A method and system for medical data visualization

Abstract

This invention provides a method and system for visualizing medical data. The method includes: Step S1: acquiring medical data and simultaneously acquiring the user's initial viewing intent; Step S2: determining a first analysis strategy based on the initial viewing intent; Step S3: analyzing the medical data based on the first analysis strategy to obtain a first analysis result; Step S4: inputting the first analysis result into a preset visualization model for visualization display. This invention's method and system for visualizing medical data directly acquires the user's initial viewing intent, determines the first analysis strategy based on the initial viewing intent, and performs corresponding analysis on the medical data, eliminating the need for user analysis and improving convenience. Furthermore, directly inputting the analysis result into the visualization model for visualization display enhances intuitiveness.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and system for implementing medical data visualization. Background Technology

[0002] Currently, when researchers analyze medical data, they need to do the analysis themselves, which is quite tedious. At the same time, they also need to organize the results themselves before displaying them, which may lead to problems such as the presentation not being intuitive enough.

[0003] Therefore, a solution is urgently needed. Summary of the Invention

[0004] One of the objectives of this invention is to provide a method and system for visualizing medical data. This method directly obtains the user's initial viewing intent, determines a first analysis strategy based on the initial viewing intent, and performs corresponding analysis on the medical data. This eliminates the need for user analysis, thus improving convenience. At the same time, the analysis results are directly input into the visualization model for visualization display, thereby improving intuitiveness.

[0005] This invention provides a method for implementing medical data visualization, comprising:

[0006] Step S1: Obtain medical data and, at the same time, obtain the user's initial viewing intent;

[0007] Step S2: Based on the first viewing intent, determine the first analysis strategy;

[0008] Step S3: Based on the first analysis strategy, analyze the medical data to obtain the first analysis result;

[0009] Step S4: Input the first analysis result into the preset visualization model for visualization display.

[0010] Preferably, in step S1, acquiring medical data includes:

[0011] Obtain a preset collection set, the collection set including: multiple first collections;

[0012] Obtain the first credit value corresponding to the first collector; if the first credit value is less than or equal to a preset first threshold, remove the corresponding first collector.

[0013] Otherwise, obtain the composition type corresponding to the first collector, whereby the composition type includes: individuals and teams;

[0014] When the composition type corresponding to the first collector is a team, obtain the team information corresponding to the first collector, and the team information includes: multiple first collectors;

[0015] Attempt to obtain the guarantee circle of the first collector. If successful, the corresponding first collector will be designated as the second collector, and the remaining first collectors will be designated as the third collectors.

[0016] Obtain the second credit value of the second collector; if the second credit value is less than or equal to a preset second threshold and / or the third collector is not within the guarantee circle, remove the corresponding first collector.

[0017] Once all the first collectors that need to be removed from the first collectors have been removed, the remaining first collectors will be used as the second collectors.

[0018] Acquire at least one new first data item newly collected by the second collector, and simultaneously acquire the collection process of the second collector collecting the first data item;

[0019] The collection process is analyzed and broken down to obtain multiple first processes;

[0020] Obtain a preset importance analysis model, input the first process into the importance analysis model, and obtain the importance value;

[0021] If the importance value is greater than or equal to a preset third threshold, the first process will be used as the second process.

[0022] The collection method corresponding to the second process is obtained, and the collection method includes: online collection and physical collection;

[0023] When the collection method corresponding to the second process is network collection, obtain the collection scenario corresponding to the second process;

[0024] Obtain a preset collection simulation model, and use the collection simulation model to simulate data collection in the collection scenario to obtain simulation records;

[0025] Obtain a preset risk assessment model, input the simulation record into the risk assessment model, and obtain the risk value;

[0026] If the risk value is greater than or equal to the preset fourth threshold, the corresponding first data item is removed;

[0027] When the collection method corresponding to the second process is actual collection, obtain the collection object corresponding to the second process and the first authentication identifier corresponding to the collection object;

[0028] Based on a preset collection object-authentication identifier library, determine the second authentication identifier corresponding to the collection object;

[0029] The first authentication identifier and the second authentication identifier are compared and analyzed to obtain a second analysis result, which includes: consistency and inconsistency.

[0030] When the second analysis result is inconsistent, the corresponding first data item is removed;

[0031] Once all the first data items that need to be removed from the first data items have been removed, the remaining first data items will be used as the second data items.

[0032] Integrate the second data items to obtain medical data and complete the acquisition.

[0033] Preferably, in step S1, obtaining the user's first viewing intent includes:

[0034] Obtain multiple second viewing intents generated by the user in history, and simultaneously obtain the generation time corresponding to the second viewing intent;

[0035] Based on the generation time, the second viewing intent is sorted chronologically to obtain an intent sequence;

[0036] The last second viewing intent in the intent sequence is selected as the third viewing intent. Meanwhile, in the intent sequence, the second viewing intents are traversed sequentially from the third viewing intent backward.

[0037] A preset intent association analysis model is obtained. The second viewing intent and the third viewing intent that have been traversed are input into the intent association analysis model. The intent association analysis model performs intent association analysis and obtains a third analysis result. The third analysis result includes: association and non-association.

[0038] When the third analysis result is not related, stop the traversal and take the second viewing intent encountered during the traversal as the fourth viewing intent;

[0039] The preset intent prediction model takes the third viewing intent and the fourth viewing intent as inputs and attempts to predict the intent.

[0040] If the intent prediction is successful, obtain the first intent prediction result, and use the first intent prediction result as the user's first viewing intent to complete the acquisition.

[0041] Otherwise, obtain the reason for the failure to obtain the intent prediction, which includes: multiple first requirement items;

[0042] Based on a preset requirement-behavior capture strategy library, at least one behavior capture strategy corresponding to the first requirement is determined;

[0043] Based on the behavior capture strategy, attempt to capture at least one first behavior item of the user corresponding to the first demand item; if the capture is successful, obtain the capture process of capturing the first behavior item.

[0044] The capture process is analyzed and broken down to obtain multiple third processes, and the sequence of the third processes is obtained.

[0045] Based on the chronological order of the processes, the third process is sorted to obtain a process sequence;

[0046] The third process that captures the first behavior item in the third process is determined and used as the fourth process;

[0047] Obtain the first capture scene corresponding to the fourth process, and at the same time, obtain the first confidence level corresponding to the first capture scene. If the first confidence level is less than or equal to a preset fifth threshold, remove the corresponding first behavior item.

[0048] Otherwise, a predetermined number of the third processes before and / or after the fourth process in the process sequence are selected and used as the fifth process;

[0049] Obtain the second capture scene corresponding to the fifth process, and at the same time, obtain the second confidence level corresponding to the second capture scene. If the second confidence level is less than or equal to the preset sixth threshold, remove the corresponding first behavior item.

[0050] After all the first behavior items that need to be removed from the first behavior items have been removed, determine whether there are any remaining first behavior items. If so, use the remaining first behavior items as second behavior items, and at the same time, use the corresponding first requirement item as the second requirement item.

[0051] The second action item is then input into the intent prediction model.

[0052] Once all the first behavioral items that need to be supplemented into the intent prediction model have been input, a third requirement item other than the second requirement item is determined from the first requirement items.

[0053] If the number of the third requirement items is 0, obtain the second prediction result obtained by the intent prediction model to re-predict the intent, and use the second prediction result as the user's first viewing intent to complete the acquisition;

[0054] Otherwise, the user is prompted to input the requirement information corresponding to the third requirement item. When the user finishes inputting, the requirement information is added to the intent prediction model, and the third prediction result obtained by the intent prediction model is obtained by re-predicting the intent. The third prediction result is used as the user's first viewing intent, and the acquisition is completed.

[0055] Preferably, step S2: determining a first analysis strategy based on the first viewing intent, including:

[0056] Build a library of viewing intent analysis strategies;

[0057] Based on the viewing intent-analysis strategy library, the first analysis strategy corresponding to the first viewing intent is determined.

[0058] Preferred methods for implementing medical data visualization also include:

[0059] Step S5: Obtain the user's operation on the visualization model, and respond to the operation.

