Traditional Chinese medicine large variety project information management method and system based on multi-source data

Abstract

The application discloses a traditional Chinese medicine large variety project information management method and system based on multi-source data, and relates to the technical field of multi-source project data management. The method obtains project execution data of a traditional Chinese medicine large variety project in project establishment, drug management, subjects, visits, quality control and statistical analysis, carries out standardized mapping processing on different source data, establishes corresponding mapping relationships, and forms a standard transaction data set; the standard transaction data set is subjected to time sequence compensation and reconstruction, candidate time sequence segments are generated, the candidate time sequence segments satisfying preset connection conditions are subjected to chain merging, and a continuous data chain for the same project transaction is constructed; finally, based on the continuous data chain, state loss and state jump abnormality in the project advancement process are identified, and whether to output final dynamic management information is judged. The continuous correlation integration of multi-source project data is realized, and the identification accuracy of the real advancement state and abnormal node condition of the traditional Chinese medicine large variety project is improved.

Description

Technical Field

[0001] This invention relates to the field of multi-source project data management technology, and in particular to a method and system for managing information on major Chinese medicine products based on multi-source data. Background Technology

[0002] The advancement of major traditional Chinese medicine (TCM) projects typically involves multiple stages, including project initiation, protocol design, center selection, subject management, clinical trial implementation, data collection, statistical analysis, and registration application support. Especially for TCM projects focused on clinical research and evaluation, the implementation process continuously generates heterogeneous data from multiple sources, including basic project data, trial protocol data, research center execution data, subject enrollment data, visit data, drug management data, data quality control data, and statistical evaluation data.

[0003] In existing technologies, a project management system for major Chinese medicine products is typically set up. This system includes a project management subsystem, a clinical trial management subsystem, an electronic data acquisition subsystem, a drug management subsystem, a document management subsystem, and a statistical analysis subsystem. Among them, the project management subsystem is used to maintain project numbers, plan nodes, and center progress information; the clinical trial management subsystem is used to maintain research center and subject execution information; the electronic data acquisition subsystem is used to maintain case report forms and visit data; the drug management subsystem is used to maintain drug dispensing and collection data; the document management subsystem is used to maintain ethics documents, research materials, and quality control documents; and the statistical analysis subsystem is used to maintain statistical results and evaluation results.

[0004] Taking the subject enrollment and visit execution phase of a major clinical research project for traditional Chinese medicine as an example, when subject S001 completes the screening, formal enrollment, first visit, and drug dispensing sequentially on the same day, corresponding data records will be generated in different subsystems. Specifically, the clinical trial management subsystem generates a record of enrollment status changes, the electronic data collection subsystem generates a set of visit form data, the drug management subsystem generates a record of the combination of drug batch number and subject number, and the document management subsystem subsequently generates monitoring records or quality control documents related to this execution process.

[0005] However, existing technologies for summarizing data from multiple subsystems typically use interface calls, report exports, or database extraction to combine data from different sources based on fields such as project number, subject number, or visit name. This approach essentially merges the outputs of each subsystem at the field level. While it achieves data aggregation and display, the differences in organization, record granularity, generation methods, and time representation among data from different sources mean that the aggregation process often relies on superficial matching based on identifier fields. This makes it difficult to identify the relationships between status change records, form data sets, drug combination records, and attachment objects corresponding to the same execution process.

[0006] Furthermore, when data from different subsystems differ in recording time precision, generation time, and backfilling rhythm—for example, some data is written in real time, while others are generated after subsequent data entry, or some data is recorded to the minute-level timestamp while others only record date information—data originally belonging to the same project item is prone to exhibiting partial missing data or correlation drift after integration. Therefore, while existing technologies can achieve aggregated management of data from multiple subsystems, they struggle to accurately identify the true progress status and abnormal nodes of major traditional Chinese medicine (TCM) projects based on data scattered across different subsystems. Ultimately, this limits the TCM project management system's ability to manage dynamic information throughout the entire project process. Summary of the Invention

[0007] Therefore, embodiments of the present invention provide a method and system for managing information on major traditional Chinese medicine products based on multi-source data. The technical solution is as follows: On the one hand, a method for managing information on major traditional Chinese medicine products based on multi-source data is provided, which includes: Obtain project execution data corresponding to major Chinese medicine projects, perform standardized mapping processing, establish a one-to-one or one-to-many relationship between project items and business nodes, and obtain a standard item dataset. The project execution data should include at least project initiation data, drug management data, subject data, visit data, quality control data, and statistical analysis data.

[0008] Perform time-series compensation and reconstruction on the standard event dataset to generate candidate time-series segments.

[0009] Within the preset project item association range, when the business node sequence corresponding to the candidate time sequence segment meets the preset connection conditions, chain merging is performed to construct a continuous data chain for the same project item; otherwise, a break mark is made and the data is retained for verification. The preset project item association range includes at least one or more of the following: the same project identifier, the same research center identifier, the same subject identifier, the same visit identifier, and the same item type identifier.

[0010] Based on continuous data links, identify missing states and abnormal state transitions during the progress of the same project item, in order to determine whether to output dynamic management information for the whole process of progress identification of major Chinese medicine projects.

[0011] On the other hand, a multi-source data-based information management system for major traditional Chinese medicine products is provided. This system includes: The standard mapping module is used to obtain the project execution data corresponding to major Chinese medicine projects, perform standardized mapping processing, establish a one-to-one or one-to-many relationship between project items and business nodes, and obtain a standard item dataset.

[0012] The temporal reconstruction module is used to perform temporal compensation and reconstruction on standard event datasets to generate candidate temporal segments.

[0013] The link building module is used to perform chain merging within the preset project item association range. When the business node sequence corresponding to the candidate time sequence segment meets the preset connection conditions, it will build a continuous data chain for the same project item. Otherwise, it will mark the break and retain it for verification.

[0014] The status recognition module is used to identify status gaps and status transition anomalies in the same project process based on continuous data chains, so as to determine whether to output dynamic management information for the whole process progress recognition of major Chinese medicine projects.

