Data processing method and device for multi-task prediction
Through multi-task prediction data processing methods and devices, key features in multi-modal data are extracted and deep learning and machine learning are integrated to solve the problem of lack of prediction models for patients with TBI combined hemorrhagic shock and the low accuracy of multi-task prediction data evaluation. , improves the reliability and efficiency of prediction.
Patent Information
- Application Number
- CN202411052190.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-11-19
AI Technical Summary
Existing prediction models have not yet been developed for patients with TBI and hemorrhagic shock, and multitasking prediction has problems with the accuracy and efficiency of data evaluation in clinical practice, especially when dealing with multiple prognostic outcome indicators.
Provide a multi-task prediction data processing method and device, by obtaining multi-modal data information, performing pre-processing and analysis processing, extracting key features, integrating the advantages of deep learning and machine learning, optimizing model performance, and automatically adjusting model parameters to Adapt to different clinical scenarios.
It improves the accuracy of data evaluation and the reliability and efficiency of task prediction, and can adapt to different clinical scenarios without changing the basic model structure and meet the needs of the medical environment.
Smart Images

Figure CN118981698A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of prediction technology, and in particular to a data processing method and device for multi-task prediction. Background Art
[0002] With the rapid progress of artificial intelligence in the medical field, some prediction models for the prognosis of patients with single TBI and single hemorrhagic shock have been developed, but no prediction models for patients with TBI combined with hemorrhagic shock have been developed. In addition, clinicians may need to consider multiple prognostic indicators because different prognostic indicators reflect different clinical information. The simultaneous prediction of multiple prognostic results can theoretically provide doctors with more comprehensive decision support information, but most of the existing models remain at the prediction of a single prognostic result. At the technical level, deep learning and machine learning, as the two pillars of artificial intelligence, each have their own unique advantages and limitations. Deep learning performs well in processing high-dimensional complex data. It can automatically learn data features and build powerful prediction models, but it is limited by the need for large amounts of data and computing resources, and the model has poor interpretability. In contrast, machine learning has the characteristics of diverse algorithms, good interpretability, and high computational efficiency, but it may not be as good as deep learning when processing extremely complex data. Therefore, a data processing method and device for multi-task prediction are provided to identify and extract key features of multimodal data information, improve the accuracy of data evaluation, and further improve the reliability and efficiency of task prediction. Summary of the invention
[0003] The technical problem to be solved by the present invention is to provide a data processing method and device for multi-task prediction, which is conducive to the key feature identification and extraction of multimodal data information, improves the accuracy of data evaluation, and further improves the reliability and efficiency of task prediction.
[0004] In order to solve the above technical problems, a first aspect of an embodiment of the present invention discloses a data processing method for multi-task prediction, the method comprising:
[0005] Acquire multimodal data information; the multimodal data information includes a plurality of modal data information to be processed; each of the modal data information to be processed includes a plurality of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period consisting of a lower limit time value and an upper limit time value; M is an integer not less than 0;
[0006] Pre-processing the multimodal data information to obtain initial processing multimodal information; the initial processing multimodal information includes a plurality of initial processing modal information;
[0007] The initially processed multimodal information is analyzed and processed to obtain target task analysis result information.
[0008] A second aspect of an embodiment of the present invention discloses a data processing device for multi-task prediction, the device comprising:
[0009] An acquisition module, used for acquiring multimodal data information; the multimodal data information includes a plurality of modal data information to be processed; each of the modal data information to be processed includes a plurality of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period consisting of a lower limit time value and an upper limit time value; and M is an integer not less than 0;
[0010] A first processing module is used to pre-process the multimodal data information to obtain initial processing multimodal information; the initial processing multimodal information includes a plurality of initial processing modal information;
[0011] The second processing module is used to analyze and process the initially processed multimodal information to obtain target task analysis result information.
[0012] A third aspect of the present invention discloses another data processing device for multi-task prediction, the device comprising:
[0013] A memory storing executable program code;
[0014] a processor coupled to the memory;
[0015] The processor calls the executable program code stored in the memory to execute part or all of the steps in the data processing method for multi-task prediction disclosed in the first aspect of the embodiment of the present invention.
[0016] The fourth aspect of the present invention discloses a computer-readable storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the data processing method for multi-task prediction disclosed in the first aspect of an embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0018] Figure 1 is a schematic diagram of a scenario of a data processing system for multi-task prediction provided by an embodiment of the present invention;
[0019] Figure 2 It is a flowchart of a data processing method for multi-task prediction disclosed in an embodiment of the present invention;
[0020] Figure 3 is a structural schematic diagram of a data processing device for multi-task prediction disclosed in an embodiment of the present invention;
[0021] Figure 4 It is a structural schematic diagram of another data processing device for multi-task prediction disclosed in an embodiment of the present invention. DETAILED DESCRIPTION
[0022] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0023] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or equipment.