[0060] This invention provides a medical data visualization implementation system, comprising:

[0061] The acquisition module is used to acquire medical data, and at the same time, to acquire the user's initial viewing intent;

[0062] The determination module is used to determine a first analysis strategy based on the first viewing intent;

[0063] An analysis module is used to analyze the medical data based on the first analysis strategy and obtain a first analysis result;

[0064] The display module is used to input the first analysis results into a preset visualization model for visualization display.

[0065] Preferably, the acquisition module performs the following operations:

[0066] Obtain a preset collection set, the collection set including: multiple first collections;

[0067] Obtain the first credit value corresponding to the first collector; if the first credit value is less than or equal to a preset first threshold, remove the corresponding first collector.

[0068] Otherwise, obtain the composition type corresponding to the first collector, whereby the composition type includes: individuals and teams;

[0069] When the composition type corresponding to the first collector is a team, obtain the team information corresponding to the first collector, and the team information includes: multiple first collectors;

[0070] Attempt to obtain the guarantee circle of the first collector. If successful, the corresponding first collector will be designated as the second collector, and the remaining first collectors will be designated as the third collectors.

[0071] Obtain the second credit value of the second collector; if the second credit value is less than or equal to a preset second threshold and / or the third collector is not within the guarantee circle, remove the corresponding first collector.

[0072] Once all the first collectors that need to be removed from the first collectors have been removed, the remaining first collectors will be used as the second collectors.

[0073] Acquire at least one new first data item newly collected by the second collector, and simultaneously acquire the collection process of the second collector collecting the first data item;

[0074] The collection process is analyzed and broken down to obtain multiple first processes;

[0075] Obtain a preset importance analysis model, input the first process into the importance analysis model, and obtain the importance value;

[0076] If the importance value is greater than or equal to a preset third threshold, the first process will be used as the second process.

[0077] The collection method corresponding to the second process is obtained, and the collection method includes: online collection and physical collection;

[0078] When the collection method corresponding to the second process is network collection, obtain the collection scenario corresponding to the second process;

[0079] Obtain a preset collection simulation model, and use the collection simulation model to simulate data collection in the collection scenario to obtain simulation records;

[0080] Obtain a preset risk assessment model, input the simulation record into the risk assessment model, and obtain the risk value;

[0081] If the risk value is greater than or equal to the preset fourth threshold, the corresponding first data item is removed;

[0082] When the collection method corresponding to the second process is actual collection, obtain the collection object corresponding to the second process and the first authentication identifier corresponding to the collection object;

[0083] Based on a preset collection object-authentication identifier library, determine the second authentication identifier corresponding to the collection object;

[0084] The first authentication identifier and the second authentication identifier are compared and analyzed to obtain a second analysis result, which includes: consistency and inconsistency.

[0085] When the second analysis result is inconsistent, the corresponding first data item is removed;

[0086] Once all the first data items that need to be removed from the first data items have been removed, the remaining first data items will be used as the second data items.

[0087] Integrate the second data items to obtain medical data and complete the acquisition.

[0088] Preferably, the acquisition module performs the following operations:

[0089] Obtain multiple second viewing intents generated by the user in history, and simultaneously obtain the generation time corresponding to the second viewing intent;

[0090] Based on the generation time, the second viewing intent is sorted chronologically to obtain an intent sequence;

[0091] The last second viewing intent in the intent sequence is selected as the third viewing intent. Meanwhile, in the intent sequence, the second viewing intents are traversed sequentially from the third viewing intent backward.

[0092] A preset intent association analysis model is obtained. The second viewing intent and the third viewing intent that have been traversed are input into the intent association analysis model. The intent association analysis model performs intent association analysis and obtains a third analysis result. The third analysis result includes: association and non-association.

[0093] When the third analysis result is not related, stop the traversal and take the second viewing intent encountered during the traversal as the fourth viewing intent;

[0094] The preset intent prediction model takes the third viewing intent and the fourth viewing intent as inputs and attempts to predict the intent.

[0095] If the intent prediction is successful, obtain the first intent prediction result, and use the first intent prediction result as the user's first viewing intent to complete the acquisition.

[0096] Otherwise, obtain the reason for the failure to obtain the intent prediction, which includes: multiple first requirement items;

[0097] Based on a preset requirement-behavior capture strategy library, at least one behavior capture strategy corresponding to the first requirement is determined;

[0098] Based on the behavior capture strategy, attempt to capture at least one first behavior item of the user corresponding to the first demand item; if the capture is successful, obtain the capture process of capturing the first behavior item.

[0099] The capture process is analyzed and broken down to obtain multiple third processes, and the sequence of the third processes is obtained.

[0100] Based on the chronological order of the processes, the third process is sorted to obtain a process sequence;

[0101] The third process that captures the first behavior item in the third process is determined and used as the fourth process;

[0102] Obtain the first capture scene corresponding to the fourth process, and at the same time, obtain the first confidence level corresponding to the first capture scene. If the first confidence level is less than or equal to a preset fifth threshold, remove the corresponding first behavior item.

[0103] Otherwise, a predetermined number of the third processes before and / or after the fourth process in the process sequence are selected and used as the fifth process;

[0104] Obtain the second capture scene corresponding to the fifth process, and at the same time, obtain the second confidence level corresponding to the second capture scene. If the second confidence level is less than or equal to the preset sixth threshold, remove the corresponding first behavior item.

[0105] After all the first behavior items that need to be removed from the first behavior items have been removed, determine whether there are any remaining first behavior items. If so, use the remaining first behavior items as second behavior items, and at the same time, use the corresponding first requirement item as the second requirement item.

[0106] The second action item is then input into the intent prediction model.

[0107] Once all the first behavioral items that need to be supplemented into the intent prediction model have been input, a third requirement item other than the second requirement item is determined from the first requirement items.

[0108] If the number of the third requirement items is 0, obtain the second prediction result obtained by the intent prediction model to re-predict the intent, and use the second prediction result as the user's first viewing intent to complete the acquisition;

[0109] Otherwise, the user is prompted to input the requirement information corresponding to the third requirement item. When the user finishes inputting, the requirement information is added to the intent prediction model, and the third prediction result obtained by the intent prediction model is obtained by re-predicting the intent. The third prediction result is used as the user's first viewing intent, and the acquisition is completed.

[0110] Preferably, the determining module performs the following operations:

[0111] Build a library of viewing intent analysis strategies;

[0112] Based on the viewing intent-analysis strategy library, the first analysis strategy corresponding to the first viewing intent is determined.

[0113] Preferably, the medical data visualization implementation system also includes:

[0114] The response module is used to obtain the user's operation on the visualization model and respond to the operation.

[0115] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0116] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0117] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0118] Figure 1 This is a schematic diagram illustrating a method for visualizing medical data according to an embodiment of the present invention;

[0119] Figure 2 This is a schematic diagram illustrating another method for implementing medical data visualization in an embodiment of the present invention;

[0120] Figure 3 This is a schematic diagram of a medical data visualization implementation system according to an embodiment of the present invention. Detailed Implementation

[0121] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0122] This invention provides a method for implementing medical data visualization, such as... Figure 1 As shown, it includes:

[0123] Step S1: Obtain medical data and, at the same time, obtain the user's initial viewing intent;

[0124] Step S2: Based on the first viewing intent, determine the first analysis strategy;

[0125] Step S3: Based on the first analysis strategy, analyze the medical data to obtain the first analysis result;

[0126] Step S4: Input the first analysis result into the preset visualization model for visualization display.

[0127] The working principle and beneficial effects of the above technical solution are as follows:

[0128] Acquire medical data (e.g., patient basic information, chief complaint, laboratory data, imaging data, diagnostic data, and treatment data), and simultaneously, obtain the user's initial viewing intent (e.g., mortality rate corresponding to each disease); based on the initial viewing intent, determine the first analysis strategy (e.g., analyze the mortality rate corresponding to each disease); based on the first analysis strategy, analyze the medical data to obtain the first analysis result (e.g., mortality rate corresponding to each disease); input the first analysis result into a preset visualization model for visualization display (e.g., using bar charts to represent the magnitude of disease mortality rates).