[0015] The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following: 1. This invention first acquires project execution data generated during the stages of project initiation, drug management, subject management, visit execution, data quality control, and statistical analysis for major traditional Chinese medicine (TCM) projects. Through standardized mapping, it establishes a one-to-one or one-to-many mapping relationship between project items and business nodes, resulting in a standardized item dataset. This unifies subject enrollment records, visit records, drug records, quality control records, and statistical records, originally scattered across different subsystems, under a single project item framework, improving the ability to merge and correlate multi-source heterogeneous data in the clinical research scenario of major TCM products. Secondly, it performs temporal compensation and reconstruction on the standardized item dataset to generate candidate temporal segments, thereby correcting the sequential misalignment caused by supplementary data entry, delayed data writing, and data with different time precisions, restoring the true progress order of the same subject, the same visit, or the same item. Furthermore, within the associated scope of the same project identifier, same research center identifier, same subject identifier, same visit identifier, and same event type identifier, candidate time-series segments that meet the preset connection conditions are chained together to construct a continuous data chain. This comprehensively represents the connection link between subject screening, enrollment, visit execution, data quality control, and statistical analysis in major traditional Chinese medicine projects. Finally, based on the continuous data chain, missing status and abnormal status transitions are identified, and dynamic management information for the entire process progress is output, thereby intuitively improving the ability to identify the true progress of major traditional Chinese medicine projects, locate abnormal nodes, and manage the entire process dynamically.

[0016] 2. First, extract the node name, status, time, and object identifier fields from the project execution data, and obtain the semantic parameters and temporal attribution parameters for each. Then, combine these parameters to obtain a comprehensive mapping parameter, which simultaneously depicts the correspondence between data records in terms of item category, status meaning, process stage, and object attribution, improving the mapping accuracy between project data from different sources and preset project items. Subsequently, based on the comprehensive mapping parameter, construct a mapping sequence between data records and project items, uniquely mapping each piece of project execution data to the highest matching project item, forming a one-to-one mapping relationship. Then, merge the project execution data to obtain a standardized merged item dataset, thus achieving standardized aggregation of scattered data. Further, when multiple business node data mapped to the same project item are identified, establish a one-to-many association mapping relationship based on the state change sequence and temporal connection of the multiple business node data. Then, perform item aggregation processing on the multiple business node data belonging to the same project item to obtain an aggregated and associated item dataset, thereby enhancing the complete expressive capability of the same project item in multi-node, multi-state scenarios.

[0017] 3. First, extract the business occurrence time, status update time, and supplementary entry time corresponding to each item data in the standardized and merged item dataset or the aggregated and related item dataset, and perform identification and grouping processing to generate an initial time series, thereby unifying the originally scattered item data under the same project item time perspective. Subsequently, calculate time compensation parameters based on the supplementary entry delay value and update time offset value, and construct a time compensation matrix in combination with time correlation relationships to uniformly correct the time position of each item data, obtaining a compensated time series, thereby effectively correcting the time sequence deviation caused by supplementary entry, delayed updates, and differences in time precision. Further, based on the compensated time series, combined with node order relationship parameters and status connection parameters, rearrange and chain the order of each item data corresponding to the same project item, generate candidate time sequence segments, and distinguish continuous segments from independent segments according to preset connection conditions, thereby improving the time sequence restoration capability of the local execution process of project items. Finally, the candidate time series segments are sorted according to their start and end times, and adjacent candidate time series segments are progressively chained and merged based on the segment connection parameters until a continuous data chain for the same project is formed, thereby enhancing the ability to fully represent the real progress chain of major Chinese medicine projects.

[0018] 4. First, the screening node, enrollment node, node occurrence time, and subject identification information corresponding to the same subject are extracted from the continuous data chain. Node association parameters are obtained by combining sequential association values, temporal association values, and attribution association values. This comprehensively characterizes the degree of continuous association between screening and enrollment nodes in terms of sequence, temporal connection, and subject attribution, improving the ability to identify the completeness of the pre-enrollment process. Furthermore, the node association parameters determine whether there are missing screening nodes, enrollment nodes appearing before screening is completed, or abnormal node temporal connections, thus accurately identifying missing states and generating corresponding markers. On the other hand, information such as the enrollment status, visit preparation status, visit execution status, status location number, and status update time corresponding to the same subject are also extracted from the continuous data chain. Status transition parameters are determined by status scrambling parameters and intermediate status incompleteness parameters, effectively characterizing the degree of missing intermediate execution steps and status scrambling during the status progression process, improving the accuracy of identifying abnormal progression behaviors from subject enrollment to visit execution. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1A flowchart illustrating the method for managing information on major Chinese medicine products based on multi-source data provided in this embodiment of the invention; Figure 2 This is a schematic diagram of the structure of the information management system for major Chinese medicine products based on multi-source data provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the abnormal progress of major traditional Chinese medicine projects provided in this embodiment of the invention; Figure 4 A schematic diagram of standardized mapping relationships provided for embodiments of the present invention; Figure 5 A schematic diagram of time compensation and sequence correction provided for an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the generation of candidate time segments provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0022] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0023] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0024] The embodiments of the present invention provide, as follows Figure 1 The flowchart shown is a method for managing information on major Chinese medicine projects based on multi-source data. By standardizing and mapping, time-series compensation and reconstruction, constructing continuous data chains, and identifying abnormal states, the method enables dynamic identification and management of the progress status of the entire project.

[0025] Embodiment 1 of the present invention Step 1: The project execution data for major traditional Chinese medicine (TCM) projects is not a single-format data set, but rather a collection of multiple data types from different business subsystems. Project initiation data can be tabular data such as project application forms, project plans, project budgets, and project schedules, or text data such as project descriptions, research protocol summaries, and project approval documents. Drug management data can be structured tabular data such as drug warehousing records, dispensing records, return records, and inventory records. Subject data can be structured tabular data such as subject screening forms, enrollment registration forms, withdrawal records, and adverse event records, or semi-structured text data such as informed consent texts and screening instructions. Visit data can be visit plans, visit execution forms, examination and test result forms, case report forms, and visit notes. Quality control data can be tabular and text data such as monitoring records, inquiry forms, verification records, deviation records, and audit opinion forms. Statistical analysis data can be statistical analysis plans, dataset version information, statistical result tables, charts, and summary texts.

[0026] After obtaining the aforementioned project execution data, a project item set and a business node set are first established in advance. The project item set is used to define the item categories that need to be uniformly identified. It can be pre-configured by technical personnel according to the project management process for major Chinese medicine products. For example, project initiation items, drug distribution items, subject screening items, subject enrollment items, visit execution items, data quality control items, and statistical analysis items can be set as different project items. The business node set is used to define the node categories of each item in the execution process. For example, pending submission, submitted, reviewed, screening, enrolled, visit in progress, and locked inventory can be set as different business nodes.

