Septicopyemia case data sharing system

Abstract

The invention discloses a septicopyemia case data sharing system, which relates to the technical field, is used for solving the problem that the similar case aggregation precision is reduced, and is used for generating an information evaluation coefficient by receiving case data of different information sources and counting the sharing recording proportion of each information source. Comparing the case data with data in a shared database, analyzing and evaluating case similarity, screening and marking case data in combination with an information evaluation coefficient, dividing the marked case data according to input time to generate case segments, splitting segment vocabularies, calculating vocabulary change amplitude, and evaluating associated features of the case segments; the method comprises the following steps: calculating and sorting correlation degrees between each case fragment and other case fragments, collecting data recording frequency and duration of each case fragment, analyzing a record density coefficient through the duration and generating information intensity characteristics, setting a matching stacking value of the case fragments, and performing stacking processing in combination with a correlation degree sorting result. And the aggregation precision and retrieval efficiency of similar cases are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of case data sharing, more particularly, the present application relates to a sepsis case data sharing system. BACKGROUND

[0002] Sepsis is a systemic inflammatory response syndrome caused by infection, its disease course is complex, and the disease changes rapidly, which puts forward high requirements for clinical diagnosis and treatment, in order to improve the diagnosis and treatment efficiency and research level of sepsis, in recent years, clinical and scientific research institutions try to establish sepsis case sharing database for case analysis, data sharing and scientific research.

[0003] The prior art has the following disadvantages: At present, the prior art usually depends on single index or simple matching rule for case comparison, which is difficult to accurately evaluate the similarity between cases, lacks the ability of fragmenting processing and correlation feature analysis of case data, resulting in low similarity case aggregation accuracy and increased sharing search efficiency, therefore, a sepsis case data sharing system is proposed.

[0004] The above information disclosed in the background section is only used to strengthen the understanding of the background of the present disclosure, so it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a sepsis case data sharing system, which solves the problems raised in the above background technology by using multi-source reception of case data, information evaluation coefficient generation, fragmentation analysis, vocabulary change amplitude evaluation, correlation feature sorting and matching superposition processing.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme, a sepsis case data sharing system, comprising a data receiving module, a fragment analysis module, a correlation analysis module and a sharing collection module, the functions of each module are as follows: The data receiving module receives case data from different information sources, accesses the sharing database to count the sharing collection proportion of each information source and generates information evaluation coefficient, compares and analyzes the case data with the case data in the sharing database to evaluate the similarity of the cases, filters and marks the case data combined with the information evaluation coefficient, and stores the marked case data into a temporary database and transmits it into the fragment analysis module; The fragment analysis module detects the input time of each marked case data in the temporary database, divides the marked case data according to the input time to generate each case fragment, splits the vocabulary of each case fragment to calculate the vocabulary change amplitude and transmits it into the correlation analysis module; The association analysis module receives the association characteristics of case fragments by assessing the magnitude of word changes, sorts the association degree of case fragments according to the association characteristics, collects the data recording frequency and duration of case fragments, analyzes the record density coefficient using the duration, and transmits the data recording frequency and record density coefficient to the shared collection module. The shared collection module receives the data record frequency and record density coefficient and generates the information intensity characteristics of the case fragments. Based on the information intensity characteristics, it sets the matching overlay value of the case fragments, and performs overlay processing on the case fragments in combination with the correlation ranking results. The overlay processed case data is then stored in the shared database.

[0007] In a preferred embodiment, the data receiving module connects to various information source databases via a multi-source data access interface to receive case data from different information sources; Case data refers to a comprehensive collection of data that reflects various medical information and personal information of patients throughout the entire diagnosis and treatment process; Match each news source with a shared database to obtain the number of sepsis cases that have been included in each news source; Count the total number of cases from the same information source in the shared database; The ratio of the number of sepsis cases included in each information source to the total number of cases in the information source is used as the sharing and inclusion ratio of the information source. The total sharing and inclusion ratio is obtained by adding up the sharing and inclusion ratios of each information source; The ratio of the shared inclusion rate of each information source to the total shared inclusion rate is used as the information evaluation coefficient for each information source.