[0129] This invention directly obtains the user's first viewing intent, determines a first analysis strategy based on the first viewing intent, and performs corresponding analysis on the medical data without requiring the user to perform the analysis, thus improving convenience. At the same time, the analysis results are directly input into a visualization model for visualization display, thus improving intuitiveness.

[0130] This invention provides a method for visualizing medical data. Step S1, acquiring medical data, includes:

[0131] Obtain a preset collection set, the collection set including: multiple first collections;

[0132] Obtain the first credit value corresponding to the first collector; if the first credit value is less than or equal to a preset first threshold, remove the corresponding first collector.

[0133] Otherwise, obtain the composition type corresponding to the first collector, whereby the composition type includes: individuals and teams;

[0134] When the composition type corresponding to the first collector is a team, obtain the team information corresponding to the first collector, and the team information includes: multiple first collectors;

[0135] Attempt to obtain the guarantee circle of the first collector. If successful, the corresponding first collector will be designated as the second collector, and the remaining first collectors will be designated as the third collectors.

[0136] Obtain the second credit value of the second collector; if the second credit value is less than or equal to a preset second threshold and / or the third collector is not within the guarantee circle, remove the corresponding first collector.

[0137] Once all the first collectors that need to be removed from the first collectors have been removed, the remaining first collectors will be used as the second collectors.

[0138] Acquire at least one new first data item newly collected by the second collector, and simultaneously acquire the collection process of the second collector collecting the first data item;

[0139] The collection process is analyzed and broken down to obtain multiple first processes;

[0140] Obtain a preset importance analysis model, input the first process into the importance analysis model, and obtain the importance value;

[0141] If the importance value is greater than or equal to a preset third threshold, the first process will be used as the second process.

[0142] The collection method corresponding to the second process is obtained, and the collection method includes: online collection and physical collection;

[0143] When the collection method corresponding to the second process is network collection, obtain the collection scenario corresponding to the second process;

[0144] Obtain a preset collection simulation model, and use the collection simulation model to simulate data collection in the collection scenario to obtain simulation records;

[0145] Obtain a preset risk assessment model, input the simulation record into the risk assessment model, and obtain the risk value;

[0146] If the risk value is greater than or equal to the preset fourth threshold, the corresponding first data item is removed;

[0147] When the collection method corresponding to the second process is actual collection, obtain the collection object corresponding to the second process and the first authentication identifier corresponding to the collection object;

[0148] Based on a preset collection object-authentication identifier library, determine the second authentication identifier corresponding to the collection object;

[0149] The first authentication identifier and the second authentication identifier are compared and analyzed to obtain a second analysis result, which includes: consistency and inconsistency.

[0150] When the second analysis result is inconsistent, the corresponding first data item is removed;

[0151] Once all the first data items that need to be removed from the first data items have been removed, the remaining first data items will be used as the second data items.

[0152] Integrate the second data items to obtain medical data and complete the acquisition.

[0153] The working principle and beneficial effects of the above technical solution are as follows:

[0154] Obtain the first credit score of the first data collector (the individual or team collecting medical data), which can be determined based on the authenticity of the medical data collected by the first data collector in the past. If the first credit score is less than a preset first threshold (e.g., 98), the data collected by the first data collector is unreliable and is directly removed. The first data collector can be either an individual or a team. When the composition type is an individual, the first credit score only needs to meet the requirements. If the composition type is a team, further verification is required. Obtain the guarantee circle of the second data collector (the guarantee circle includes personnel guaranteed by the first data collector). If all third data collectors are within the guarantee circle and the second data collector's second credit score (personal credit score) is high, the data collection is considered successful. If the data collection method reaches a preset second threshold (e.g., 97), the first data collector is considered trustworthy; otherwise, it is discarded. The collection process of the remaining second data collectors' new mobile phone data items (e.g., collection online and collection at a hospital) is obtained. This process is analyzed and broken down to obtain multiple first processes. A preset importance analysis model (generated using machine learning algorithms to learn from a large number of records of manual importance analysis of the collection process) is obtained. The importance of each first process is analyzed to obtain an importance value (the higher the importance value, the greater the importance). If the importance value is greater than or equal to a preset third threshold (e.g., 75), the corresponding first process is considered important. The second process involves two data collection methods: online collection and offline collection. When online collection is used, the process involves acquiring the corresponding collection scenario (e.g., a hospital website and a medical database), obtaining a pre-defined collection simulation model (a model generated by learning from a large number of manually collected online records using machine learning algorithms), simulating data collection in the specified scenario, and obtaining simulated records. A pre-defined risk assessment model (a model generated by learning from a large number of manually assessed risk records using machine learning algorithms) is then acquired, and the simulated records are used to conduct a risk assessment (e.g., the simulated records contain multiple third-party jumps). (Forwarding links carries significant risk, including the possibility of inaccurate data collection.) A risk value is obtained (the higher the risk value, the greater the risk). If the risk value is greater than or equal to a preset fourth threshold (e.g., 10), the corresponding first data item is removed. When the collection method is actual data collection, the corresponding collection object (e.g., a hospital) and the object's first authentication identifier (identity identifier on the collected data) are obtained. Based on a preset collection object-authentication identifier database (a database containing authentication identifiers corresponding to different collection objects), a second authentication identifier is determined. The two identifiers are compared and analyzed; if they are inconsistent, the first data item is unreliable and is removed. Integrating the remaining second data items completes the acquisition of medical data.

[0155] In this embodiment of the invention, when acquiring medical data, the first collector is initially verified based on its first credit value. Then, verification is performed separately based on the different composition types of the first collector, and the first collector that fails verification is eliminated to ensure the reliability of the personnel collecting medical data. Then, the first data items newly collected by the remaining second collectors are obtained to verify the collection process. Based on the importance value, important second processes are verified to save verification resources. Verification is performed separately based on different collection methods to ensure the accuracy of medical data acquisition.

[0156] This invention provides a method for visualizing medical data. In step S1, obtaining the user's first viewing intent includes:

[0157] Obtain multiple second viewing intents generated by the user in history, and simultaneously obtain the generation time corresponding to the second viewing intent;

[0158] Based on the generation time, the second viewing intent is sorted chronologically to obtain an intent sequence;

[0159] The last second viewing intent in the intent sequence is selected as the third viewing intent. Meanwhile, in the intent sequence, the second viewing intents are traversed sequentially from the third viewing intent backward.

[0160] A preset intent association analysis model is obtained. The second viewing intent and the third viewing intent that have been traversed are input into the intent association analysis model. The intent association analysis model performs intent association analysis and obtains a third analysis result. The third analysis result includes: association and non-association.

[0161] When the third analysis result is not related, stop the traversal and take the second viewing intent encountered during the traversal as the fourth viewing intent;

[0162] The preset intent prediction model takes the third viewing intent and the fourth viewing intent as inputs and attempts to predict the intent.

[0163] If the intent prediction is successful, obtain the first intent prediction result, and use the first intent prediction result as the user's first viewing intent to complete the acquisition.

[0164] Otherwise, obtain the reason for the failure to obtain the intent prediction, which includes: multiple first requirement items;

[0165] Based on a preset requirement-behavior capture strategy library, at least one behavior capture strategy corresponding to the first requirement is determined;

[0166] Based on the behavior capture strategy, attempt to capture at least one first behavior item of the user corresponding to the first demand item; if the capture is successful, obtain the capture process of capturing the first behavior item.

[0167] The capture process is analyzed and broken down to obtain multiple third processes, and the sequence of the third processes is obtained.

[0168] Based on the chronological order of the processes, the third process is sorted to obtain a process sequence;

[0169] The third process that captures the first behavior item in the third process is determined and used as the fourth process;

[0170] Obtain the first capture scene corresponding to the fourth process, and at the same time, obtain the first confidence level corresponding to the first capture scene. If the first confidence level is less than or equal to a preset fifth threshold, remove the corresponding first behavior item.