[0027] Subsequently, the node name, status, time, and object identifier fields are extracted from each project execution data. The node name field can be derived from table column names, process node names, form signatures, or text titles; the status field can be derived from status columns, review result columns, execution result columns, or status description statements in the text; the time field can be derived from creation time, submission time, update time, visit time, medication dispensing time, quality control time, etc.; the object identifier field can be derived from project number, protocol number, subject number, visit number, drug batch number, form number, etc. For tabular data, the field names are read directly; for text data, the corresponding fields are extracted from paragraph headings, fixed tags, and sentence structures using preset keyword extraction rules, such as extracting corresponding content from fixed expressions like project number, subject number, and visit time.

[0028] Based on this, the node name and status content in the current data record are matched with the item name, item description, and corresponding typical status description in the preset project item set, respectively, to obtain the corresponding results of the current data record in terms of item category and status meaning. Specifically: first, the name matching value A is calculated, then the status matching value B is calculated, and then the item semantic parameter C is obtained through addition, where C = A + B. Both the name matching value A and the status matching value B are obtained by the ratio of the number of overlapping terms to the total number of terms. The number of overlapping terms refers to the number of terms that are the same as those in the target item's corresponding field after the current field is split, and the total number of terms refers to the total number of terms contained in the current field after splitting.

[0029] For example, if a data record's node name is "Enrollment Registration" and its status is "Enrollment Completed," then its name matching value A with the subject enrollment item can be 1, and its status matching value B can be 1. Therefore, the semantic parameter C = 1 + 1 = 2. If this data record is matched with the medication dispensing item, then the name matching value A can be 0, and the status matching value B can be 0. Therefore, the semantic parameter C = 0. Similarly, if a data record's node name is "Medication Dispensing Confirmation" and its status is "Dispensed," then its name matching value A with the medication dispensing item can be 1, and its status matching value B can be 1. Therefore, the semantic parameter C = 2. This method allows for a clear numerical reflection of the semantic proximity between the current data record and different item matters.

[0030] Simultaneously, the data record is first identified by the object identifier field to determine its project, subject, visit, or drug batch type; then, its occurrence order within the same object's execution process is determined by the time field. Specifically: first, the object attribution value D is determined, then the stage sequence value E is determined, and finally, the time sequence attribution parameter F is obtained through addition, where F = D + E. The object attribution value D indicates whether the current data record clearly belongs to the same object, and can be set to 1 if the project number, subject number, visit number, or drug batch number all match, and 0 if they do not match; the stage sequence value E indicates the stage position of the current data record in the execution chain, which can be obtained by predefined sequence numbers. For example, in the subject execution chain, screening and registration is 1, screening qualified is 2, waiting to be enrolled is 3, enrolled is 4, visit preparation is 5, visit execution is 6, and data quality control is 7. For example, if a record has subject number S001, visit number V02, date March 10, 2025, and node name "Enrolled," then the subject attribution value D is 1, the stage sequence value E is 4, and therefore the time sequence attribution parameter F = 1 + 4 = 5. Another record, with the same project number, subject number S001, earlier date, and node name "screening and registration," would have a subject attribution value D of 1 and a stage sequence value E of 1, thus the time sequence attribution parameter F = 1 + 1 = 2. This allows for a clear numerical representation of which subject the data record belongs to and which process stage it is in.

[0031] Then, the semantic parameters of the event and the time sequence attribution parameters are combined to obtain the comprehensive mapping parameter. Specifically, let the comprehensive mapping parameter G = (C × F) / K, where K is the normalization coefficient, which can be the maximum stage number in the current project event chain plus 1, or a fixed constant, such as 10. For example, for a data record of enrollment registration / completed enrollment, if the event semantic parameter C = 2, the time sequence attribution parameter F = 5, and K = 10, then the comprehensive mapping parameter G = (2 × 5) / 10 = 1; if this data is matched with the drug dispensing event, its event semantic parameter C = 0, and the time sequence attribution parameter F can still be 5, then the comprehensive mapping parameter G = (0 × 5) / 10 = 0. For each project execution data, the comprehensive mapping parameter G between it and each preset project event is calculated, and these values ​​are arranged in the order of project events to form a mapping sequence.

[0032] Subsequently, the item with the largest comprehensive mapping parameter in the mapping sequence is selected as the highest matching item for that item's execution data, thus establishing a one-to-one mapping relationship between that item's execution data and the highest matching item. For example, if a data point has a comprehensive mapping parameter of 1 for subject enrollment, 0.2 for drug dispensing, and 0 for data quality control, then subject enrollment is determined as the unique corresponding item for that data point.

[0033] After completing the one-to-one mapping, data mapped to the same project item are merged. The merging method can be done by grouping, sorting, and then merging: first, group by project number, subject number, visit number, and project item name; then sort by time field from smallest to largest; finally, concatenate the data records within the same group in chronological order to form a single item data set. For example, if three data records all map to a subject's enrollment item, and have the same project number and subject number, these three records are first grouped together, then sorted in ascending chronological order, ultimately forming the subject's enrollment item data set. This merging method yields a standard merged item dataset.

[0034] After completing the one-to-one mapping, if multiple business node data are identified as being mapped to the same project item, a one-to-many association mapping relationship is further established. For example... Figure 4The diagram illustrates a standardized mapping relationship. The left side shows a one-to-one correspondence between project execution data and project items; each piece of project execution data corresponds to only one highest-matching project item after being determined by comprehensive mapping parameters. The right side shows a one-to-many association mapping between the same project item and multiple business nodes during subsequent item aggregation; that is, the same project item can be further associated with multiple preceding and following business node data. Based on this mapping method, while clearly defining that each piece of original project execution data uniquely belongs to a specific project item, it also allows for the preservation of the multi-node development trajectory of the project item during its actual implementation.

[0035] Specifically, the establishment of a one-to-many association mapping relationship can employ explicit numerical and computational methods. First, pre-assign node sequence numbers H to different business nodes under the same project item. Then, calculate the time interval value I between adjacent business nodes. Finally, calculate the node association value J, where J = H difference + I judgment value. Here, the H difference represents the result of subtracting the previous node number from the subsequent node number; the I judgment value can be set as follows: 1 when the time interval between two adjacent nodes is within a preset connection interval, and 0 when it is not within the preset interval. If the H difference = 1 and the I judgment value = 1, it indicates that the two business nodes are sequentially continuous and reasonably connected in time, and these two nodes can be established as a one-to-many association node under the same project item.