[0024] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0025] In this application, the word "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described in this application as "exemplary" is not necessarily to be construed as being preferred or advantageous over other embodiments. The following description is given to enable any technician in the field to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present application can be implemented without using these specific details. In other instances, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present application with unnecessary details. Therefore, the present application is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present application.
[0026] It should be noted that since the method of the embodiment of the present application is executed in a computer device, the processing objects of each computer device exist in the form of data or information. For example, time is actually time information. It can be understood that if size, quantity, position, etc. are mentioned in subsequent embodiments, they are all corresponding data for processing by the computer device. The details will not be repeated here.
[0027] The embodiments of the present application provide a data processing method, apparatus, computer device, and computer-readable storage medium for multi-task prediction, which are described in detail below.
[0028] See also Figure 1 , Figure 1 Schematic diagram of a scenario of a data processing system for multi-task prediction provided in an embodiment of the present application. The data processing system for multi-task prediction may include a computer device 100, in which a data processing device for multi-task prediction is integrated, such as Figure 1 Computer equipment in.
[0029] In the embodiment of the present application, the computer device 100 is mainly used to obtain multimodal data information; the multimodal data information includes a plurality of modal data information to be processed; each modal data information to be processed includes a plurality of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period consisting of a lower limit time value and an upper limit time value; M is an integer not less than 0;
[0030] Pre-processing the multimodal data information to obtain initial processing multimodal information; the initial processing multimodal information includes a plurality of initial processing modal information;
[0031] The initial processed multimodal information is analyzed and processed to obtain the target task analysis result information.
[0032] It can identify and extract key features of multimodal data information, improve the accuracy of data evaluation, and thus improve the reliability and efficiency of task prediction.
[0033] In the embodiment of the present application, the computer device 100 may be an independent server, or a server network or server cluster composed of servers. For example, the computer device 100 described in the embodiment of the present application includes but is not limited to a computer, a network host, a single network server, a plurality of network server sets or a cloud server composed of a plurality of servers. The cloud server is composed of a large number of computers or network servers based on cloud computing.
[0034] It is understandable that the computer device 100 used in the embodiments of the present application may be a device including both receiving and transmitting hardware, that is, a device having receiving and transmitting hardware capable of performing two-way communication on a two-way communication link. Such a device may include: a cellular or other communication device having a single-line display or a multi-line display or a cellular or other communication device without a multi-line display. The specific computer device 100 may be a desktop terminal or a mobile terminal, and the computer device 100 may also be one of a mobile phone, a tablet computer, a laptop computer, etc.
[0035] Those skilled in the art will understand that Figure 1 The application environment shown in the figure is only one application scenario of the present application solution and does not constitute a limitation on the application scenario of the present application solution. Other application environments may also include Figure 1 More or less computer equipment as shown in Figure 1 Only one computer device is shown. It can be understood that the data processing system for multi-task prediction may also include one or more other services, which are not specifically limited here.
[0036] In addition, if Figure 1 As shown, the data processing system for multi-task prediction may also include a memory 200 for storing data, such as image data, location information, etc.
[0037] It should be noted that Figure 1 The scenario diagram of the data processing system for multi-task prediction shown is merely an example. The data processing system and scenario for multi-task prediction described in the embodiment of the present application are intended to more clearly illustrate the technical solution of the embodiment of the present application, and do not constitute a limitation on the technical solution provided in the embodiment of the present application. A person of ordinary skill in the art can appreciate that with the evolution of the data processing system for multi-task prediction and the emergence of new business scenarios, the technical solution provided in the embodiment of the present application is equally applicable to similar technical problems.
[0038] The present invention discloses a data processing method and device for multi-task prediction, which is beneficial to identify and extract key features of multi-modal data information, improve the accuracy of data evaluation, and further improve the reliability and efficiency of task prediction. The following are detailed descriptions.
[0039] Embodiment 1
[0040] See also Figure 2 , Figure 2 is a flow chart of a data processing method for multi-task prediction disclosed in an embodiment of the present invention. Figure 2 The data processing method for multi-task prediction described is applied to a management system, such as a local server or a cloud server for management, and the embodiments of the present invention do not limit this. Figure 2 As shown, the data processing method for multi-task prediction may include the following operations:
[0041] 101. Obtain multimodal data information.