[0171] Otherwise, a predetermined number of the third processes before and / or after the fourth process in the process sequence are selected and used as the fifth process;

[0172] Obtain the second capture scene corresponding to the fifth process, and at the same time, obtain the second confidence level corresponding to the second capture scene. If the second confidence level is less than or equal to the preset sixth threshold, remove the corresponding first behavior item.

[0173] After all the first behavior items that need to be removed from the first behavior items have been removed, determine whether there are any remaining first behavior items. If so, use the remaining first behavior items as second behavior items, and at the same time, use the corresponding first requirement item as the second requirement item.

[0174] The second action item is then input into the intent prediction model.

[0175] Once all the first behavioral items that need to be supplemented into the intent prediction model have been input, a third requirement item other than the second requirement item is determined from the first requirement items.

[0176] If the number of the third requirement items is 0, obtain the second prediction result obtained by the intent prediction model to re-predict the intent, and use the second prediction result as the user's first viewing intent to complete the acquisition;

[0177] Otherwise, the user is prompted to input the requirement information corresponding to the third requirement item. When the user finishes inputting, the requirement information is added to the intent prediction model, and the third prediction result obtained by the intent prediction model is obtained by re-predicting the intent. The third prediction result is used as the user's first viewing intent, and the acquisition is completed.

[0178] The working principle and beneficial effects of the above technical solution are as follows:

[0179] When obtaining a user's first viewing intent, we can obtain the user's historical second viewing intents for intent prediction. However, the user's latest first viewing intent... Figure 1It must be related to the most recently generated second viewing intent. Furthermore, the most recently generated second viewing intent needs to be correlated (for example, if the user's most recent second viewing intents are 1, to view disease death cases; 2, to view mortality rates for different diseases, then the user's latest first viewing intent can be predicted to be 3, to view the disease with the highest mortality rate and related cases). Therefore, starting from the last third viewing sequence in the intent sequence, the second viewing intents are traversed backwards sequentially. The traversed second and third viewing intents are input into a preset intent association analysis model (a model generated by learning from a large number of records of manual intent association analysis using machine learning algorithms) to analyze the intents of both. The process involves identifying and identifying related intents. If a connection exists, the traversal continues forward; otherwise, it stops, and the second viewing intent is used as the fourth viewing intent. The third and fourth viewing intents are then input into a pre-defined intent prediction model (generated using machine learning algorithms to learn from a large number of manually predicted intent records). Theoretically, this should lead to a successful prediction. However, due to limitations in intent correlation, the available data for intent prediction may be insufficient. Therefore, the intent prediction model can automatically output reasons for intent prediction failure. These reasons include multiple primary requirements (e.g., which disease death cases were viewed). Based on a preset requirement-capture strategy library (a database containing capture strategies corresponding to different first requirement items, specifically, for example, capturing user access records), a capture strategy is determined, and the first behavior (access behavior) is captured. If capture is successful, the capture process is obtained, the capture process is analyzed and decomposed, and the process sequence is obtained. The fourth process (usually the last in the process sequence) that captures the first behavior item is determined, and the first capture scenario corresponding to the fourth process (e.g., a certain access record database) is obtained. The first confidence level of the first capture scenario is obtained. If the first confidence level is less than or equal to a preset fifth threshold (e.g., ... For example, if the fourth process is unreliable (e.g., 99), the corresponding first row item is removed. Otherwise, due to the correlation between processes, the fifth process is related to the preset number (e.g., 3) before and / or after the fourth process. The second credibility of the second capture scene corresponding to the fifth process also needs to be verified. If the second credibility is less than or equal to the preset sixth threshold (e.g., 98), the corresponding first row item is removed. The remaining second row items are input into the supplementary input intent prediction model. If there are no remaining third requirement items, it means that all requirements are met, and the second prediction result is directly obtained as the user's first viewing intent. If there are remaining third requirement items, the user inputs them himself, and the intent prediction is performed again.

[0180] In this embodiment of the invention, when obtaining a user's first viewing intent, intent prediction is performed based on historical second viewing intents. This fully considers the special requirements of using second viewing intents for intent prediction (related to recent intents, and recent intents need to be correlated). Based on these special requirements, a traversal verification is performed. Verified third and fourth viewing intents are input into the intent prediction model for further intent prediction. When prediction fails, a capture strategy is used based on the first requirement item for the reason for prediction failure to further capture the intent. When captured, the capture process is verified to ensure accuracy. Only when a third requirement item exists does the user need to input it manually. In data research, users often generate multiple intents in a series, and inputting them one by one is cumbersome. This invention greatly improves convenience, ensuring accuracy in intent prediction while being more user-friendly.

[0181] This invention provides a method for visualizing medical data. Step S2: Based on the first viewing intent, determine a first analysis strategy, including:

[0182] Build a library of viewing intent analysis strategies;

[0183] Based on the viewing intent-analysis strategy library, the first analysis strategy corresponding to the first viewing intent is determined.

[0184] The working principle and beneficial effects of the above technical solution are as follows:

[0185] Construct a viewing intent-analysis strategy library (a database containing the first analysis strategy corresponding to different viewing intents). Based on this viewing intent-analysis strategy library, determine the first analysis strategy corresponding to the first viewing intent, thus completing the determination.

[0186] This invention provides a method for implementing medical data visualization, wherein the construction of a viewing intent-analysis strategy library includes:

[0187] Obtain a preset set of viewing intents, the set of viewing intents including: multiple fifth viewing intents;

[0188] Obtain multiple alternative second analysis strategies corresponding to the fifth viewing intent;

[0189] Obtain at least one test record item corresponding to the second analysis strategy, wherein the test record item includes: a first test strategy and a first test result;

[0190] Based on a preset test strategy and an adaptation feature library, at least one first feature adapted to the first test strategy and an importance value corresponding to the first feature are determined.

[0191] The second analysis strategy is used to perform feature analysis and extraction to obtain at least one second feature and a feature value corresponding to the second feature;

[0192] Perform feature matching between the first feature and the second feature. If a match is found, obtain the matching value between the first feature and the second feature.

[0193] The determination index of the first testing strategy is calculated based on the importance value, feature value, and matching value, using the following formula:

[0194]

[0195] Where γ is the decision index of the first testing strategy, and α t To match the t-th first feature with the corresponding second feature, σ t To match the importance value corresponding to the t-th first feature that meets the criteria, d t The feature value of the t-th matching second feature is O, where O is the total number of matching first or second features, and ε1 and ε2 are preset weight values.

[0196] If the judgment index is greater than or equal to the preset seventh threshold, the first test strategy will be used as the second test strategy, the value corresponding to the second test strategy will be obtained, and the first test result corresponding to the second test strategy will be used as the second test result.

[0197] Extract the result value from the second test result, and calculate the ranking index based on the value and result value. The calculation formula is as follows:

[0198]

[0199] Where β is the sorting index, μ i For the i-th second test strategy, l i is the result value in the second test result corresponding to the i-th second test strategy, and m is the total number of the second test strategies;

[0200] The second analysis strategy corresponding to the largest ranking index is paired with the corresponding fifth viewing intent to obtain a pairing item, and the pairing item is input into a preset blank database.

[0201] Once all required matching items for the blank database have been entered, the blank database will be used as the viewing intent-analysis strategy library, thus completing the construction.

[0202] The working principle and beneficial effects of the above technical solution are as follows:

[0203] The fifth viewing intent is the user's potential viewing intent; obtain the second analysis strategy (multiple alternative analysis strategies) corresponding to the fifth viewing intent; obtain the test record items corresponding to the second analysis strategy, which include the first test strategy (e.g., test analysis effect) and the first test result; based on the preset test strategy-adaptation feature library (a database containing adaptation features corresponding to different test strategies, specifically, the features of the analysis strategy suitable for testing the test strategy), determine the first adaptation feature and its importance value (the importance value represents the degree of importance of the first feature adaptation); perform feature analysis and extraction on the first analysis strategy to obtain the second feature and its feature values ​​(the larger the feature value, the greater the proportion of the second feature in the total analysis strategy). The larger the ratio, the better the first test strategy is. Match the first feature and the second feature to obtain the matching value. Calculate the judgment index based on the importance value, feature value, and matching value (in the formula, the importance value, feature value, and matching value are all positively correlated with the judgment index). The larger the judgment index, the more suitable the first test strategy is. If the judgment index is greater than or equal to the preset seventh threshold (e.g., 75), the corresponding first test strategy is used as the second test strategy. Obtain the value of the second test strategy (representing the test value). Calculate the ranking index based on the value and the result value (in the formula, the result value and value are both positively correlated with the ranking index). The larger the ranking index, the better the corresponding second analysis strategy is. Pair the second analysis strategy corresponding to the maximum ranking index with the fifth viewing intent and input it into the preset blank database.