[0036] For example, subject enrollment can correspond to multiple business nodes such as pending screening, qualified screening, pending enrollment, and enrolled, and assigned values ​​of 1, 2, 3, and 4 respectively. If a subject's data record is sequentially pending screening, qualified screening, pending enrollment, and enrolled, then the difference between adjacent node numbers is 1, and the adjacent time intervals all fall within a preset range. These nodes can then be collectively associated with the same subject enrollment event. Similarly, data quality control events can correspond to multiple business nodes such as responded to and closed. If their node numbers are consecutive and the time sequence is reasonable, a one-to-many association mapping relationship can also be established.

[0037] After establishing a one-to-many association mapping, data from multiple business nodes belonging to the same project item can be aggregated based on project number, subject number, project item name, node sequence number, and time order. The aggregation process can be performed by grouping, sorting, and then concatenating. First, business node data with the same project number, subject number, and project item name are grouped together. Then, they are sorted by node sequence number from smallest to largest. If node sequence numbers are the same, they are arranged by time order from earliest to latest. Finally, the sorted business node data are concatenated sequentially to form multi-node trajectory data under the same project item. This method yields an aggregated and associated item dataset. This approach clearly identifies which project item each piece of original data belongs to while preserving the multi-node progression trajectory within the same project item, facilitating subsequent time-series compensation and reconstruction.

[0038] Step Two: There are two entry paths for the data foundation faced by time-series compensation and reconstruction: one is when a project item corresponds to only a single business node, or although it comes from multiple original records, it can independently express the item after the aforementioned one-to-one mapping. In this case, the standard merged item dataset can be directly used as the time-series processing object. The other is when a project item corresponds to multiple sequential business nodes in the actual execution process. For example, the subject enrollment item may simultaneously include nodes such as pending screening, qualified screening, pending enrollment, and enrolled, or the data quality control item may include nodes such as responded and closed. In this case, it is difficult to fully reflect the progress of the item by using only the standard merged single-point record. It is necessary to further call the aggregated and associated item dataset as the time-series processing object.

[0039] After determining the source of the input data, the first step is to extract the business occurrence time, status update time, and supplementary entry time corresponding to each item's data, and then perform identification and grouping processing to generate an initial time series. Here, the business occurrence time refers to the time when the corresponding business action actually occurs during execution, such as the enrollment completion time, medication dispensing completion time, visit execution time, and quality control response time; the status update time refers to the time when the status of the item's data is modified or refreshed in the business system; and the supplementary entry time refers to the data writing time formed after the original business action occurs due to delayed entry, manual supplementation, or cross-system synchronization lag. After extraction, the data is then identified and grouped according to project number, subject number, visit number, drug batch number, and project item name, that is, data with the same object belonging to the same category are grouped into the same data group. Within each data group, the data is then initially sorted according to the business occurrence time, status update time, and supplementary entry time in sequence to generate the corresponding initial time series. In other words, the initial time series is essentially the original time arrangement result of the same project item under the same project object.

[0040] After obtaining the initial time series, the supplementary entry delay value and update time offset value are further calculated to generate time compensation parameters. The supplementary entry delay value is obtained by subtracting the business occurrence time from the supplementary entry time, and is used to characterize the delay between the actual occurrence of the data and its entry into the system. The update time offset value is obtained by subtracting the business occurrence time from the status update time, and is used to characterize the degree of deviation of the status change record relative to the original business action. Then, the supplementary entry delay value and the update time offset value are added to obtain the time compensation parameter: Time compensation parameter = Supplementary entry delay value + Update time offset value. The larger the time compensation parameter, the more significant the deviation in the completeness of the time record and the accuracy of the time position of the data.

[0041] After obtaining the time compensation parameters, a time compensation matrix is ​​further constructed. The time compensation matrix can be understood as a time correction relationship table established for each item's data within the same data group. Each element in the matrix represents the time position of a certain item's data after correction by the time compensation parameters. Specifically, the original business occurrence time of each item's data can be used as the base time, and the corresponding time compensation parameter can be subtracted or added to the base time to obtain the corrected time position. If the data is a supplementary delayed record, the correction time is usually calculated as: Correction Time = Business Occurrence Time + Time Compensation Parameter Correction Amount. If the data has a status update time shift deviation, the correction time can be calculated as: Correction Time = Status Update Time - Time Compensation Parameter Correction Amount. After writing the correction times of each item's data within the same data group into the matrix in sequence, the time compensation matrix is ​​formed. Then, by re-sorting the correction times of each item's data in this matrix, the compensation time sequence can be obtained, i.e., sorted according to the chronological order of the correction times of each item's data from earliest to latest; when the correction times are the same, they are then sorted according to the business node sequence number from smallest to largest. The compensation time series reflects the order of events in the same project after eliminating delays in data entry, offsets in status updates, and cross-system write errors, which more closely approximate the actual execution process.

[0042] After obtaining the compensated time series, the node sequence relationship parameter and state connection parameter are further combined to rearrange and chain the data of each item corresponding to the same project item, generating candidate time series segments. The node sequence relationship parameter is used to characterize the sequential relationship between two adjacent item data in the business process. It can be calculated based on the difference in node sequence numbers. For example, when the difference between the next node number and the previous node number equals 1, it means that the two nodes are continuous in the business process. The state connection parameter is used to characterize whether the state changes of two adjacent item data have a continuous evolution relationship. It can be determined based on whether there is a natural connection between the previous state and the next state. For example, there is a connection between "pending group entry" and "already entered group entry," while there is usually no connection between "pending screening" and "directly proceeding to visit execution." The data in the compensated time series are compared sequentially according to the corrected time order. If two adjacent data are both true in terms of node sequence relationship and state connection relationship, they are concatenated into the same candidate time series segment; if they are not true, they are truncated at that point, and a new candidate time series segment is started.