[0042] In an embodiment of the present invention, the multimodal data information includes a number of modal data information to be processed; each modal data information to be processed includes a number of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period composed of a lower limit time value and an upper limit time value; M is an integer not less than 0.
[0043] 102. Pre-process the multimodal data information to obtain initial processed multimodal information.
[0044] In the embodiment of the present invention, the initial processing multimodal information includes a plurality of initial processing modal information.
[0045] 103. Analyze and process the initially processed multimodal information to obtain target task analysis result information.
[0046] It should be noted that the above-mentioned target task analysis result information is a data prediction result value, which is used to evaluate the user status and is not limited in the embodiment of the present invention.
[0047] It should be noted that the above-mentioned modal data to be processed may be data representing a physiological state, audio data, or image data, which is not limited in the embodiment of the present invention.
[0048] It should be noted that the time interval in the above-mentioned time modal data information is a time period defined in hours, such as 8:00-9:00, which is not limited in the embodiment of the present invention. Further, the time modal value is the user modal data collected within the time interval, which can be multiple or empty, which is not limited in the embodiment of the present invention.
[0049] It should be noted that the data processing method for multi-task prediction can be deployed in a data processing system for multi-task prediction. Further, the data processing system for multi-task prediction may include a multimodal data perception subsystem and an intelligent multi-task dynamic prediction subsystem. Further, the multimodal data perception subsystem is used to pre-process the multimodal data information to obtain the initial processing multimodal information. Further, the intelligent multi-task dynamic prediction subsystem is used to analyze and process the initial processing multimodal information to obtain the target task analysis result information. First, the user's physiological parameters and experimental parameters are monitored in real time through the multimodal data perception subsystem. The newly monitored data will be transmitted to the intelligent feature extraction subsystem in real time through the multimodal data perception subsystem. The intelligent feature extraction subsystem analyzes the patient's physiological and experimental data during the learning window in real time, and identifies key features from them and passes them to the intelligent multi-task dynamic prediction subsystem. The system combines the advantages of deep learning and machine learning to optimize model performance. Deep learning will be used as an automatic feature extraction tool to identify key features from clinical time series data of users with TBI combined with hemorrhagic shock. These features will then be used as input data for multiple machine learning models to learn, and then the final prediction results will be obtained by integrating multiple machine learning prediction results. The system can automatically adjust the model parameters to adapt to different clinical scenarios without changing its basic model structure, thereby ensuring the high accuracy and strong robustness of the model to meet the application needs in actual medical environments.
[0050] It should be noted that the above-mentioned pre-processing of multimodal data information to obtain the initial processed multimodal information utilizes deep learning technology specifically for processing time series to analyze the patient data within the learning window in the multimodal data perception subsystem, and identify key features closely related to the patient's prognosis from these data, and finally transmit these features to the machine learning model in the intelligent multi-task prediction subsystem for further learning.
[0051] It should be noted that the above-mentioned modal data information to be processed represents the time modal values collected in different time intervals, such as sound, facial image, heart rate, diastolic blood pressure, systolic blood pressure, mean arterial pressure, arterial oxygen saturation, arterial oxygen partial pressure, arterial carbon dioxide partial pressure, pH value, body temperature, respiratory rate, pulse, lactate, glucose, base excess, hemoglobin, international normalized ratio, partial thromboplastin time, platelet count, complete blood cell count, white blood cells and other data representing user modal information, which is not limited in the embodiments of the present invention.
[0052] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0053] In an optional embodiment, the pre-processing of the multimodal data information to obtain the initial processed multimodal information includes:
[0054] For any modal data information to be processed in the multimodal data information, determine whether there is time modal data information with M greater than 1 in the time modal data information corresponding to the modal data information to be processed, and obtain a first data determination result;
[0055] When the first data judgment result is yes, for any time modal data information where M is greater than 1, other time modal values except the time modal value closest to the upper limit time value corresponding to the time modal data information are removed from the time modal data information;
[0056] When the first data judgment result is no, ending the judgment process corresponding to the first data judgment result;
[0057] Determine whether there is time modal data information with M equal to 0 in the time modal data information corresponding to the modal data information to be processed, and obtain a second data determination result;
[0058] When the second data judgment result is yes, for any time modal data information where M is equal to 0, the time modal value corresponding to the time modal data information is estimated using the time modal data information where M is not equal to 0 in the modal data information to be processed, to obtain updated time modal data information;
[0059] When the second data judgment result is no, the judgment process corresponding to the second data judgment result is terminated;
[0060] The modal data information to be processed is determined as an initial processing modal information; any time interval in the initial processing modal information corresponds to a unique time modal value.