[0204] This invention constructs a viewing intent-analysis strategy library. Based on adaptation considerations, it calculates a judgment index to quickly eliminate unqualified first test record items. Based on the result values ​​of the remaining second test records and the value of the corresponding second test strategy, it calculates a ranking index to quickly determine the optimal analysis strategy and pair it with the corresponding fifth viewing intent, thereby improving the system's work efficiency.

[0205] This invention provides a method for implementing medical data visualization, such as... Figure 2 As shown, it also includes:

[0206] Step S5: Obtain the user's operation on the visualization model, and respond to the operation.

[0207] The working principle and beneficial effects of the above technical solution are as follows:

[0208] Get the user's actions on the visualization model (e.g., zoom in, zoom out, and adjust colors) and respond accordingly.

[0209] This invention provides a system for implementing medical data visualization, such as... Figure 3 As shown, it includes:

[0210] Module 1 is used to acquire medical data and, at the same time, the user's initial viewing intent.

[0211] Module 2 is used to determine a first analysis strategy based on the first viewing intent;

[0212] Analysis module 3 is used to analyze the medical data based on the first analysis strategy and obtain a first analysis result;

[0213] Display module 4 is used to input the first analysis result into a preset visualization model for visualization display.

[0214] The working principle and beneficial effects of the above technical solution are as follows:

[0215] Acquire medical data (e.g., patient basic information, chief complaint, laboratory data, imaging data, diagnostic data, and treatment data), and simultaneously, obtain the user's initial viewing intent (e.g., mortality rate corresponding to each disease); based on the initial viewing intent, determine the first analysis strategy (e.g., analyze the mortality rate corresponding to each disease); based on the first analysis strategy, analyze the medical data to obtain the first analysis result (e.g., mortality rate corresponding to each disease); input the first analysis result into a preset visualization model for visualization display (e.g., using bar charts to represent the magnitude of disease mortality rates).

[0216] This invention directly obtains the user's first viewing intent, determines a first analysis strategy based on the first viewing intent, and performs corresponding analysis on the medical data without requiring the user to perform the analysis, thus improving convenience. At the same time, the analysis results are directly input into a visualization model for visualization display, thus improving intuitiveness.

[0217] This invention provides a system for visualizing medical data, wherein the acquisition module 1 performs the following operations:

[0218] Obtain a preset collection set, the collection set including: multiple first collections;

[0219] Obtain the first credit value corresponding to the first collector; if the first credit value is less than or equal to a preset first threshold, remove the corresponding first collector.

[0220] Otherwise, obtain the composition type corresponding to the first collector, whereby the composition type includes: individuals and teams;

[0221] When the composition type corresponding to the first collector is a team, obtain the team information corresponding to the first collector, and the team information includes: multiple first collectors;

[0222] Attempt to obtain the guarantee circle of the first collector. If successful, the corresponding first collector will be designated as the second collector, and the remaining first collectors will be designated as the third collectors.

[0223] Obtain the second credit value of the second collector; if the second credit value is less than or equal to a preset second threshold and / or the third collector is not within the guarantee circle, remove the corresponding first collector.

[0224] Once all the first collectors that need to be removed from the first collectors have been removed, the remaining first collectors will be used as the second collectors.

[0225] Acquire at least one new first data item newly collected by the second collector, and simultaneously acquire the collection process of the second collector collecting the first data item;

[0226] The collection process is analyzed and broken down to obtain multiple first processes;

[0227] Obtain a preset importance analysis model, input the first process into the importance analysis model, and obtain the importance value;

[0228] If the importance value is greater than or equal to a preset third threshold, the first process will be used as the second process.

[0229] The collection method corresponding to the second process is obtained, and the collection method includes: online collection and physical collection;

[0230] When the collection method corresponding to the second process is network collection, obtain the collection scenario corresponding to the second process;

[0231] Obtain a preset collection simulation model, and use the collection simulation model to simulate data collection in the collection scenario to obtain simulation records;

[0232] Obtain a preset risk assessment model, input the simulation record into the risk assessment model, and obtain the risk value;

[0233] If the risk value is greater than or equal to the preset fourth threshold, the corresponding first data item is removed;

[0234] When the collection method corresponding to the second process is actual collection, obtain the collection object corresponding to the second process and the first authentication identifier corresponding to the collection object;

[0235] Based on a preset collection object-authentication identifier library, determine the second authentication identifier corresponding to the collection object;

[0236] The first authentication identifier and the second authentication identifier are compared and analyzed to obtain a second analysis result, which includes: consistency and inconsistency.

[0237] When the second analysis result is inconsistent, the corresponding first data item is removed;

[0238] Once all the first data items that need to be removed from the first data items have been removed, the remaining first data items will be used as the second data items.

[0239] Integrate the second data items to obtain medical data and complete the acquisition.

[0240] The working principle and beneficial effects of the above technical solution are as follows:

[0241] Obtain the first credit score of the first data collector (the individual or team collecting medical data), which can be determined based on the authenticity of the medical data collected by the first data collector in the past. If the first credit score is less than a preset first threshold (e.g., 98), the data collected by the first data collector is unreliable and is directly removed. The first data collector can be either an individual or a team. When the composition type is an individual, the first credit score only needs to meet the requirements. If the composition type is a team, further verification is required. Obtain the guarantee circle of the second data collector (the guarantee circle includes personnel guaranteed by the first data collector). If all third data collectors are within the guarantee circle and the second data collector's second credit score (personal credit score) is high, the data collection is considered successful. If the data collection method reaches a preset second threshold (e.g., 97), the first data collector is considered trustworthy; otherwise, it is discarded. The collection process of the remaining second data collectors' new mobile phone data items (e.g., collection online and collection at a hospital) is obtained. This process is analyzed and broken down to obtain multiple first processes. A preset importance analysis model (generated using machine learning algorithms to learn from a large number of records of manual importance analysis of the collection process) is obtained. The importance of each first process is analyzed to obtain an importance value (the higher the importance value, the greater the importance). If the importance value is greater than or equal to a preset third threshold (e.g., 75), the corresponding first process is considered important. The second process involves two data collection methods: online collection and offline collection. When online collection is used, the process involves acquiring the corresponding collection scenario (e.g., a hospital website and a medical database), obtaining a pre-defined collection simulation model (a model generated by learning from a large number of manually collected online records using machine learning algorithms), simulating data collection in the specified scenario, and obtaining simulated records. A pre-defined risk assessment model (a model generated by learning from a large number of manually assessed risk records using machine learning algorithms) is then acquired, and the simulated records are used to conduct a risk assessment (e.g., the simulated records contain multiple third-party jumps). (Forwarding links carries significant risk, including the possibility of inaccurate data collection.) A risk value is obtained (the higher the risk value, the greater the risk). If the risk value is greater than or equal to a preset fourth threshold (e.g., 10), the corresponding first data item is removed. When the collection method is actual data collection, the corresponding collection object (e.g., a hospital) and the object's first authentication identifier (identity identifier on the collected data) are obtained. Based on a preset collection object-authentication identifier database (a database containing authentication identifiers corresponding to different collection objects), a second authentication identifier is determined. The two identifiers are compared and analyzed; if they are inconsistent, the first data item is unreliable and is removed. Integrating the remaining second data items completes the acquisition of medical data.