[0043] Furthermore, to determine whether adjacent data items can be concatenated into the same candidate time series segment, it is necessary to assess whether the connection conditions are met. These connection conditions are not abstract judgments, but rather specific analytical results based on the parameters mentioned earlier. First, check whether the business nodes corresponding to the two adjacent data items conform to a sequential relationship, i.e., whether the latter node follows the former node. Second, check whether the state change directions of the two adjacent data items are consistent, i.e., whether they naturally progress from the former state to the latter. Finally, check whether the time interval between the two falls within the allowable range. This allowable range can be obtained based on the historical execution time statistics of similar items. For example, first, statistically analyze the common time spans between similar business nodes, and then select the average fluctuation range above and below as the time tolerance range. When the business node sequence is valid, the state transition direction is consistent, and the time interval is within the aforementioned tolerance range, the adjacent data items can be considered to meet the connection conditions, and the two can be connected into the same candidate time series segment. If any condition is not met, the current data item is retained as an independent candidate time series segment, awaiting subsequent verification or participating in the next step of processing separately.

[0044] Through the above process, the data of each item in the initial time series are gradually corrected, rearranged, and pieced together into one or more candidate time series segments. The result of this processing is that it can not only eliminate the interference caused by the deviation of recording granularity, supplementary recording rhythm, and update time between different systems in the time dimension, but also restore the true progress order between the nodes within the same project item in the business process dimension.

[0045] Step 3: First, extract the start time, end time, start node, and end node of each candidate time series segment from all candidate time series segments formed under the same project item. Then, sort them according to the start time from earliest to latest. When multiple candidate time series segments have the same start time, sort them again according to the end time from earliest to latest to generate a data chain segment sequence to be merged.

[0046] After sorting, it is necessary to calculate the segment connection parameter for adjacent candidate time series segments in the data chain segment sequence. The segment connection parameter can be determined by both node connection relationship and time interval relationship. The node connection relationship is used to characterize whether there is a process continuity between the termination node of the previous candidate time series segment and the starting node of the next candidate time series segment. For example, if the previous segment terminates at the screening qualification and the next segment starts at the waiting group, then the two have the possibility of connection in the business chain; if the previous segment terminates at the waiting screening and the next segment starts directly at the visit execution, then there is usually no normal connection relationship. The time interval relationship is used to characterize whether the interval between the termination time of the previous segment and the start time of the next segment is reasonable, and it can be calculated by the time difference between the two. Specifically, the node connection value is 1 when the node connection is established and 0 when it is not established; the time interval value is 1 when the time difference is within the allowable range, and 0 otherwise; then the segment connection parameter is obtained according to the formula: segment connection parameter = node connection value × time interval value. The larger the segment connection parameter, the more likely that the adjacent candidate time series segments belong to the same continuous execution process.

[0047] After obtaining the segment connection parameters, it is necessary to determine whether they meet the merging conditions. These merging conditions can be reflected by whether the segment connection parameters reach a set value. For example, when using the multiplication method, the segment connection parameter is only 1 if both the node connection value and the time interval value are 1, indicating that the two candidate time series segments can be merged. If the segment connection parameter is 0, it indicates at least a process break or time break, and merging is not performed. When the segment connection parameters corresponding to adjacent candidate time series segments in the data link segment sequence meet the merging conditions, the termination node of the previous candidate time series segment is connected end-to-end with the starting node of the next candidate time series segment to form a new merged segment. The start time of this merged segment is taken from the start time of the previous candidate time series segment, and the end time is taken from the end time of the next candidate time series segment. The node trajectory within the segment is formed by sequentially splicing the nodes from the two segments. If the segment connection parameters do not meet the merging conditions, the independent segment state of the current candidate time series segment is retained, and a break marker is added to the end of the current segment, indicating that there is a business link interruption or time connection anomaly at that point. The merging conditions can be jointly determined based on the connection relationship between adjacent business nodes in historical continuous samples, the distribution of time intervals, and the statistical boundaries of segment connection parameters.

[0048] After a merged segment is formed, the process doesn't end immediately. Instead, the merged segment is treated as a new candidate time series segment and continues to be evaluated for connection and chained merging with subsequent candidate time series segments. In other words, chained merging uses a progressive processing method: first, it determines whether the first two segments can be merged. If they can, the merged new segment is compared with the third segment. If the third segment also meets the merging criteria, the connection continues; if not, the current merged segment is fixed, and subsequent segments are processed separately. In this way, multiple scattered local candidate time series segments can be gradually expanded into longer and more complete segment trajectories. Until no adjacent candidate time series segments satisfy the merging criteria remain in the sequence, the currently formed merged segment is determined as a continuous data chain for the same project.

[0049] Through the aforementioned chain-like merging process, not only can the complete progress trajectory of the same project across different business nodes be restored based on chronological order, but also fragmented sections that were separated due to delays in data entry, system switching, or differences in status update rhythms can be reassembled into a continuous chain. The resulting continuous data chain preserves the sequential relationship between nodes at each stage within the project while clearly identifying potential breakpoints in the chain.

[0050] Step Four: After completing the construction of the continuous data chain, it is necessary to further identify state gaps and state transition anomalies in the same project's progress based on the continuous data chain, and determine whether to output dynamic management information for the entire process of major traditional Chinese medicine projects. This dynamic management information is not arbitrarily triggered, but rather a management result generated based on a comprehensive judgment of the anomaly identification results, the degree of anomaly, and the scope of impact. Specifically, state gap identification and state transition anomaly identification can be completed separately first, then the anomaly judgment results can be determined based on the two types of identification results, and finally, the decision on whether to output dynamic management information can be made based on the anomaly judgment results.

[0051] The missing state identification process includes the following steps: First, extract the screening node, enrollment node, node occurrence time, and subject identification information corresponding to the same subject from the continuous data chain. The subject identification information includes at least the project number, subject number, and visit number to ensure that the comparison subjects belong to the same subject execution chain. Then, determine whether a screening node exists in the continuous data chain and whether it precedes the enrollment node. If a screening node exists and its time position is earlier than the enrollment node, assign a sequence association value; if a screening node is missing, or its time position is later than the enrollment node, assign an abnormal sequence value. The sequence association value can be 1, and the abnormal sequence value can be 0. Next, calculate the time interval between the screening node and the enrollment node, which can be obtained by subtracting the screening node occurrence time from the enrollment node occurrence time. To determine whether this time interval is reasonable, a time tolerance range also needs to be obtained.