[0008] In a preferred embodiment, in the data receiving module, key fields are extracted from each case data from each information source to obtain a case content set; The case content set is matched with the case data in the shared database to obtain the case similarity of each case data; The information evaluation coefficients of each information source and the case similarity of each case data are standardized to obtain the information evaluation factor and the similarity factor; The square root of the product of the information evaluation factor and the similarity factor from the same information source is used to obtain the labeling features of each case data under the information source. If the labeling characteristics of the case data are greater than the preset labeling threshold, then the case data is determined to be labeled. If the labeling characteristics of the case data are less than or equal to the preset labeling threshold, then the case data will not be labeled. The tagged case data is stored in a temporary database.

[0009] In a preferred embodiment, in the fragment analysis module, the entry time of each labeled case data in the temporary database is detected, and labeled case data with an entry time interval less than a preset time interval threshold are merged into a case data set; Keyword extraction is performed on each set of case data. The patient ID content is extracted from each labeled case data and traversed through the labeled case data. Case data with identical patient IDs are grouped into separate case segments.

[0010] In a preferred embodiment, in the fragment analysis module, a dictionary-based Chinese word segmenter segments the case description content in each case fragment, and removes stop words and punctuation marks. Each word in the case description of each labeled case data in each case segment is mapped to a word vector using the one-hot encoding method. The case vector of the labeled case data is obtained by summing the word vectors of the case description content in a case segment; For each case segment, calculate the cosine similarity between the case vectors of all labeled case data; The maximum value minus the minimum value of cosine similarity is used as the lexical change range.

[0011] In a preferred embodiment, in the association analysis module, a case segment is randomly selected, and any case segments other than the case segment are selected and merged into a case segment group. The difference between the magnitude of word changes in two case segments within a case segment group is used as the word change factor for the segment group. The case vector factor of the fragment group is obtained by subtracting the total case vector values ​​of the two case fragments in the case fragment group. The association features of case fragment groups were calculated by combining the lexical change factor and the case vector factor. The association features of case segments with other case segments are compared with preset feature thresholds to determine whether the association features should be used as the degree of association. When the correlation feature is used as the correlation degree, the correlation degree of case segments with other case segments is ranked to obtain the correlation degree ranking result.

[0012] In a preferred embodiment, in the correlation analysis module, a preset statistical time is used to count the number of times case data is recorded within the statistical time of each case segment, and the data recording frequency of each case segment is obtained by dividing the preset statistical time by the number of times the data is recorded. The patient names in each case segment are matched with the databases of each information source to obtain the case data recorded in each case segment; The duration of a case segment is obtained by subtracting the minimum value from the maximum value of the recorded time content in all recorded case data of a case segment. The ratio of the data recording frequency to the duration of each case segment is used as the recording density coefficient of each case segment.

[0013] In a preferred embodiment, in the shared recording module, the data recording frequency and recording density coefficient of each case segment are standardized to obtain the recording frequency factor and density coefficient factor. The information intensity characteristics of each case segment are obtained by weighted summation of the recording frequency factor and the density coefficient factor; The information intensity feature of each case segment is multiplied by the preset maximum matching overlay value to obtain the matching overlay value of each case segment, and the matching overlay value of each case segment is rounded down. If the information intensity feature of each case segment is greater than the preset intensity feature threshold, then the case segment is determined to be a high information intensity case segment; If the information intensity feature of each case segment is less than or equal to the preset intensity feature threshold, then the case segment is determined to be a case segment with low information intensity.