[0242] In this embodiment of the invention, when acquiring medical data, the first collector is initially verified based on its first credit value. Then, verification is performed separately based on the different composition types of the first collector, and the first collector that fails verification is eliminated to ensure the reliability of the personnel collecting medical data. Then, the first data items newly collected by the remaining second collectors are obtained to verify the collection process. Based on the importance value, important second processes are verified to save verification resources. Verification is performed separately based on different collection methods to ensure the accuracy of medical data acquisition.

[0243] This invention provides a system for visualizing medical data, wherein the acquisition module 1 performs the following operations:

[0244] Obtain multiple second viewing intents generated by the user in history, and simultaneously obtain the generation time corresponding to the second viewing intent;

[0245] Based on the generation time, the second viewing intent is sorted chronologically to obtain an intent sequence;

[0246] The last second viewing intent in the intent sequence is selected as the third viewing intent. Meanwhile, in the intent sequence, the second viewing intents are traversed sequentially from the third viewing intent backward.

[0247] A preset intent association analysis model is obtained. The second viewing intent and the third viewing intent that have been traversed are input into the intent association analysis model. The intent association analysis model performs intent association analysis and obtains a third analysis result. The third analysis result includes: association and non-association.

[0248] When the third analysis result is not related, stop the traversal and take the second viewing intent encountered during the traversal as the fourth viewing intent;

[0249] The preset intent prediction model takes the third viewing intent and the fourth viewing intent as inputs and attempts to predict the intent.

[0250] If the intent prediction is successful, obtain the first intent prediction result, and use the first intent prediction result as the user's first viewing intent to complete the acquisition.

[0251] Otherwise, obtain the reason for the failure to obtain the intent prediction, which includes: multiple first requirement items;

[0252] Based on a preset requirement-behavior capture strategy library, at least one behavior capture strategy corresponding to the first requirement is determined;

[0253] Based on the behavior capture strategy, attempt to capture at least one first behavior item of the user corresponding to the first demand item; if the capture is successful, obtain the capture process of capturing the first behavior item.

[0254] The capture process is analyzed and broken down to obtain multiple third processes, and the sequence of the third processes is obtained.

[0255] Based on the chronological order of the processes, the third process is sorted to obtain a process sequence;

[0256] The third process that captures the first behavior item in the third process is determined and used as the fourth process;

[0257] Obtain the first capture scene corresponding to the fourth process, and at the same time, obtain the first confidence level corresponding to the first capture scene. If the first confidence level is less than or equal to a preset fifth threshold, remove the corresponding first behavior item.

[0258] Otherwise, a predetermined number of the third processes before and / or after the fourth process in the process sequence are selected and used as the fifth process;

[0259] Obtain the second capture scene corresponding to the fifth process, and at the same time, obtain the second confidence level corresponding to the second capture scene. If the second confidence level is less than or equal to the preset sixth threshold, remove the corresponding first behavior item.

[0260] After all the first behavior items that need to be removed from the first behavior items have been removed, determine whether there are any remaining first behavior items. If so, use the remaining first behavior items as second behavior items, and at the same time, use the corresponding first requirement item as the second requirement item.

[0261] The second action item is then input into the intent prediction model.

[0262] Once all the first behavioral items that need to be supplemented into the intent prediction model have been input, a third requirement item other than the second requirement item is determined from the first requirement items.

[0263] If the number of the third requirement items is 0, obtain the second prediction result obtained by the intent prediction model to re-predict the intent, and use the second prediction result as the user's first viewing intent to complete the acquisition;

[0264] Otherwise, the user is prompted to input the requirement information corresponding to the third requirement item. When the user finishes inputting, the requirement information is added to the intent prediction model, and the third prediction result obtained by the intent prediction model is obtained by re-predicting the intent. The third prediction result is used as the user's first viewing intent, and the acquisition is completed.

[0265] The working principle and beneficial effects of the above technical solution are as follows:

[0266] When obtaining a user's first viewing intent, we can obtain the user's historical second viewing intents for intent prediction. However, the user's latest first viewing intent... Figure 1It must be related to the most recently generated second viewing intent. Furthermore, the most recently generated second viewing intent needs to be correlated (for example, if the user's most recent second viewing intents are 1, to view disease death cases; 2, to view mortality rates for different diseases, then the user's latest first viewing intent can be predicted to be 3, to view the disease with the highest mortality rate and related cases). Therefore, starting from the last third viewing sequence in the intent sequence, the second viewing intents are traversed backwards sequentially. The traversed second and third viewing intents are input into a preset intent association analysis model (a model generated by learning from a large number of records of manual intent association analysis using machine learning algorithms) to analyze the intents of both. The process involves identifying and identifying related intents. If a connection exists, the traversal continues forward; otherwise, it stops, and the second viewing intent is used as the fourth viewing intent. The third and fourth viewing intents are then input into a pre-defined intent prediction model (generated using machine learning algorithms to learn from a large number of manually predicted intent records). Theoretically, this should lead to a successful prediction. However, due to limitations in intent correlation, the available data for intent prediction may be insufficient. Therefore, the intent prediction model can automatically output reasons for intent prediction failure. These reasons include multiple primary requirements (e.g., which disease death cases were viewed). Based on a preset requirement-capture strategy library (a database containing capture strategies corresponding to different first requirement items, specifically, for example, capturing user access records), a capture strategy is determined, and the first behavior (access behavior) is captured. If capture is successful, the capture process is obtained, the capture process is analyzed and decomposed, and the process sequence is obtained. The fourth process (usually the last in the process sequence) that captures the first behavior item is determined, and the first capture scenario corresponding to the fourth process (e.g., a certain access record database) is obtained. The first confidence level of the first capture scenario is obtained. If the first confidence level is less than or equal to a preset fifth threshold (e.g., ... For example, if the fourth process is unreliable (e.g., 99), the corresponding first row item is removed. Otherwise, due to the correlation between processes, the fifth process is related to the preset number (e.g., 3) before and / or after the fourth process. The second credibility of the second capture scene corresponding to the fifth process also needs to be verified. If the second credibility is less than or equal to the preset sixth threshold (e.g., 98), the corresponding first row item is removed. The remaining second row items are input into the supplementary input intent prediction model. If there are no remaining third requirement items, it means that all requirements are met, and the second prediction result is directly obtained as the user's first viewing intent. If there are remaining third requirement items, the user inputs them himself, and the intent prediction is performed again.

[0267] In this embodiment of the invention, when obtaining a user's first viewing intent, intent prediction is performed based on historical second viewing intents. This fully considers the special requirements of using second viewing intents for intent prediction (related to recent intents, and recent intents need to be correlated). Based on these special requirements, a traversal verification is performed. Verified third and fourth viewing intents are input into the intent prediction model for further intent prediction. When prediction fails, a capture strategy is used based on the first requirement item for the reason for prediction failure to further capture the intent. When captured, the capture process is verified to ensure accuracy. Only when a third requirement item exists does the user need to input it manually. In data research, users often generate multiple intents in a series, and inputting them one by one is cumbersome. This invention greatly improves convenience, ensuring accuracy in intent prediction while being more user-friendly.

[0268] This invention provides a system for visualizing medical data, wherein the determining module 2 performs the following operations:

[0269] Build a library of viewing intent analysis strategies;

[0270] Based on the viewing intent-analysis strategy library, the first analysis strategy corresponding to the first viewing intent is determined.

[0271] The working principle and beneficial effects of the above technical solution are as follows:

[0272] Construct a viewing intent-analysis strategy library (a database containing the first analysis strategy corresponding to different viewing intents). Based on this viewing intent-analysis strategy library, determine the first analysis strategy corresponding to the first viewing intent, thus completing the determination.

[0273] This invention provides a system for visualizing medical data, wherein the determining module 2 performs the following operations:

[0274] Obtain a preset set of viewing intents, the set of viewing intents including: multiple fifth viewing intents;

[0275] Obtain multiple alternative second analysis strategies corresponding to the fifth viewing intent;

[0276] Obtain at least one test record item corresponding to the second analysis strategy, wherein the test record item includes: a first test strategy and a first test result;

[0277] Based on a preset test strategy and an adaptation feature library, at least one first feature adapted to the first test strategy and an importance value corresponding to the first feature are determined.