[0052] The time tolerance interval can be obtained as follows: First, statistically analyze the time interval samples between the screening node and the enrollment node in historical projects. Then, calculate the sample mean and sample fluctuation range, and determine the fluctuation range of the mean as the time tolerance interval. If the current time interval falls within this interval, assign a time correlation value of 1; otherwise, assign a time correlation value of 0. Next, determine whether the subject identifiers corresponding to the screening node and the enrollment node are consistent. If they are consistent, assign an attribution correlation value of 1; otherwise, assign an attribution correlation value of 0. Subsequently, add the sequential correlation value, the time correlation value, and the attribution correlation value to obtain the node correlation parameter. The node correlation parameter can be written as: Node Correlation Parameter = Sequential Correlation Value + Time Correlation Value + Attribution Correlation Value. Thus, the value range of the node correlation parameter is 0 to 3.

[0053] If the node association parameter is less than a predetermined missing value threshold, for example, less than 2, it indicates that there is a missing screening node, an entry node appears before the screening node is completed, or there is an abnormal time connection between the screening node and the entry node in the current continuous data chain. In this case, it is determined that there is a missing state in the continuous data chain, and a missing state marker is generated. If the node association parameter is greater than or equal to the missing value threshold, it is determined that the connection state between the screening node and the entry node in the continuous data chain is normal, and a non-missing state marker is generated. The missing value threshold is represented by the sum and average of the historical node association parameters corresponding to the historical screening to entry connection samples in historical projects.

[0054] On the other hand, the identification of state transition anomalies includes the following process: First, extract the enrollment state, visit preparation state, and visit execution state corresponding to the same subject from the continuous data chain, as well as the state position number, state update time, and subject identification information corresponding to each state. The state position number can be predefined, for example, the enrollment state number is 1, the visit preparation state number is 2, and the visit execution state number is 3. Then, calculate the state transition parameter based on the state position number. The state transition parameter can be obtained by subtracting the previous state position number from the subsequent state position number and then subtracting 1. If the enrollment state number is 1 and the visit execution state number is 3, then the state transition parameter = 3 - 1 - 1 = 1, indicating that an intermediate state has been transitioned. Next, determine whether the visit preparation state exists in the continuous data chain. If the visit preparation state is missing, assign an intermediate state missing value of 1; if it exists, assign an intermediate state missing value of 0.

[0055] Then, the update time interval between the group entry status and the visit execution status is calculated, and a time anomaly value is assigned based on whether this interval exceeds the normal progress range: 1 if it exceeds, and 0 if it does not. The time progress range can be obtained in the same way as described above, i.e., by statistically analyzing the update time interval samples between the group entry status and the visit execution status in historical projects to obtain the normal fluctuation range. Subsequently, the missing intermediate status value and the time anomaly value are added together to obtain the intermediate status incomplete parameter. Then, the status transition parameter is added together with the intermediate status incomplete parameter to obtain the status transition parameter. If the status transition parameter exceeds the preset anomaly threshold, for example, greater than or equal to 2, a status transition anomaly is determined to exist in the continuous data chain, and a transition anomaly mark is generated; if the status transition parameter does not reach the preset anomaly threshold, the status progress between the group entry status and the visit execution status is determined to be normal, and a non-transition status mark is generated. The preset anomaly threshold is represented by the sum and average of the historical status transition parameters corresponding to the progress samples in historical projects.

[0056] After obtaining the missing state marker, non-missing state marker, transition anomaly marker, and non-transition state marker, it is necessary to further determine whether to output dynamic management information for the entire process of progress identification of major traditional Chinese medicine projects. The specific implementation method is as follows: First, define a comprehensive anomaly judgment parameter to uniformly characterize the anomaly situation in the current continuous data chain. The comprehensive anomaly judgment parameter can be obtained by adding the missing state marker value and the transition anomaly marker value, where a missing state marker is 1 if it exists and 0 if it does not exist; a transition anomaly marker is 1 if it exists and 0 if it does not exist. The calculation formula for the comprehensive anomaly judgment parameter can be written as: Comprehensive anomaly judgment parameter = Missing state marker value + Transition anomaly marker value.

[0057] If the comprehensive anomaly judgment parameter is equal to 0, it means that there is no obvious anomaly in the current continuous data chain. In this case, no abnormal dynamic management information is triggered, and only normal progress prompt information can be output. If the comprehensive anomaly judgment parameter is equal to 1, it means that there is a single anomaly. In this case, local anomaly prompt information is output, such as the absence of a pre-enrollment step for the subject or an abnormal progress in the subject visit status. If the comprehensive anomaly judgment parameter is equal to 2, it means that there are both status missing and status jump anomalies in the continuous data chain. In this case, anomaly warning level dynamic management information is output, and the current project item is marked as a key verification item.

[0058] To enhance the operability of dynamic management information, the output can be further supplemented with information on the location, type, object, and time of any anomalies. Specifically, the project number, subject number, visit number, anomaly node name, anomaly occurrence time, and anomaly determination result can be written into the dynamic management information record and pushed to the project management interface, quality control review interface, or statistical analysis interface in chronological order. Project managers can not only identify whether there are any anomalies in the overall progress identification of major traditional Chinese medicine projects, but also directly pinpoint the missing or abrupt status changes at the corresponding subject, visit stage, or business node, enabling timely verification, supplementation, correction, or early warning. Through this process, the output of dynamic management information is no longer a simple data display, but a proactive project progress management result built on the foundation of continuous data chain anomaly identification.

[0059] Embodiment 1 of this invention describes the process for identifying missing states in the context of subject screening and subject enrollment, and the process for identifying abnormal state transitions in the context of subject enrollment and visit execution. However, it does not address the link drift that may occur during cross-node connection, cross-time supplementation, or cross-system connection of data from different business nodes. Therefore, Embodiment 2 is proposed to describe the link drift identification process in the context of visit execution and data quality control.

[0060] Embodiment 2 of the present invention First, extract the visit execution node, case data recording node, and data quality control node corresponding to the same subject from the continuous data chain, along with their corresponding object identification information, node occurrence time, status update time, and supplementary entry time. Then, perform a consistency comparison of the data at each node according to the subject number, visit number, and project number, i.e., determine whether the subject number, visit number, and project number corresponding to the visit execution node, case data recording node, and data quality control node are consistent. When all three are consistent, it is determined that the data at each node belongs to the same subject in the same project execution chain under the same visit stage; otherwise, it is determined that there is an inconsistency in object attribution. Subsequently, check whether there is a corresponding case data recording node after the visit execution node, and whether there is a corresponding data quality control node after the case data recording node. If there is an inconsistency in object attribution, a misalignment of node order, or a missing adjacent node, it is determined that there is a link drift in the continuous data chain, and a corresponding link drift marker is generated; otherwise, it is determined that the link association between the visit execution node and the data quality control node is normal, and a corresponding non-drift marker is generated.

