference gene sequence to obtain the chromosome position sequence; and sort and combine the absolute gene position of each gene unit in the gene variation sequence and the absolute gene position of each deleted gene unit to obtain the absolute gene position sequence.

16. The device according to claim 13, characterized in that The state acquisition module includes: An information acquisition unit, used to acquire text status information, numerical status information and image status information of the target object; An information mapping unit, used to perform data mapping on the text state information according to a preset text state conversion relationship to obtain text feature information of the target object; A vectorization processing unit, used for performing vectorization processing on the image state information to obtain image data information of the image state information, and summarizing the numerical state information and the image data information to obtain numerical feature information of the target object; An information summarizing unit is used to obtain the detection status information based on the text feature information and the numerical feature information.

17. The device according to any one of claims 13 to 16, characterized in that: The state prediction model includes a coding network, a conversion network and a decoding network; the state analysis module includes: a feature encoding unit, configured to extract an abnormal sequence feature of the abnormal sequence information based on the encoding network, and to extract a detection state feature of the detection state information; A feature concatenation unit, configured to concatenate the abnormal sequence feature and the detection state feature based on the conversion network, and perform dimensionality increase processing on the concatenated features to obtain a multimodal biometric feature; A feature decoding unit is used to perform dimensionality reduction processing on the multimodal biometric feature based on the decoding network and output target prediction state information of the target object.

18. The device according to claim 17, characterized in that The coding network includes a sequence anomaly coding layer; the abnormal sequence information includes a gene reference sequence, a gene variation sequence, a chromosome position sequence and a gene absolute position sequence; The feature encoding unit is also used to identify the gene reference sequence and the gene variation sequence through the sequence anomaly encoding layer, determine the anomaly type of each gene unit in the abnormal sequence information, generate an anomaly type vector of each gene unit based on the anomaly type of each gene unit, and combine the anomaly type vectors of each gene unit to obtain a type matrix of the abnormal sequence information; identify the chromosome position sequence through the sequence anomaly encoding layer, determine the chromosome position of each gene unit in the abnormal sequence information, generate a position vector of each gene unit based on the chromosome position of each gene unit, and combine the position vectors of each gene unit to obtain a chromosome position matrix of the abnormal sequence information; splice the type matrix, the chromosome position matrix and the gene absolute position sequence to obtain an abnormal feature matrix, and determine the abnormal feature matrix as the abnormal sequence feature.

19. The device according to claim 18, characterized in that The feature encoding unit is also used to construct an initial type vector of any gene unit; the initial type vector includes a newly added gene identifier, a deleted gene identifier, multiple types of reference gene identifiers and multiple types of variant gene identifiers; if the gene unit is a newly added gene unit, the value of the newly added gene identifier in the initial type vector and the value of the variant gene identifier that matches the type of the gene unit are set to a first preset value, and the values ​​of other identifiers in the initial type vector are set to a second preset value to generate an abnormal type vector of the gene unit; if the gene unit is a deleted gene unit, the initial type vector is set to a second preset value. The numerical value of the deleted gene identification bit in the vector and the numerical value of the reference gene identification bit matching the type of the gene unit are set to the first preset value, and the numerical values ​​of other identification bits in the initial type vector are set to the second preset value to generate an abnormal type vector of the gene unit; if the gene unit is a replacement gene unit, the historical gene type of the gene unit is determined, the numerical value of the reference gene identification bit matching the historical gene type and the numerical value of the variant gene identification bit matching the gene unit are set to the first preset value, and the numerical values ​​of other identification bits in the initial type vector are set to the second preset value to generate an abnormal type vector of the gene unit.

20. The device according to claim 18, characterized in that The feature encoding unit is also used to construct an initial position vector of any gene unit; the initial position vector includes multiple candidate chromosome identification positions; among the multiple candidate chromosome identification positions, determine the target chromosome identification position corresponding to the chromosome position of the gene unit, and set the value of the target chromosome identification position to a first preset value, and set the values ​​of other chromosome identification positions in the initial position vector to a second preset value, so as to obtain the position vector of the gene unit.

21. The device according to claim 17, characterized in that The encoding network includes a text encoding layer and a numerical encoding layer; The feature encoding unit is further used to extract text features in the detection status information based on the text encoding layer, and to extract numerical features in the detection status information based on the numerical encoding layer; The detection status feature is determined according to the text feature and the numerical feature.

22. The device according to claim 17, characterized in that The decoding network includes a pooling layer and a classification layer; The feature decoding unit is used to perform dimensionality reduction processing on the multimodal biometric feature based on the pooling layer to obtain a one-dimensional abnormal sequence feature of the target object; Based on the classification layer, the one-dimensional abnormal sequence features are reduced in dimension, the matching probability between the target object and each candidate state information in the candidate state information set is predicted, and the candidate state information whose matching probability is in a preset probability interval is determined as the target predicted state information of the target object.

23. The device according to any one of claims 13 to 16, characterized in that: The state prediction device comprises: A data set acquisition module is used to acquire a state information data set; the state information data set includes historical abnormal sequence information, historical detection state information and historical state information labels of multiple objects; A model training module is used to predict the historical abnormal sequence information and the historical detection state information through a state prediction model to obtain historical predicted state information, and to train the state prediction model according to the loss value of the historical predicted state information and the historical state information label to obtain the state prediction model.

24. The device according to any one of claims 13 to 16, characterized in that The state prediction model is a gene disease prediction model; the state prediction device comprises: A genomic information acquisition module, used to obtain the gene reference sequence, gene variation sequence, chromosome position sequence and gene absolute position sequence of the target object; An imaging omics information acquisition module, used to acquire clinical information and imaging information of the target object; The disease type prediction module is used to input the gene reference sequence, the gene variation sequence, the chromosome position sequence, the gene absolute position sequence, the clinical information and the imaging information into the gene disease prediction model to extract the multimodal biological characteristics of the target object, and analyze the multimodal biological characteristics to predict the gene disease type of the target object.

25. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 12 are implemented.

26. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

27. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images