[0278] The second analysis strategy is used to perform feature analysis and extraction to obtain at least one second feature and a feature value corresponding to the second feature;

[0279] Perform feature matching between the first feature and the second feature. If a match is found, obtain the matching value between the first feature and the second feature.

[0280] The determination index of the first testing strategy is calculated based on the importance value, feature value, and matching value, using the following formula:

[0281]

[0282] Where γ is the decision index of the first testing strategy, and α t To match the t-th first feature with the corresponding second feature, σ t To match the importance value corresponding to the t-th first feature that meets the criteria, d t The feature value of the t-th matching second feature is O, where O is the total number of matching first or second features, and ε1 and ε2 are preset weight values.

[0283] If the judgment index is greater than or equal to the preset seventh threshold, the first test strategy will be used as the second test strategy, the value corresponding to the second test strategy will be obtained, and the first test result corresponding to the second test strategy will be used as the second test result.

[0284] Extract the result value from the second test result, and calculate the ranking index based on the value and result value. The calculation formula is as follows:

[0285]

[0286] Where β is the sorting index, μ i For the i-th second test strategy, l i is the result value in the second test result corresponding to the i-th second test strategy, and m is the total number of the second test strategies;

[0287] The second analysis strategy corresponding to the largest ranking index is paired with the corresponding fifth viewing intent to obtain a pairing item, and the pairing item is input into a preset blank database.

[0288] Once all required matching items for the blank database have been entered, the blank database will be used as the viewing intent-analysis strategy library, thus completing the construction.

[0289] The working principle and beneficial effects of the above technical solution are as follows:

[0290] The fifth viewing intent is the user's potential viewing intent; obtain the second analysis strategy (multiple alternative analysis strategies) corresponding to the fifth viewing intent; obtain the test record items corresponding to the second analysis strategy, which include the first test strategy (e.g., test analysis effect) and the first test result; based on the preset test strategy-adaptation feature library (a database containing adaptation features corresponding to different test strategies, specifically, the features of the analysis strategy suitable for testing the test strategy), determine the first adaptation feature and its importance value (the importance value represents the degree of importance of the first feature adaptation); perform feature analysis and extraction on the first analysis strategy to obtain the second feature and its feature values ​​(the larger the feature value, the greater the proportion of the second feature in the total analysis strategy). The larger the ratio, the better the first test strategy is. Match the first feature and the second feature to obtain the matching value. Calculate the judgment index based on the importance value, feature value, and matching value (in the formula, the importance value, feature value, and matching value are all positively correlated with the judgment index). The larger the judgment index, the more suitable the first test strategy is. If the judgment index is greater than or equal to the preset seventh threshold (e.g., 75), the corresponding first test strategy is used as the second test strategy. Obtain the value of the second test strategy (representing the test value). Calculate the ranking index based on the value and the result value (in the formula, the result value and value are both positively correlated with the ranking index). The larger the ranking index, the better the corresponding second analysis strategy is. Pair the second analysis strategy corresponding to the maximum ranking index with the fifth viewing intent and input it into the preset blank database.

[0291] This invention constructs a viewing intent-analysis strategy library. Based on adaptation considerations, it calculates a judgment index to quickly eliminate unqualified first test record items. Based on the result values ​​of the remaining second test records and the value of the corresponding second test strategy, it calculates a ranking index to quickly determine the optimal analysis strategy and pair it with the corresponding fifth viewing intent, thereby improving the system's work efficiency.

[0292] This invention provides a system for visualizing medical data, which further includes:

[0293] The response module is used to obtain the user's operation on the visualization model and respond to the operation.

[0294] The working principle and beneficial effects of the above technical solution are as follows:

[0295] Get the user's actions on the visualization model (e.g., zoom in, zoom out, and adjust colors) and respond accordingly.

[0296] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for implementing medical data visualization, characterized in that, include: Step S1: Obtain medical data and, at the same time, obtain the user's initial viewing intent; Step S2: Based on the first viewing intent, determine the first analysis strategy; Step S3: Based on the first analysis strategy, analyze the medical data to obtain the first analysis result; Step S4: Input the first analysis result into a preset visualization model for visualization display; In step S1, obtaining the user's first viewing intent includes: Obtain multiple second viewing intents generated by the user in history, and simultaneously obtain the generation time corresponding to the second viewing intent; Based on the generation time, the second viewing intent is sorted chronologically to obtain an intent sequence; The last second viewing intent in the intent sequence is selected as the third viewing intent. Meanwhile, in the intent sequence, the second viewing intents are traversed sequentially from the third viewing intent backward. A preset intent association analysis model is obtained. The second viewing intent and the third viewing intent that have been traversed are input into the intent association analysis model. The intent association analysis model performs intent association analysis and obtains a third analysis result. The third analysis result includes: association and non-association. When the third analysis result is not related, stop the traversal and take the second viewing intent encountered during the traversal as the fourth viewing intent; The preset intent prediction model takes the third viewing intent and the fourth viewing intent as inputs and attempts to predict the intent. If the intent prediction is successful, obtain the first intent prediction result, and use the first intent prediction result as the user's first viewing intent to complete the acquisition. Otherwise, obtain the reason for the failure of intent prediction, which includes: multiple first requirement items; Based on a preset requirement-behavior capture strategy library, at least one behavior capture strategy corresponding to the first requirement is determined; Based on the behavior capture strategy, attempt to capture at least one first behavior item of the user corresponding to the first demand item; if the capture is successful, obtain the capture process of capturing the first behavior item. The capture process is analyzed and broken down to obtain multiple third processes, and the sequence of the third processes is obtained. Based on the chronological order of the processes, the third process is sorted to obtain a process sequence; The third process that captures the first behavior item in the third process is determined and used as the fourth process; Obtain the first capture scene corresponding to the fourth process, and at the same time, obtain the first confidence level corresponding to the first capture scene. If the first confidence level is less than or equal to a preset fifth threshold, remove the corresponding first behavior item. Otherwise, a predetermined number of the third processes before and / or after the fourth process in the process sequence are selected and used as the fifth process; Obtain the second capture scene corresponding to the fifth process, and at the same time, obtain the second confidence level corresponding to the second capture scene. If the second confidence level is less than or equal to the preset sixth threshold, remove the corresponding first behavior item. After all the first behavior items that need to be removed from the first behavior items have been removed, determine whether there are any remaining first behavior items. If so, use the remaining first behavior items as second behavior items, and at the same time, use the corresponding first requirement item as the second requirement item. The second action item is then input into the intent prediction model. Once all the first behavioral items that need to be supplemented into the intent prediction model have been input, a third requirement item other than the second requirement item is determined from the first requirement items. If the number of the third requirement items is 0, obtain the second prediction result obtained by the intent prediction model to re-predict the intent, and use the second prediction result as the user's first viewing intent to complete the acquisition; Otherwise, the user is prompted to input the requirement information corresponding to the third requirement item. When the user finishes inputting, the requirement information is added to the intent prediction model, and the third prediction result obtained by the intent prediction model is obtained by re-predicting the intent. The third prediction result is used as the user's first viewing intent, and the acquisition is completed.