[0061] It should be noted that, the existence of time modal data information with M greater than 1 in the time modal data information corresponding to the above-mentioned modal data information to be processed indicates that multiple time modal values are collected in the same time interval.
[0062] It should be noted that the time modal data information in which M is equal to 0 indicates that no time modal value is collected within the time interval, or the time modal value is abnormal, which is not limited in the embodiment of the present invention.
[0063] It should be noted that the above-mentioned removal of other time modal values from the time modal data information except the time modal value closest to the upper limit time value corresponding to the time modal data information is to retain only one time modal value in the same time interval, and the time modal value is the upper limit time value closest to the time interval. For example, when in the time interval of 8:00-9:00, a time modal value of 1 is collected at 8:05, a time modal value of 3 is collected at 8:15, and a time modal value of 6 is collected at 8:55, then only the time modal value 6 collected at 8:55 is retained, and 1 and 3 are removed, which is not limited in the embodiment of the present invention.
[0064] In this optional embodiment, as an optional implementation manner, the above-mentioned using the time modal data information in which M is not equal to 0 in the modal data information to be processed to estimate the time modal value corresponding to the time modal data information to obtain the updated time modal data information includes:
[0065] The time modal data information with the highest time order and M not equal to 0 in the modal data information to be processed is used as the first target time modal data information;
[0066] The time modal data information whose time order in the modal data information to be processed is after the time modal data information and closest to the time modal data information and whose M is not equal to 0 is used as the second target time modal data information;
[0067] Using the estimation model to estimate the first target time modal data information and the second target time modal data information to obtain an estimated time modal value;
[0068]
[0069] Wherein, GJMTZ is the estimated time modal value; DYMT is the time modal value corresponding to the first target time modal data information; DEMT is the time modal value corresponding to the second target time modal data information; DYSJ is the upper limit time value corresponding to the first target time modal data information; DESJ is the upper limit time value corresponding to the second target time modal data information; MTSJ is the upper limit time value corresponding to the time modal data information;
[0070] The estimated time modal value is determined as the time modal value corresponding to the time modal data information, and updated time modal data information is obtained.
[0071] For example, when the modal data information to be processed is the following table data:
[0072] Upper limit time value 1 2 3 4 5 6 Time modal value 3 10 11 Missing Values 15 16
[0073] Among them, the time modal data information with the most forward time order M not equal to 0 in the processed modal data information is the time modal data information composed of the upper limit time value of 1 and the time modal value of 3, then the upper limit time value and time modal value corresponding to the first target time modal data information are 2 and 10 respectively; correspondingly, the upper limit time value and time modal value corresponding to the second target time modal data information are 5 and 15 respectively, then the estimated time modal value is
[0074] It should be noted that the updated time modal data information is to fill in and update the missing time modal values in the time modal data information where M is equal to 0, which is not limited in the embodiment of the present invention.
[0075] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0076] In another optional embodiment, the initially processed multimodal information is analyzed and processed to obtain target task analysis result information, including:
[0077] Based on the initial processing multimodal information, target modal feature information is determined; the target modal feature information includes a plurality of target modal feature value information arranged in sequence; each target modal feature value information includes a modal sequence value and a modal feature value;
[0078] Based on the target modal feature information, the target task analysis result information is determined.
[0079] In this optional embodiment, as an optional implementation, the above-mentioned determination of target modal feature information based on the initial processing of multimodal information includes:
[0080] Using the first initial processing layer to perform data filling processing on the initially processed multimodal information to obtain filled modal information;
[0081] The second initial processing layer is used to perform feature extraction processing on the filled modal information to obtain the target modal feature information.
[0082] It should be noted that the first initial processing layer is a Masking layer. When processing the initial processing modal information that represents time series and natural language processing tasks, since the lengths of sentences corresponding to different initial processing modal information may be inconsistent, 0 padding values will be used in the first initial processing layer to fill in the sentences so that the lengths of all sentences are uniform. The role of the first initial processing layer is to make it clear to the model which parts are filled so that these parts can be ignored in subsequent processing. After being processed by the first initial processing layer, the data will be sent to the second initial processing layer built based on a bidirectional long short-term memory network. In the second initial processing layer, two independent LSTM networks will be trained at the same time, one responsible for processing the forward information of the sequence and the other for processing the reverse information. The second initial processing layer can learn key features from past and future information at the same time, thereby more effectively capturing the dependency features in the time series.