[0014] In a preferred embodiment, in the shared collection module, a case fragment with high information intensity is selected as the main fragment; Based on the correlation ranking results, case fragments with higher correlation to the main fragment are selected and stacked in descending order until the upper limit of the matching stacking value of the case fragments corresponding to the main fragment is reached. Each stacked sub-fragment matches its own stacking value and is then stacked as a sub-main fragment to the next layer of fragments, forming a hierarchical stacking structure. The stacked case data is stored in a shared database.

[0015] The technical effects and advantages of this invention are as follows: This invention receives case data from different information sources, calculates the sharing and inclusion ratio of each source to generate an information evaluation coefficient, compares and analyzes the case data with data in a shared database to assess case similarity, and uses the information evaluation coefficient to filter and label case data. The labeled case data is then divided into case segments based on the entry time. The segment vocabulary is split and the magnitude of vocabulary changes is calculated to assess the association characteristics of the case segments. The association degree of each case segment with other case segments is calculated and ranked. The data recording frequency and duration of each case segment are collected. The record density coefficient is analyzed based on the duration to generate information intensity features. A matching overlay value for case segments is set, and overlay processing is performed based on the association degree ranking results. This improves the aggregation accuracy and shared retrieval efficiency of similar cases while maintaining the semantic integrity between segments, providing technical support for the dynamic updating and multi-source collaborative analysis of case information. Attached Figure Description

[0016] Fig. 1 This is a flowchart illustrating the implementation of a sepsis case data sharing system according to the present invention.

[0017] Fig. 2 This is a module framework diagram of a sepsis case data sharing system according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] This invention receives case data from different information sources, calculates the sharing and inclusion ratio of each information source to generate an information evaluation coefficient, compares and analyzes the case data with data in a shared database to evaluate case similarity, and uses the information evaluation coefficient to filter and mark case data. The marked case data is then divided into case segments based on the entry time. The segment vocabulary is split and the magnitude of vocabulary change is calculated to evaluate the association features of the case segments. The association degree of each case segment with other case segments is calculated and ranked. The data recording frequency and duration of each case segment are collected. The record density coefficient is analyzed through the duration to generate information intensity features. A matching overlay value for case segments is set, and overlay processing is performed based on the association degree ranking results. This improves the aggregation accuracy and shared retrieval efficiency of similar cases while maintaining the semantic integrity between segments.

[0020] Example 1: A sepsis case data sharing system, such as Figs. 1-2As shown, it includes a data receiving module, a fragment analysis module, a correlation analysis module, and a shared recording module. The modules interact with each other through signal connections. The functions of each module are as follows: The data receiving module receives case data from different information sources, accesses the shared database to calculate the sharing and inclusion ratio of each information source and generates information evaluation coefficients, compares and analyzes the case data with the case data in the shared database to evaluate the case similarity, combines the information evaluation coefficients to filter and label case data, stores the labeled case data in a temporary database and transmits it to the fragment analysis module. The fragment analysis module detects the entry time of each labeled case data in the temporary database, divides the labeled case data into case fragments according to the entry time, splits the vocabulary of each case fragment, calculates the vocabulary change range, and inputs it into the association analysis module. The association analysis module receives the association characteristics of case fragments by assessing the magnitude of word changes, sorts the association degree of case fragments according to the association characteristics, collects the data recording frequency and duration of case fragments, analyzes the record density coefficient using the duration, and transmits the data recording frequency and record density coefficient to the shared collection module. The shared collection module receives the data record frequency and record density coefficient and generates the information intensity characteristics of the case fragments. Based on the information intensity characteristics, it sets the matching overlay value of the case fragments, and performs overlay processing on the case fragments in combination with the correlation ranking results. The overlay processed case data is then stored in the shared database.

[0021] The specific implementation is as follows: In the data receiving module, the system connects to various information source databases through a multi-source data access interface to receive case data from different information sources. It should be noted that the multi-source data access interface is a functional interface structure used to receive data input from different information sources, databases, or terminal devices; various information source databases refer to a collection of multiple types of external or internal data storage systems used to provide case data, including but not limited to hospital information systems, laboratory information systems, image archiving and communication systems, electronic medical record systems, and research case databases; case data refers to a comprehensive data collection used to reflect various medical information and personal information of patients throughout the entire diagnosis and treatment process, used to describe the patient's basic condition, disease evolution, examination and test results, and treatment response.