2. The method for implementing medical data visualization as described in claim 1, characterized in that, In step S1, acquiring medical data includes: Obtain a preset collection set, the collection set including: multiple first collections; Obtain the first credit value corresponding to the first collector; if the first credit value is less than or equal to a preset first threshold, remove the corresponding first collector. Otherwise, obtain the composition type corresponding to the first collector, whereby the composition type includes: individuals and teams; When the composition type corresponding to the first collector is a team, obtain the team information corresponding to the first collector, and the team information includes: multiple first collectors; Attempt to obtain the guarantee circle of the first collector. If successful, the corresponding first collector will be designated as the second collector, and the remaining first collectors will be designated as the third collectors. Obtain the second credit value of the second collector; if the second credit value is less than or equal to a preset second threshold and / or the third collector is not within the guarantee circle, remove the corresponding first collector. Once all the first collectors that need to be removed from the first collectors have been removed, the remaining first collectors will be used as the second collectors. Acquire at least one new first data item newly collected by the second collector, and simultaneously acquire the collection process of the second collector collecting the first data item; The collection process is analyzed and broken down to obtain multiple first processes; Obtain a preset importance analysis model, input the first process into the importance analysis model, and obtain the importance value; If the importance value is greater than or equal to a preset third threshold, the first process will be used as the second process. The collection method corresponding to the second process is obtained, and the collection method includes: online collection and physical collection; When the collection method corresponding to the second process is network collection, obtain the collection scenario corresponding to the second process; Obtain a preset collection simulation model, and use the collection simulation model to simulate data collection in the collection scenario to obtain simulation records; Obtain a preset risk assessment model, input the simulation record into the risk assessment model, and obtain the risk value; If the risk value is greater than or equal to the preset fourth threshold, the corresponding first data item is removed; When the collection method corresponding to the second process is actual collection, obtain the collection object corresponding to the second process and the first authentication identifier corresponding to the collection object; Based on a preset collection object-authentication identifier library, determine the second authentication identifier corresponding to the collection object; The first authentication identifier and the second authentication identifier are compared and analyzed to obtain a second analysis result, which includes: consistency and inconsistency. When the second analysis result is inconsistent, the corresponding first data item is removed; Once all the first data items that need to be removed from the first data items have been removed, the remaining first data items will be used as the second data items. Integrate the second data items to obtain medical data and complete the acquisition.

3. The method for implementing medical data visualization as described in claim 1, characterized in that, Step S2: Based on the first viewing intent, determine the first analysis strategy, including: Build a library of viewing intent analysis strategies; Based on the viewing intent-analysis strategy library, the first analysis strategy corresponding to the first viewing intent is determined.

4. The method for implementing medical data visualization as described in claim 1, characterized in that, Also includes: Step S5: Obtain the user's operation on the visualization model, and respond to the operation.

5. A system for visualizing medical data, characterized in that, include: The acquisition module is used to acquire medical data, and at the same time, to acquire the user's initial viewing intent; The determination module is used to determine a first analysis strategy based on the first viewing intent; An analysis module is used to analyze the medical data based on the first analysis strategy and obtain a first analysis result; The display module is used to input the first analysis result into a preset visualization model for visualization display; The acquisition module performs the following operations: Obtain multiple second viewing intents generated by the user in history, and simultaneously obtain the generation time corresponding to the second viewing intent; Based on the generation time, the second viewing intent is sorted chronologically to obtain an intent sequence; The last second viewing intent in the intent sequence is selected as the third viewing intent. Meanwhile, in the intent sequence, the second viewing intents are traversed sequentially from the third viewing intent backward. A preset intent association analysis model is obtained. The second viewing intent and the third viewing intent that have been traversed are input into the intent association analysis model. The intent association analysis model performs intent association analysis and obtains a third analysis result. The third analysis result includes: association and non-association. When the third analysis result is not related, stop the traversal and take the second viewing intent encountered during the traversal as the fourth viewing intent; The preset intent prediction model takes the third viewing intent and the fourth viewing intent as inputs and attempts to predict the intent. If the intent prediction is successful, obtain the first intent prediction result, and use the first intent prediction result as the user's first viewing intent to complete the acquisition. Otherwise, obtain the reason for the failure of intent prediction, which includes: multiple first requirement items; Based on a preset requirement-behavior capture strategy library, at least one behavior capture strategy corresponding to the first requirement is determined; Based on the behavior capture strategy, attempt to capture at least one first behavior item of the user corresponding to the first demand item; if the capture is successful, obtain the capture process of capturing the first behavior item. The capture process is analyzed and broken down to obtain multiple third processes, and the sequence of the third processes is obtained. Based on the chronological order of the processes, the third process is sorted to obtain a process sequence; The third process that captures the first behavior item in the third process is determined and used as the fourth process; Obtain the first capture scene corresponding to the fourth process, and at the same time, obtain the first confidence level corresponding to the first capture scene. If the first confidence level is less than or equal to a preset fifth threshold, remove the corresponding first behavior item. Otherwise, a predetermined number of the third processes before and / or after the fourth process in the process sequence are selected and used as the fifth process; Obtain the second capture scene corresponding to the fifth process, and at the same time, obtain the second confidence level corresponding to the second capture scene. If the second confidence level is less than or equal to the preset sixth threshold, remove the corresponding first behavior item. After all the first behavior items that need to be removed from the first behavior items have been removed, determine whether there are any remaining first behavior items. If so, use the remaining first behavior items as second behavior items, and at the same time, use the corresponding first requirement item as the second requirement item. The second action item is then input into the intent prediction model. Once all the first behavioral items that need to be supplemented into the intent prediction model have been input, a third requirement item other than the second requirement item is determined from the first requirement items. If the number of the third requirement items is 0, obtain the second prediction result obtained by the intent prediction model to re-predict the intent, and use the second prediction result as the user's first viewing intent to complete the acquisition; Otherwise, the user is prompted to input the requirement information corresponding to the third requirement item. When the user finishes inputting, the requirement information is added to the intent prediction model, and the third prediction result obtained by the intent prediction model is obtained by re-predicting the intent. The third prediction result is used as the user's first viewing intent, and the acquisition is completed.

6. The medical data visualization system as described in claim 5, characterized in that, The acquisition module performs the following operations: Obtain a preset collection set, the collection set including: multiple first collections; Obtain the first credit value corresponding to the first collector; if the first credit value is less than or equal to a preset first threshold, remove the corresponding first collector. Otherwise, obtain the composition type corresponding to the first collector, whereby the composition type includes: individuals and teams; When the composition type corresponding to the first collector is a team, obtain the team information corresponding to the first collector, and the team information includes: multiple first collectors; Attempt to obtain the guarantee circle of the first collector. If successful, the corresponding first collector will be designated as the second collector, and the remaining first collectors will be designated as the third collectors. Obtain the second credit value of the second collector; if the second credit value is less than or equal to a preset second threshold and / or the third collector is not within the guarantee circle, remove the corresponding first collector. Once all the first collectors that need to be removed from the first collectors have been removed, the remaining first collectors will be used as the second collectors. Acquire at least one new first data item newly collected by the second collector, and simultaneously acquire the collection process of the second collector collecting the first data item; The collection process is analyzed and broken down to obtain multiple first processes; Obtain a preset importance analysis model, input the first process into the importance analysis model, and obtain the importance value; If the importance value is greater than or equal to a preset third threshold, the first process will be used as the second process. The collection method corresponding to the second process is obtained, and the collection method includes: online collection and physical collection; When the collection method corresponding to the second process is network collection, obtain the collection scenario corresponding to the second process; Obtain a preset collection simulation model, and use the collection simulation model to simulate data collection in the collection scenario to obtain simulation records; Obtain a preset risk assessment model, input the simulation record into the risk assessment model, and obtain the risk value; If the risk value is greater than or equal to the preset fourth threshold, the corresponding first data item is removed; When the collection method corresponding to the second process is actual collection, obtain the collection object corresponding to the second process and the first authentication identifier corresponding to the collection object; Based on a preset collection object-authentication identifier library, determine the second authentication identifier corresponding to the collection object; The first authentication identifier and the second authentication identifier are compared and analyzed to obtain a second analysis result, which includes: consistency and inconsistency. When the second analysis result is inconsistent, the corresponding first data item is removed; Once all the first data items that need to be removed from the first data items have been removed, the remaining first data items will be used as the second data items. Integrate the second data items to obtain medical data and complete the acquisition.

7. The medical data visualization system as described in claim 5, characterized in that, The determining module performs the following operations: Build a library of viewing intent analysis strategies; Based on the viewing intent-analysis strategy library, the first analysis strategy corresponding to the first viewing intent is determined.

8. The medical data visualization system as described in claim 5, characterized in that, Also includes: The response module is used to obtain the user's operation on the visualization model and respond to the operation.

Images