[0083] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0084] In yet another optional embodiment, based on the target modal feature information, target task analysis result information is determined, including:
[0085] Utilizing the first modal feature calculation model to calculate and process the target modal feature information, to obtain first task analysis result information; the first task analysis result information includes a plurality of first task analysis result values arranged in sequence;
[0086] Analyzing and processing the target modal feature information to obtain second task analysis result information; the second task analysis result information includes a plurality of second task analysis result values arranged in sequence;
[0087] The target modal feature information is calculated and processed using the second modal feature calculation model to obtain third task analysis result information; the third task analysis result information includes a plurality of third task analysis result values arranged in sequence;
[0088] The first task analysis result information, the second task analysis result information and the third task analysis result information are summed and averaged to obtain the target task analysis result information.
[0089] It should be noted that the above-mentioned summing and averaging of the first task analysis result information, the second task analysis result information and the third task analysis result information represents that a majority voting integrator is used to determine the final prediction result, which is selected by counting the categories that appear most frequently in the prediction results of each algorithm.
[0090] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0091] In yet another optional embodiment, the target modal feature information is calculated and processed using the first modal feature calculation model to obtain first task analysis result information, including:
[0092] Acquire first modal coefficient information; the first modal coefficient information includes a plurality of first modal coefficient values and a first modal constant value;
[0093] Utilizing the first modal feature calculation model to calculate and process the target modal feature information and the first modal coefficient information to obtain first task analysis result information;
[0094] Among them, the first modal feature calculation model is:
[0095]
[0096] In the formula, DY i is the i-th first task analysis result value in the first task analysis result information; x j is the jth first modal coefficient value; y1 is the first modal constant value; MT j is the modal eigenvalue corresponding to the jth target modal eigenvalue information; MS j is a vector composed in sequence of the modal sequence values corresponding to the first j target modal eigenvalue information in the target modal feature information.
[0097] It should be noted that the above first modal coefficient values are arranged in sequence, which is not limited in the embodiment of the present invention.
[0098] It should be noted that the first modal coefficient information may be a value set by a user according to needs, or may be determined by a system according to historical data, and the embodiment of the present invention does not limit this.
[0099] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0100] In an optional embodiment, the target modal feature information is analyzed and processed to obtain the second task analysis result information, including:
[0101] Obtain a decision estimation module set; the decision estimation module set includes a plurality of decision estimation modules;
[0102] For any target modal eigenvalue information in the target modal eigenvalue information, the modal eigenvalues corresponding to the target modal eigenvalue information are estimated in turn by using the decision estimation modules in the decision estimation module set to obtain a plurality of task estimation result values; each decision estimation module corresponds to a unique task estimation result value;
[0103] The estimated result values of all tasks are summed and averaged to obtain the second task analysis result value corresponding to the target modal eigenvalue information.
[0104] It should be noted that the above-mentioned decision estimation module can be a decision tree set by the user, or a classification prediction model constructed by a neural network such as CNN, which is not limited in the embodiment of the present invention. Furthermore, each decision estimation module is inconsistent.
[0105] It should be noted that the above analysis and processing of the target modal feature information improves the versatility and robustness of a single decision estimation module, and can reduce the overfitting of a single decision estimation module, and has better prediction performance than a single decision estimation module.
[0106] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0107] In another optional embodiment, the target modal feature information is calculated and processed using the second modal feature calculation model to obtain third task analysis result information, including:
[0108] Performing fitting processing on target modal characteristic value information in the target modal characteristic information to obtain fitting parameter information; the fitting parameter information includes a first fitting parameter value and a second fitting parameter value;
[0109] The target modal feature information and the fitting parameter information are calculated and processed by using the second modal feature calculation model to obtain the third task analysis result information;
[0110] Among them, the second modal feature calculation model is:
[0111]
[0112] Where, DS l MT is the lth third task analysis result value in the third task analysis result information; l is the modal eigenvalue corresponding to the lth target modal eigenvalue information; z1 and z2 are the first fitting parameter value and the second fitting parameter value respectively.
[0113] It should be noted that the above-mentioned fitting process of the target modal characteristic value information in the target modal characteristic information is to perform linear fitting on the data pair consisting of the modal order value and the modal characteristic value, thereby obtaining two linear parameters, namely the first fitting parameter value and the second fitting parameter value. For example, when the target modal characteristic value information is the data in the following table:
[0114] Modal Order Value 1 2 3 4 5 6 Modal eigenvalues 3 5 7 9 11 13
[0115] Then, taking (0,0) as the coordinate origin, the modal order value as the horizontal coordinate value, and the modal characteristic value as the vertical coordinate value, a linear fitting is performed, and the fitted linear relationship can be obtained as Y=2*X+1. The corresponding first fitting parameter value and the second fitting parameter value are 2 and 1 respectively, which is not limited in the embodiment of the present invention.