[0061] In summary, after completing the state missing identification and state transition anomaly identification in Example 1 and the link drift identification in Example 2, the missing state marker value, transition anomaly marker value, and link drift marker value are added together to obtain the comprehensive anomaly judgment parameter. When the comprehensive anomaly judgment parameter is equal to 0, normal progress information is output; when the comprehensive anomaly judgment parameter is equal to 1, local anomaly prompt information is output; when the comprehensive anomaly judgment parameter is greater than or equal to 2, anomaly warning level dynamic management information is output, and the corresponding project item is marked as a key verification item.

[0062] By jointly identifying state missing, state transition anomalies, and link drift, the actual progress status of major traditional Chinese medicine projects can be accurately reflected at the continuous data chain level. This enhances the ability to locate abnormal nodes, abnormal links, and abnormal progress behaviors, thereby improving the accuracy of progress identification and the timeliness of dynamic management throughout the entire process.

[0063] The embodiments of the present invention provide, as follows Figure 2 The diagram shown illustrates the structure of a multi-source data-based information management system for major Chinese herbal medicine products. This system includes: The standard mapping module is used to obtain the project execution data corresponding to major Chinese medicine projects, and perform standardized mapping processing to establish a one-to-one or one-to-many relationship between project items and business nodes, thereby obtaining a standard item dataset. The temporal reconstruction module is used to perform temporal compensation and reconstruction on standard event datasets to generate candidate temporal segments. The link building module is used to perform chain merging within the preset project item association range. When the business node sequence corresponding to the candidate time sequence segment meets the preset connection conditions, it will build a continuous data chain for the same project item. Otherwise, it will mark the break and retain it for verification. The status recognition module is used to identify status gaps and status transition anomalies in the same project process based on continuous data chains, so as to determine whether to output dynamic management information for the whole process progress recognition of major Chinese medicine projects.

[0064] It should be added that, such as Figure 3 The diagram illustrates abnormalities in the progress of major traditional Chinese medicine (TCM) projects. From left to right, it shows three types of anomalies: missing status, abrupt changes, and link drift. The left-hand section shows missing status nodes (circled by dashed lines) indicating that intermediate nodes did not proceed correctly, ending at the product application stage. The missing status nodes are marked below. The middle section shows abrupt changes (red zigzag lines) where subsequent nodes jump upwards from project initiation to R&D, directly connecting to market launch, with a sudden status change marked below. The right-hand section shows the main link from project initiation, R&D, clinical trials to market launch, but also includes a deviating, curved connection path, marked as "link drift" below, with a "link path anomaly" indicator at the bottom. This diagram visually represents these three types of anomalies during project progress.

[0065] like Figure 5 The diagram illustrates time compensation and sequence correction. The top of the diagram shows the initial time series before compensation, with data points A1, A2, A3, A4, and A5 as the objects. The timelines T1 to T7 are marked with the business occurrence time, status update time, and supplementary entry time, respectively. Blue dots represent business occurrence times, orange dots represent status update times, and green dots represent supplementary entry times. The diagram also includes a time compensation matrix listing the corresponding time compensation values ​​for A1 to A5. The bottom of the diagram shows the time series after compensation. A1, A2, A3, A4, and A5 are repositioned on the timeline using gray rectangles, indicating the corrected position of the data points. The legend clearly shows four ways to identify the business occurrence time, status update time, supplementary entry time, and the corrected position of the data points. This diagram demonstrates how a time compensation matrix is ​​constructed based on the time offset relationship between the business occurrence time, status update time, and supplementary entry time, and how the data points are repositioned to their corrected time positions based on the compensation results.

[0066] like Figure 6The diagram illustrates the generation of candidate time series segments. The top of the diagram shows the original node sequence, including D1: Node A / Pending Review, D2: Node B / Screening Passed, D3: Node C / Center Initiation, D4: Node D / Subject Enrollment, D5: Node E / Visit Execution, and D6: Node F / Quality Control Correction. The middle section, titled "Based on Whether Adjacent Nodes Meet the Connection Conditions," shows the judgment results between adjacent nodes. D1→D2, D2→D3, D3→D4, and D5→D6 are marked as satisfied, while D4→D5 is marked as not satisfied. The bottom of the diagram shows the generated candidate time series segments. D1, D2, D3, and D4 are classified as candidate time series segment F1, and D5 and D6 are classified as candidate time series segment F2. F1 and F2 are separated by two red vertical lines. This diagram demonstrates how different candidate time series segments are generated by chaining or preserving breaks in the original node sequence based on whether adjacent nodes meet the connection conditions.

[0067] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0069] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0070] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0071] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0072] 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 managing information on major traditional Chinese medicine products based on multi-source data, characterized in that: The method includes: Obtain the project execution data corresponding to major Chinese medicine projects, perform standardized mapping processing, establish a one-to-one or one-to-many relationship between project items and business nodes, and obtain a standard item dataset. The project execution data includes at least project initiation data, drug management data, subject data, visit data, quality control data, and statistical analysis data; The standard event dataset is subjected to time series compensation and reconstruction to generate candidate time series segments; Within the preset project item association range, when the business node sequence corresponding to the candidate time sequence segment meets the preset connection conditions, chain merging is performed to construct a continuous data chain for the same project item; otherwise, a break mark is made and the data is retained for verification. The preset project item association scope includes at least one or more of the following: the same project identifier, the same research center identifier, the same subject identifier, the same visit identifier, and the same item type identifier; Based on the continuous data chain, identify missing states and abnormal state transitions in the same project process to determine whether to output dynamic management information for the whole process progress identification of major Chinese medicine projects.

2. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 1, characterized in that, The standardized mapping process includes: Extract the node name field, status field, time field, and object identifier field from the project execution data; The semantic parameters of the item are obtained based on the node name field and the status field. The semantic parameters of the item are used to characterize the degree of correspondence between the current data record and the preset project item in terms of item category and status meaning. The time-series attribution parameter is obtained based on the time field and the object identifier field. The time-series attribution parameter is used to characterize the current data record’s stage position in the project process and the project object to which it belongs. The semantic parameters of the items and the time sequence attribution parameters are combined to obtain the comprehensive mapping parameters, which are used to characterize the overall correspondence between the current data record and each preset item. Based on the comprehensive mapping parameters, a mapping sequence between data records and project items is constructed, and a one-to-one mapping relationship between the project execution data and the highest matching project item is established according to the highest matching project item corresponding to each project execution data in the mapping sequence. Based on the one-to-one mapping relationship, the project execution data is merged to obtain a standardized merged data set.

3. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 1, characterized in that, The standardized mapping process also includes: Based on the completion of the one-to-one mapping between the project execution data and the highest matching project item, if multiple business node data mapped to the same project item are identified, a one-to-many association mapping relationship between the project execution data and the multiple business nodes is established based on the state change sequence and time connection relationship corresponding to the multiple business node data. Based on the one-to-many association mapping relationship, the data of multiple business nodes belonging to the same project item are aggregated to obtain an aggregated and associated item dataset.

4. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 2 or 3, characterized in that, The timing compensation and reconstruction include: Extract the data of each item from the standard merged item dataset or the aggregated associated item dataset, including the corresponding business occurrence time, status update time, and supplementary entry time, and perform identification and grouping processing to generate an initial time series; Based on the data entry delay value and update time offset value of each item in the initial time series, a time compensation parameter is calculated. The time compensation parameter is used to characterize the degree of deviation of the current item data in terms of time record completeness and time position accuracy. A time compensation matrix is ​​constructed based on the time correlation between the time compensation parameters and the data of each item, so as to uniformly correct the time position corresponding to each item data and obtain a compensation time series. The compensation time series is used to characterize the chronological arrangement of each item data corresponding to the same project item after time deviation correction.

5. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 4, characterized in that, The timing compensation and reconstruction also include: Based on the compensation time series, and combined with the node order relationship parameters and state connection parameters between the data of each item, the data of each item corresponding to the same project item are rearranged and chained together to generate candidate time series segments. When adjacent data items meet the preset connection conditions, the corresponding adjacent data items are connected into the same candidate time sequence segment; otherwise, the data items that do not meet the preset connection conditions are retained as independent candidate time sequence segments. The preset connection conditions include at least the following conditions: matching the order of business nodes, consistent state transition direction, and time interval within a preset tolerance range.

6. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 5, characterized in that, The chain merging process is as follows: The candidate time series segments are sorted according to their start and end times to generate a data chain segment sequence to be merged. When the segment connection parameters corresponding to the data link segment sequence meet the preset merging conditions, the adjacent candidate time series segments corresponding to the data link segment sequence are connected end to end to form a merged segment; otherwise, the independent segment state of the current candidate time series segment is retained and a break mark is made. Using the merged segment as a new candidate time sequence segment, continue to perform connection judgment and chain merging with subsequent candidate time sequence segments until there are no adjacent candidate time sequence segments that meet the preset merging conditions. The final merged segment is then determined as a continuous data chain for the same project item. The segment connection parameters are obtained by matching and analyzing the node connection relationship and time interval relationship between adjacent candidate time segments.

7. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 6, characterized in that, The identification of missing states and abnormal state transitions during the progress of the same project includes at least the following: Identification of state gaps in the transition between subject screening and subject enrollment, and identification of abnormal state transitions in the transition between subject enrollment and visit execution; The state missing identification is used to identify whether there are situations in the continuous data chain corresponding to the same subject where the subject's pre-enrollment node is missing, the pre-enrollment node is incomplete, or the connection between the pre-enrollment node and the enrollment node is abnormal. The state transition anomaly identification is used to identify whether there are cases in the continuous data chain corresponding to the same subject where the subject directly enters the subsequent visit state without going through a preset intermediate execution state.

8. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 7, characterized in that, The state missing identification includes: Extract the screening node, enrollment node, node occurrence time, and subject identification information corresponding to the same subject from the continuous data chain; If a filter node exists in a continuous data chain and the filter node is located before the in group node, assign an order association value; otherwise, assign an abnormal order value. Calculate the time interval between the screening node and the group entry node, and assign a time-related value based on whether the time interval is within a preset tolerance range; Determine whether the subject identifiers corresponding to the screening node and the enrollment node are consistent, and assign an attribution association value based on the consistency result; The sequence association value, time association value, and affiliation association value are combined to obtain the node association parameters; If the node association parameter indicates that the filtering node is missing, the filtering node is not completed before the entry node appears, or the time connection between the filtering node and the entry node is abnormal, then it is determined that there is a state missing in the continuous data chain, and a corresponding missing state marker is generated. Otherwise, the connection status between the filtering node and the grouping node in the continuous data chain is determined to be normal, and a corresponding non-missing status marker is generated.

9. The method for managing information on major traditional Chinese medicine products based on multi-source data as described in claim 7, characterized in that, The state transition anomaly identification includes: Extract the enrollment status, visit preparation status, visit execution status, and status location number, status update time, and subject identification information corresponding to the same subject from the continuous data chain. The state traversal parameter is calculated based on the state position number. The state traversal parameter is used to characterize the number of intermediate states traversed between two adjacent states in the current continuous data chain. Based on the missing status of the visit preparation state in the continuous data chain and the update time interval between the group entry state and the visit execution state, the intermediate state missing parameter is calculated. The intermediate state missing parameter is used to characterize the degree of missing intermediate execution links in the current state advancement process. The state transition parameters are determined based on the state crossing parameters and the intermediate state incomplete parameters. When the state transition parameter exceeds the preset abnormal threshold, it is determined that there is a state transition abnormality in the continuous data chain, and a corresponding transition abnormality flag is generated. Otherwise, it is determined that the state progression between the group entry state and the visit execution state in the continuous data chain is normal, and a corresponding non-jump state flag is generated.

10. A project information management system for major traditional Chinese medicine products based on multi-source data, characterized in that: The system includes: The standard mapping module is used to obtain the project execution data corresponding to major Chinese medicine projects, and perform standardized mapping processing to establish a one-to-one or one-to-many relationship between project items and business nodes, thereby obtaining a standard item dataset. The temporal reconstruction module is used to perform temporal compensation and reconstruction on standard event datasets to generate candidate temporal segments. The link building module is used to perform chain merging within the preset project item association range. When the business node sequence corresponding to the candidate time sequence segment meets the preset connection conditions, it will build a continuous data chain for the same project item. Otherwise, it will mark the break and retain it for verification. The status recognition module is used to identify status gaps and status transition anomalies in the same project process based on continuous data chains, so as to determine whether to output dynamic management information for the whole process progress recognition of major Chinese medicine projects.

Images