[0022] Match each news source with a shared database to obtain the number of sepsis cases that have been included in each news source; Count the total number of cases from the same information source in the shared database; The ratio of the number of sepsis cases already included in each information source to the total number of cases in that information source is taken as the shared inclusion ratio of that information source. Repeat the above steps to obtain the sharing and inclusion ratio of each information source; The total sharing and inclusion ratio is obtained by adding up the sharing and inclusion ratios of each information source; The ratio of the shared inclusion rate of each information source to the total number of shared inclusion rates is used as the information evaluation coefficient for each information source. It should be noted that the purpose of the information assessment coefficient is to quantify the reliability and data contribution quality of different information sources. The more case data that an information source has already shared, the higher the potential value of its newly provided data is considered.

[0023] Key fields were extracted from each case data from each information source, including patient ID, infection site, gender, and age; The specific process for extracting key fields is as follows: Format the data for each case from each information source; Iterate through the formatted data of each case from various information sources; When the patient ID field is encountered during iteration, extract the content of that field. When the infection site field is encountered during iteration, extract the content of that field. Repeat the above steps to continue extracting gender and age information; The content extracted from each case data from various information sources is integrated into a collection of case content; Set a similarity value for each piece of content in each case content set, with an initial value of 0; Match the collection of case data with the case data in the shared database; If each case content set from each information source has the same content as the case data in the shared database, then the similarity value of the part with the same content is incremented by 1; Conversely, if no action is taken, matching continues. The total similarity score of a case data set is obtained by summing the similarity scores of each item in the case data set. The case similarity of each case data is obtained through the above steps; It should be noted that the patient ID is a unique identifier assigned to each patient, used to uniquely identify a patient entity in the medical information system, ensuring that case data from different sources, times, and hospitals can accurately correspond to the same patient; formatting refers to the unification and standardization of case data from different information sources, giving the data a consistent structure and a format that facilitates subsequent processing. Specific processing includes: field unification (unifying all different names into standard fields), data type standardization (unifying age as an integer type), removal or conversion of non-numeric characters, field value standardization (standardizing or mapping field values), removal of redundant information (removing extra spaces, punctuation marks, or special characters); and case similarity assessment, which aims to evaluate the similarity between unentered case data and existing shared database case data. Higher similarity scores indicate a better match with existing shared database case data.

[0024] The information evaluation coefficients of each information source and the case similarity of each case data are standardized to obtain the information evaluation factor and the similarity factor; The labeling features of case data from the same information source are calculated by combining information evaluation factors and similarity factors. The calculation formula is as follows: ,in, Information evaluation factors from the same information source Similarity factors for information from the same source The labeling features of each case data under this information source; It should be noted that the larger the information evaluation factor and the larger the similarity factor of the same information source, the higher the reliability and reference value of the information source, and the larger the labeling features of each case data; conversely, the lower the reliability and reference value of the information source, the smaller the labeling features of each case data.

[0025] The above steps are used to obtain the labeling features of each case data; Based on a comparison of the labeling features of each case data with a preset labeling threshold, the following judgments are made: If the labeling characteristics of the case data are greater than the preset labeling threshold, then the case data will be labeled. If the labeling characteristics of the case data are less than or equal to the preset labeling threshold, then the case data will not be labeled. The labeled case data is stored in a temporary database and then passed to the fragment analysis module.

[0026] It should be explained that the standardization methods include, but are not limited to, standard linear transformation based on interval scaling, statistical Z-Score standardization, or normalization based on nonlinear mapping functions. The application methods of standardization will not be elaborated here. The preset labeling threshold is an important parameter for determining whether to label the case data. The distribution of labeling features of case data in the historical shared database is collected. For example, the mean, variance, median and other indicators of the labeling features are statistically analyzed. The threshold can be set to the mean or median to ensure that case data with labeling features higher than the average level are selected.