[0116] It should be noted that the computational cost of calculating and processing the target modal feature information using the second modal feature calculation model is not high, and different features of the task can be efficiently and quickly predicted and analyzed with good robustness, which is not limited in the embodiments of the present invention.
[0117] It can be seen that implementing the data processing method for multi-task prediction described in the embodiment of the present invention is conducive to identifying and extracting key features of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0118] Embodiment 2
[0119] See also Figure 3 , Figure 3 is a schematic diagram of the structure of a data processing device for multi-task prediction disclosed in an embodiment of the present invention. Figure 3 The described device can be applied to a management system, such as a local server or a cloud server for management, etc., which is not limited in the embodiments of the present invention. Figure 3 As shown, the device may include:
[0120] The acquisition module 201 is used to acquire multimodal data information; the multimodal data information includes a plurality of modal data information to be processed; each modal data information to be processed includes a plurality of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period consisting of a lower limit time value and an upper limit time value; M is an integer not less than 0;
[0121] The first processing module 202 is used to pre-process the multimodal data information to obtain initial processing multimodal information; the initial processing multimodal information includes a plurality of initial processing modal information;
[0122] The second processing module 203 is used to analyze and process the initially processed multimodal information to obtain target task analysis result information.
[0123] It can be seen that implementation Figure 3 The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0124] In another optional embodiment, Figure 3 As shown, the first processing module 202 performs pre-processing on the multimodal data information to obtain initial processed multimodal information, including:
[0125] For any modal data information to be processed in the multimodal data information, determine whether there is time modal data information with M greater than 1 in the time modal data information corresponding to the modal data information to be processed, and obtain a first data determination result;
[0126] When the first data judgment result is yes, for any time modal data information where M is greater than 1, other time modal values except the time modal value closest to the upper limit time value corresponding to the time modal data information are removed from the time modal data information;
[0127] When the first data judgment result is no, ending the judgment process corresponding to the first data judgment result;
[0128] Determine whether there is time modal data information with M equal to 0 in the time modal data information corresponding to the modal data information to be processed, and obtain a second data determination result;
[0129] When the second data judgment result is yes, for any time modal data information where M is equal to 0, the time modal value corresponding to the time modal data information is estimated using the time modal data information where M is not equal to 0 in the modal data information to be processed, to obtain updated time modal data information;
[0130] When the second data judgment result is no, the judgment process corresponding to the second data judgment result is terminated;
[0131] The modal data information to be processed is determined as an initial processing modal information; any time interval in the initial processing modal information corresponds to a unique time modal value.
[0132] It can be seen that implementation Figure 3 The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0133] In yet another optional embodiment, Figure 3 As shown, the second processing module 203 analyzes and processes the initial processing multimodal information to obtain target task analysis result information, including:
[0134] Based on the initial processing multimodal information, target modal feature information is determined; the target modal feature information includes a plurality of target modal feature value information arranged in sequence; each target modal feature value information includes a modal sequence value and a modal feature value;
[0135] Based on the target modal feature information, the target task analysis result information is determined.
[0136] It can be seen that the implementation Figure 3 The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0137] In yet another optional embodiment, Figure 3 As shown, the second processing module 203 determines the target task analysis result information based on the target modal feature information, including:
[0138] Utilizing the first modal feature calculation model to calculate and process the target modal feature information, to obtain first task analysis result information; the first task analysis result information includes a plurality of first task analysis result values arranged in sequence;
[0139] Analyzing and processing the target modal feature information to obtain second task analysis result information; the second task analysis result information includes a plurality of second task analysis result values arranged in sequence;
[0140] The target modal feature information is calculated and processed using the second modal feature calculation model to obtain third task analysis result information; the third task analysis result information includes a plurality of third task analysis result values arranged in sequence;
[0141] The first task analysis result information, the second task analysis result information and the third task analysis result information are summed and averaged to obtain the target task analysis result information.
[0142] It can be seen that the implementation Figure 3 The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0143] In yet another optional embodiment, Figure 3As shown, the second processing module 203 uses the first modal feature calculation model to calculate and process the target modal feature information to obtain the first task analysis result information, including:
[0144] Acquire first modal coefficient information; the first modal coefficient information includes a plurality of first modal coefficient values and a first modal constant value;
[0145] Utilizing the first modal feature calculation model to calculate and process the target modal feature information and the first modal coefficient information to obtain first task analysis result information;
[0146] Among them, the first modal feature calculation model is:
[0147]
[0148] In the formula, DY i is the i-th first task analysis result value in the first task analysis result information; x j is the jth first modal coefficient value; y1 is the first modal constant value; MT j is the modal eigenvalue corresponding to the jth target modal eigenvalue information; MS j is a vector composed in sequence of the modal sequence values corresponding to the first j target modal eigenvalue information in the target modal feature information.
[0149] It can be seen that implementation Figure 3 The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0150] In yet another optional embodiment, Figure 3 As shown, the second processing module 203 analyzes and processes the target modal feature information to obtain second task analysis result information, including:
[0151] Obtain a decision estimation module set; the decision estimation module set includes a plurality of decision estimation modules;
[0152] For any target modal eigenvalue information in the target modal eigenvalue information, the modal eigenvalues corresponding to the target modal eigenvalue information are estimated in turn by using the decision estimation modules in the decision estimation module set to obtain a plurality of task estimation result values; each decision estimation module corresponds to a unique task estimation result value;
[0153] The estimated result values of all tasks are summed and averaged to obtain the second task analysis result value corresponding to the target modal eigenvalue information.
[0154] It can be seen that implementation Figure 3The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0155] In yet another optional embodiment, Figure 3 As shown, the second processing module 203 uses the second modal feature calculation model to calculate and process the target modal feature information to obtain the third task analysis result information, including:
[0156] Performing fitting processing on target modal characteristic value information in the target modal characteristic information to obtain fitting parameter information; the fitting parameter information includes a first fitting parameter value and a second fitting parameter value;
[0157] The target modal feature information and the fitting parameter information are calculated and processed by using the second modal feature calculation model to obtain the third task analysis result information;
[0158] Among them, the second modal feature calculation model is:
[0159]
[0160] Where, DS l MT is the lth third task analysis result value in the third task analysis result information; l is the modal eigenvalue corresponding to the lth target modal eigenvalue information; z1 and z2 are the first fitting parameter value and the second fitting parameter value respectively.
[0161] It can be seen that the implementation Figure 3 The described data processing device for multi-task prediction is conducive to the key feature identification and extraction of multimodal data information, improving the accuracy of data evaluation, and further improving the reliability and efficiency of task prediction.
[0162] Embodiment 3
[0163] See also Figure 4 , Figure 4 is a schematic diagram of the structure of another data processing device for multi-task prediction disclosed in an embodiment of the present invention. Figure 4 The described device can be applied to a management system, such as a local server or a cloud server for management, etc., which is not limited in the embodiments of the present invention. Figure 4 As shown, the device may include:
[0164] A memory 301 storing executable program codes;
[0165] a processor 302 coupled to the memory 301;
[0166] The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the data processing method for multi-task prediction described in the first embodiment.
[0167] Embodiment 4
[0168] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the data processing method for multi-task prediction described in the first embodiment.
[0169] Embodiment 5
[0170] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps in the data processing method for multi-task prediction described in the first embodiment.
[0171] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.
[0172] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.
[0173] Finally, it should be noted that the data processing method and device for multi-task prediction disclosed in the embodiments of the present invention disclose only the preferred embodiments of the present invention, which are only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A data processing method for multi-task prediction, characterized in that: The method comprises: Acquire multimodal data information; the multimodal data information includes a plurality of modal data information to be processed; each of the modal data information to be processed includes a plurality of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period consisting of a lower limit time value and an upper limit time value; M is an integer not less than 0; Pre-processing the multimodal data information to obtain initial processing multimodal information; the initial processing multimodal information includes a plurality of initial processing modal information; The initially processed multimodal information is analyzed and processed to obtain target task analysis result information.
2. The data processing method for multi-task prediction according to claim 1, characterized in that: The pre-processing of the multimodal data information to obtain initial processed multimodal information includes: For any of the to-be-processed modal data information in the multimodal data information, determine whether the time modal data information corresponding to the to-be-processed modal data information includes the time modal data information where M is greater than 1, and obtain a first data determination result; When the first data judgment result is yes, for any of the time modal data information where M is greater than 1, the other time modal values except the time modal value closest to the upper limit time value corresponding to the time modal data information are removed from the time modal data information; When the first data judgment result is no, ending the judgment process corresponding to the first data judgment result; Determine whether the time modal data information corresponding to the modal data information to be processed includes the time modal data information in which M is equal to 0, and obtain a second data determination result; When the second data judgment result is yes, for any of the time modal data information where M is equal to 0, the time modal value corresponding to the time modal data information is estimated using the time modal data information where M is not equal to 0 in the modal data information to be processed, to obtain updated time modal data information; When the second data judgment result is no, ending the judgment process corresponding to the second data judgment result; The modal data information to be processed is determined as the initial processing modal information; any one of the time intervals in the initial processing modal information corresponds to a unique time modal value.