[0027] In the fragment analysis module, the entry time of each marked case data in the temporary database is detected. Marked case data with an entry time interval less than a preset time interval threshold are merged into a case data set. Keyword extraction is performed on each case data set. The patient ID content in the marked case data is extracted by traversing each marked case data set. Collect patient ID data with identical tagged cases into separate case segments; A dictionary-based Chinese word segmenter segments the case descriptions in each case fragment and removes stop words and punctuation marks. Each word in the case description of each labeled case data in each case segment is mapped to a word vector using the one-hot encoding method. The case vector of the labeled case data is obtained by summing the word vectors of the case description content in a case segment; Repeat the above steps to obtain the case vectors for each labeled case data in each case segment; For each case segment, calculate the cosine similarity between the case vectors of all its labeled case data, using the following formula: ,in, For a certain case segment, the first A case vector of labeled case data, For a certain case segment, the first A case vector of labeled case data, For the first case in the labeled case data The and the first Cosine similarity between labeled case data; The maximum value minus the minimum value of cosine similarity is used as the lexical change range; It should be noted that the specific method for setting the preset time interval threshold is usually based on statistical analysis of the data entry time distribution of historical cases, calculating the mean, median or quantile of consecutive entry time intervals, and selecting the time value that ensures that the relevant data of the same patient is completely aggregated and avoids erroneous merging across periods as the preset time interval threshold; the magnitude of word change aims to quantify the degree of difference in the case description text of the same patient at different time points. A large magnitude indicates that the patient's condition description, diagnosis or treatment has changed significantly, which may correspond to an important clinical turning point.

[0028] The total value of the case vector for each case segment is obtained by summing the case vectors of each labeled case data in each case segment; The lexical change magnitude of each case segment and the total value of the case vector are input into the association analysis module.

[0029] It needs to be explained that a dictionary-based Chinese word segmenter refers to a word segmentation technique that uses a dictionary database containing a large number of known Chinese words and their part-of-speech tags to scan, match, and segment the input text using a dictionary matching algorithm to achieve automatic word boundary recognition of continuous Chinese character sequences. Word segmentation, stop word removal, and punctuation processing are text preprocessing stages in natural language processing, the purpose of which is to transform the original case description into structured, computable lexical vectors or semantic units. One-hot encoding is a text feature vectorization method used to convert discrete textual information into a numerical form that computers can understand.

[0030] In the association analysis module, the lexical change magnitude of each case segment and the total value of the case vector are standardized: Randomly select a case segment, and then select any other case segments to merge into a case segment group; The difference between the magnitude of word changes in two case segments within a case segment group is used as the word change factor for that segment group. The case vector factor of the segment group is obtained by subtracting the total case vector values ​​of two case segments in the segment group. The association features of a case fragment group are calculated by combining the lexical change factor and the case vector factor. The calculation formula is as follows: ,in, This is the lexical change factor for this segment group. This is the case vector factor for this segment group. This refers to the association features of this segment group; Repeat the above steps until the case segment and any other case segment have been merged into a case segment group, and obtain the association characteristics between the case segment and other case segments; Perform the above operation on each case segment to obtain the association features between each case segment and other case segments; It should be noted that the larger the vocabulary change factor and the larger the case vector factor, the greater the difference between the two case segments and the smaller the correlation between the case segment and other case segments; conversely, the smaller the difference between the two case segments and the greater the correlation between the case segment and other case segments.

[0031] The correlation features of this case segment with other case segments and preset feature thresholds are compared and judged: If the correlation feature between the case segment and other case segments is greater than a preset feature threshold, then the correlation feature between the case segment and other case segments is determined as the correlation degree between the case segment and other case segments. Conversely, if the correlation between the case segment and other case segments is not found, the correlation between the case segment and other case segments is deemed invalid, and the correlation between the case segment and other case segments is not recorded. Rank the correlation between this case segment and other case segments; Repeat the above steps to obtain the correlation and ranking of each case segment with other case segments; It should be noted that the preset feature threshold is an important parameter for judging the correlation between each case segment and other case segments. All correlation features can be sorted and the high percentile value can be selected as the threshold, such as 75%, 80% or 90%.

[0032] The system sets a preset statistical time period and matches the patient names in each case segment with the databases of each information source to obtain the case data recorded within the preset statistical time period for each case segment. The frequency of case data recorded within the statistical time period for each case segment is counted and divided by the preset statistical time to obtain the data recording frequency for each case segment. The patient names in each case segment are matched again with the databases of each information source to obtain the case data recorded in each case segment; The duration of a case segment is obtained by subtracting the minimum value from the maximum value of the recorded time content in all recorded case data of a case segment. Repeat the above steps to obtain the duration of each case segment; The ratio of the data recording frequency to the duration of each case segment is used as the recording density coefficient of each case segment. The data recording frequency, recording density coefficient, and correlation ranking results of each case segment are transmitted to the shared collection module.

[0033] It should be explained that the preset statistical time is a time window used to calculate the data recording frequency of each case segment. Based on the historical update patterns of sepsis cases and business needs, a reasonable fixed time window is selected. For example, 7 days or 14 days are selected for short-term monitoring, which is suitable for observing recent update frequency. 30 days or 60 days are selected for medium-term statistics, which is used to assess recent trends and active segments. 90 days or longer are selected for long-term statistics, which covers the recording of most segments.

[0034] In the shared data collection module, the data recording frequency and recording density coefficient of each case segment are standardized to obtain the recording frequency factor and the density coefficient factor; The information intensity characteristics of each case segment are calculated by combining the recording frequency factor and the density factor. The calculation formula is as follows: ,in, To record frequency factors, For the density coefficient factor, and To preset the weighting coefficients, Information intensity characteristics of each case segment; It should be noted that the larger the recording frequency factor and the larger the density coefficient factor, the more information the case segment contains, the higher its value in the shared database, and the stronger the information intensity characteristics of each case segment. Conversely, the smaller the recording frequency factor and the dense coefficient factor, the less information the case segment contains, the lower its value in the shared database, and the weaker the information intensity characteristics of each case segment. The preset weighting coefficient is an important parameter for balancing the influence of the recording frequency factor and the density coefficient factor on the information intensity characteristics of each case segment. It needs to be verified through experimental data and test results in practical applications. It is necessary to evaluate the accuracy of the information intensity characteristics of each case segment under different weighting coefficients by testing different recording frequency factors and density coefficient factors to ensure that the final weighting coefficient can provide the optimal calculation effect.

[0035] The information intensity feature of each case segment is multiplied by the preset maximum matching overlay value to obtain the matching overlay value of each case segment, and the matching overlay value of each case segment is rounded down. The information intensity features of each case segment are compared with a preset intensity feature threshold for judgment. If the information intensity feature of each case segment is greater than the preset intensity feature threshold, then the case segment is determined to be a high information intensity case segment; If the information intensity feature of each case segment is less than or equal to the preset intensity feature threshold, then the case segment is determined to be a case segment with low information intensity. The case fragments were overlaid. The specific stacking process is as follows: High-information-intensity case segments were selected as the main segments; Based on the correlation ranking results, case segments with higher correlation to the main segment are selected and stacked in descending order until the upper limit of the matching stacking value of the case segments corresponding to the main segment is reached. Each stacked sub-fragment matches its own stacking value and is then stacked as a sub-main fragment to the next layer of fragments, forming a hierarchical stacking structure. Each stacking operation follows the matching stacking value constraint to ensure that the number of stacks matches the information intensity, and the number of layers cannot exceed the preset maximum stacking layer limit; It should be noted that overlay refers to the hierarchical organization of case fragments based on information intensity and relevance. The case fragment with high information intensity is the core, and case fragments with high relevance to it are selected and added to form a data combination with a tree diagram structure.