3. The data processing method for multi-task prediction according to claim 1, characterized in that: The analyzing and processing the initially processed multimodal information to obtain target task analysis result information includes: Based on the initially processed multimodal information, target modal feature information is determined; the target modal feature information includes a plurality of target modal feature value information arranged in sequence; each of the target modal feature value information includes a modal sequence value and a modal feature value; Based on the target modal feature information, target task analysis result information is determined.
4. The data processing method for multi-task prediction according to claim 2, characterized in that: Determining target task analysis result information based on the target modal feature information includes: The target modal feature information is calculated and processed using a first modal feature calculation model to obtain first task analysis result information; the first task analysis result information includes a plurality of first task analysis result values arranged in sequence; Analyzing and processing the target modal feature information to obtain second task analysis result information; the second task analysis result information includes a plurality of second task analysis result values arranged in sequence; The target modal feature information is calculated and processed using the second modal feature calculation model to obtain third task analysis result information; the third task analysis result information includes a plurality of third task analysis result values arranged in sequence; The first task analysis result information, the second task analysis result information and the third task analysis result information are summed and averaged to obtain target task analysis result information.
5. The data processing method for multi-task prediction according to claim 4, characterized in that: The step of calculating and processing the target modal feature information using the first modal feature calculation model to obtain first task analysis result information includes: Acquire first modal coefficient information; the first modal coefficient information includes a plurality of first modal coefficient values and a first modal constant value; Using a first modal feature calculation model to calculate and process the target modal feature information and the first modal coefficient information to obtain first task analysis result information; Wherein, the first modal feature calculation model is: Where DY i is the i-th first task analysis result value in the first task analysis result information; x j is the jth value of the first modal coefficient; y1 is the value of the first modal constant; MT j is the modal eigenvalue corresponding to the j-th target modal eigenvalue information; MS j It is a vector composed in sequence of the modal sequence values corresponding to the first j target modal feature value information in the target modal feature information.
6. The data processing method for multi-task prediction according to claim 4, characterized in that: The analyzing and processing the target modal feature information to obtain the second task analysis result information includes: Acquire a decision estimation module set; the decision estimation module set includes a plurality of decision estimation modules; For any of the target modal feature value information in the target modal feature information, the modal feature values corresponding to the target modal feature value information are estimated in turn using the decision estimation modules in the decision estimation module set to obtain a plurality of task estimation result values; each of the decision estimation modules corresponds to a unique task estimation result value; All the task estimation result values are summed and averaged to obtain the second task analysis result value corresponding to the target modal eigenvalue information.
7. The data processing method for multi-task prediction according to claim 4, characterized in that: The method of calculating and processing the target modal feature information by using the second modal feature calculation model to obtain third task analysis result information includes: Performing fitting processing on the target modal feature value information in the target modal feature information to obtain fitting parameter information; the fitting parameter information includes a first fitting parameter value and a second fitting parameter value; Using the second modal feature calculation model to calculate and process the target modal feature information and the fitting parameter information to obtain third task analysis result information; Wherein, the second modal feature calculation model is: Where, DS l is the lth third task analysis result value in the third task analysis result information; MT l is the modal eigenvalue corresponding to the lth target modal eigenvalue information; z1 and z2 are the first fitting parameter value and the second fitting parameter value respectively.
8. A data processing device for multi-task prediction, characterized in that: The device comprises: An acquisition module, used for acquiring multimodal data information; the multimodal data information includes a plurality of modal data information to be processed; each of the modal data information to be processed includes a plurality of time modal data information arranged in chronological order; the time modal data information includes a time interval and M time modal values; the time interval represents a time period consisting of a lower limit time value and an upper limit time value; and M is an integer not less than 0; A first processing module is used to pre-process the multimodal data information to obtain initial processing multimodal information; the initial processing multimodal information includes a plurality of initial processing modal information; The second processing module is used to analyze and process the initially processed multimodal information to obtain target task analysis result information.
9. A data processing device for multi-task prediction, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the data processing method for multi-task prediction according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the data processing method for multi-task prediction according to any one of claims 1 to 7.