[0036] The stacked case data is stored in a shared database.

[0037] It should be explained that the preset maximum matching stacking value refers to the maximum number of sub-fragments allowed to be stacked in the stacking structure for each case fragment. Experiments were conducted based on the stacking effect or retrieval efficiency of historical databases to observe the impact of different matching stacking values ​​on system performance, and the optimal matching stacking value was selected as the preset maximum matching stacking value. The preset intensity feature threshold is an important parameter for judging the intensity of case fragment information. The impact of different thresholds on the stacking structure effect was verified through experiments or historical databases, and the optimal threshold was selected as the preset intensity feature threshold. The preset maximum stacking layer limit is to prevent the data structure from becoming too complex due to excessively deep layers, which would reduce query and calculation efficiency. Based on historical data analysis, a balance point between query efficiency and data integrity was found at different layer numbers, and an optimal layer number was selected as the preset maximum stacking layer number.

[0038] Finally, it should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0039] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0040] In this document, the singular forms “a,” “an,” and “the” may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that terms such as “comprising / including” or “having” specify the presence of the stated features, integrals, steps, operations, components, parts, or combinations thereof, but do not preclude the possibility of the presence or addition of one or more other features, integrals, steps, operations, components, parts, or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0041] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0042] The above description of the disclosed embodiments will enable those skilled in the art to make or use various modifications to these embodiments. It will be readily apparent to those skilled in the art that the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A sepsis case data sharing system, characterized in that: It includes a data receiving module, a fragment analysis module, a correlation analysis module, and a shared collection module. The functions of each module are as follows: The data receiving module receives case data from different information sources, accesses the shared database to calculate the sharing and inclusion ratio of each information source and generates information evaluation coefficients, compares and analyzes the case data with the case data in the shared database to evaluate the case similarity, combines the information evaluation coefficients to filter and label case data, stores the labeled case data in a temporary database and transmits it to the fragment analysis module. The fragment analysis module detects the entry time of each labeled case data in the temporary database, divides the labeled case data into case fragments according to the entry time, splits the vocabulary of each case fragment, calculates the vocabulary change range, and inputs it into the association analysis module. The association analysis module receives the association characteristics of case fragments by assessing the magnitude of word changes, sorts the association degree of case fragments according to the association characteristics, collects the data recording frequency and duration of case fragments, analyzes the record density coefficient using the duration, and transmits the data recording frequency and record density coefficient to the shared collection module. The shared collection module receives the data record frequency and record density coefficient and generates the information intensity characteristics of the case fragments. Based on the information intensity characteristics, it sets the matching overlay value of the case fragments, and performs overlay processing on the case fragments in combination with the correlation ranking results. The overlay processed case data is then stored in the shared database.

2. The sepsis case data sharing system according to claim 1, characterized in that: In the data receiving module, the system connects to various information source databases through a multi-source data access interface to receive case data from different information sources. Case data refers to a comprehensive collection of data that reflects various medical information and personal information of patients throughout the entire diagnosis and treatment process; Match each news source with a shared database to obtain the number of sepsis cases that have been included in each news source; Count the total number of cases from the same information source in the shared database; The ratio of the number of sepsis cases included in each information source to the total number of cases in the information source is used as the sharing and inclusion ratio of the information source. The total sharing and inclusion ratio is obtained by adding up the sharing and inclusion ratios of each information source; The ratio of the shared inclusion rate of each information source to the total shared inclusion rate is used as the information evaluation coefficient for each information source.

3. The sepsis case data sharing system according to claim 1, characterized in that: In the data receiving module, key fields are extracted from each case data from each information source to obtain a set of case content; The case content set is matched with the case data in the shared database to obtain the case similarity of each case data; The information evaluation coefficients of each information source and the case similarity of each case data are standardized to obtain the information evaluation factor and the similarity factor; The square root of the product of the information evaluation factor and the similarity factor from the same information source is used to obtain the labeling features of each case data under the information source. If the labeling characteristics of the case data are greater than the preset labeling threshold, then the case data is determined to be labeled. If the labeling characteristics of the case data are less than or equal to the preset labeling threshold, then the case data will not be labeled. The tagged case data is stored in a temporary database.

4. The sepsis case data sharing system according to claim 1, characterized in that: In the fragment analysis module, the entry time of each labeled case data in the temporary database is detected, and labeled case data with an entry time interval less than a preset time interval threshold are merged into a case data set. Keyword extraction is performed on each set of case data. The patient ID content is extracted from each labeled case data and traversed through the labeled case data. Case data with identical patient IDs are grouped into separate case segments.

5. A sepsis case data sharing system according to claim 1, characterized in that: In the fragment analysis module, a dictionary-based Chinese word segmenter segments the case descriptions in each case fragment, removing stop words and punctuation marks. Each word in the case description of each labeled case data in each case segment is mapped to a word vector using the one-hot encoding method. The case vector of the labeled case data is obtained by summing the word vectors of the case description content in a case segment; For each case segment, calculate the cosine similarity between the case vectors of all labeled case data; The maximum value minus the minimum value of cosine similarity is used as the lexical change range.

6. A sepsis case data sharing system according to claim 5, characterized in that: In the association analysis module, a case segment is randomly selected, and any case segments other than the case segment are selected and merged into a case segment group; The difference between the magnitude of word changes in two case segments within a case segment group is used as the word change factor for the segment group. The case vector factor of the fragment group is obtained by subtracting the total case vector values ​​of the two case fragments in the case fragment group. The association features of case fragment groups were calculated by combining the lexical change factor and the case vector factor. The association features of case segments with other case segments are compared with preset feature thresholds to determine whether the association features should be used as the degree of association. When the correlation feature is used as the correlation degree, the correlation degree of case segments with other case segments is ranked to obtain the correlation degree ranking result.

7. A sepsis case data sharing system according to claim 1, characterized in that: In the correlation analysis module, a preset statistical time is used to count the number of times case data is recorded within the statistical time of each case segment, and the data recording frequency of each case segment is obtained by dividing the preset statistical time by the number of times the data is recorded. The patient names in each case segment are matched with the databases of each information source to obtain the case data recorded in each case segment; The duration of a case segment is obtained by subtracting the minimum value from the maximum value of the recorded time content in all recorded case data of a case segment. The ratio of the data recording frequency to the duration of each case segment is used as the recording density coefficient of each case segment.

8. A sepsis case data sharing system according to claim 7, characterized in that: In the shared data collection module, the data recording frequency and recording density coefficient of each case segment are standardized to obtain the recording frequency factor and the density coefficient factor; The information intensity characteristics of each case segment are obtained by weighted summation of the recording frequency factor and the density coefficient factor; The information intensity feature of each case segment is multiplied by the preset maximum matching overlay value to obtain the matching overlay value of each case segment, and the matching overlay value of each case segment is rounded down. If the information intensity feature of each case segment is greater than the preset intensity feature threshold, then the case segment is determined to be a high information intensity case segment; If the information intensity feature of each case segment is less than or equal to the preset intensity feature threshold, then the case segment is determined to be a case segment with low information intensity.

9. A sepsis case data sharing system according to claim 8, characterized in that: In the shared collection module, case fragments with high information intensity are selected as the main fragments; Based on the correlation ranking results, case fragments with higher correlation to the main fragment are selected and stacked in descending order until the upper limit of the matching stacking value of the case fragments corresponding to the main fragment is reached. Each stacked sub-fragment matches its own stacking value and is then stacked as a sub-main fragment to the next layer of fragments, forming a hierarchical stacking structure. The stacked case data is stored in a shared database.

Images