A drug traceability association method and system based on patient longitudinal visit trajectory

Abstract

The present application relates to the technical field of data analysis, in particular to a drug trace correlation method and system based on patient longitudinal medical trajectory, comprising: when a drug correlation event is obtained, evidence slices are extracted from a longitudinal drug event set according to the occurrence time point of the drug correlation event; a drug active overlapping section is constructed in a candidate concurrent window; exposure risk is identified, specific drug events and supporting evidence are matched, bridge information is constructed, and output is performed. In view of the problem that the drug prescription trace of the prior art is only directed to the prescription, and it is difficult to accurately locate the specific drug exposure event, the longitudinal drug event set is constructed by processing medical data, evidence slices are extracted according to the drug correlation event when the drug correlation event occurs, and a candidate concurrent window is constructed, then the section with drug active overlap is found out, the exposure risk is identified, and finally the specific drug event and supporting evidence are located to construct the bridge information for traceability.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to a method and system for drug traceability and association based on a patient's longitudinal medical visit trajectory. Background Technology

[0002] Drug co-exposure refers to the phenomenon where the body is simultaneously or sequentially exposed to two or more drug components, resulting in alterations in systemic exposure (drug concentration) and biological effects. Clinically, drug co-exposure typically occurs when patients receive treatment for chronic disease management or within a short period, and are given two or more prescriptions of medication sequentially or simultaneously. If prescribing history or treatment details are omitted, there is a potential for further risks. Therefore, there is a strict need for co-exposure risk assessment in clinical practice. Existing technologies already exist for managing this type of risk.

[0003] For example, patent application CN202110209289.8 discloses a method, system, and computer storage medium for providing medication recommendations for children. The method includes the following steps: Step S1: Obtain a genetic testing report; Step S2: Import the report into the system; Step S3: Obtain adverse drug reaction data; Step S4: Store the adverse drug reaction data from Step S3; Step S5: Record the user's vaccination data and growth and development records; Step S6: Establish an independent and traceable file for the user; Step S7: Send medication recommendations to the terminal device. Sampling is convenient and comprehensive, requiring only a small number of samples, and various types of samples can be submitted for testing; a personalized file is established for each user, recording growth and development records, vaccination records, and other information; medication recommendations based on genetic testing are made more accurate; users can use the terminal device to query adverse drug reaction data, user vaccination data, and growth and development records, or to seek medication or medical advice, greatly facilitating user access.

[0004] For example, patent application CN202310698491.0 discloses a trusted method for the circulation of electronic medical prescriptions. This method includes: the electronic prescription initiator setting the circulation order and generating corresponding initial key pairs for all participants; the electronic prescription initiator updating a temporary key and using the temporary key to sign the constructed circulation data, obtaining signed circulation data; when a participant receives the signed circulation data, they update their temporary key and perform identity verification; the verified participant receives the circulation data, decrypts and approves the data, and performs signature circulation processing based on the temporary key after approval. According to the proposed circulation method, before circulation, the data must be signed using the approver's own temporary key, ensuring the transparency and traceability of the electronic prescription approval process.

[0005] However, in actual implementation, the inventors found that such technical solutions usually only trace a single prescription and lack effective identification of problems such as repeated medication, short-term continuous switching of similar drugs, and long-term continuous exposure. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, a drug traceability and association method based on the patient's longitudinal medical visit trajectory is provided. On the other hand, a system for implementing this method is also provided.

[0007] The specific technical solution is as follows: A drug traceability and association method based on a patient's longitudinal medical visit trajectory includes: Step S1: Acquire the patient's medical data and establish a longitudinal pharmacological event set; Step S2: When a drug-related event is obtained, trace and extract evidence slices from the longitudinal drug event set according to the occurrence time of the drug-related event; Step S3: Construct candidate concurrency windows for the vertical drug event set, and construct drug activity overlap segments within the candidate concurrency windows; Step S4: Identify the exposure risk in the active overlapping segment of the drug and match specific drug events and supporting evidence to construct bridging information and output it.

[0008] On the other hand, step S1 includes: Step S11: Obtain the medical data and standardize the medical data to obtain standardized medication records; Step S12: Establish a longitudinal drug event trajectory for the standardized medication records according to time sequence, and generate traceability identifiers for each drug event; The longitudinal drug event trajectory is composed of an arrangement of multiple drug events; Step S13: Obtain the longitudinal drug event set based on the longitudinal drug event trajectory combination.

[0009] On the other hand, step S2 includes: Step S21: For the time point of occurrence of the drug-related event, extract the drug-related prescription and historical medical history respectively; Step S22: Extract the current evidence slice based on the drug-related prescription, and extract the longitudinal evidence slice based on the historical medical history; Step S23: Establish the evidence slice based on the current evidence slice and the longitudinal evidence slice. On the other hand, step S3 includes: Step S31: Obtain the start and end times of the events from the evidence slices, and perform a sliding window traversal on the drug events to obtain the candidate concurrent windows; Step S32: Extract the event boundary of each drug event in the candidate concurrency window; Step S33: Investigate the active overlapping segments that overlap in time according to the event boundaries.

[0010] On the other hand, step S4 includes: Step S41: Combine the active overlapping segments with the evidence fragments to identify problems and obtain a set of problem objects; Step S42: Perform reverse matching between the set of problem objects and the set of vertical drug events to obtain the specific drug event; Step S43: Obtain the supporting evidence based on the specific drug event matching, and construct the bridging information output.

[0011] On the other hand, in step S41, the active overlapping segment is pre-identified using the first track, and when a preset condition is triggered, the second track is used for identification.

[0012] A drug traceability and association system based on the patient's longitudinal medical visit trajectory, used to implement the above-mentioned drug traceability and association method; The drug traceability system includes: The acquisition module acquires the patient's medical data and establishes a longitudinal set of pharmacological events; The evidence slice generation module, when acquiring a drug-related event, traces and extracts evidence slices from the longitudinal drug event set based on the occurrence time of the drug-related event. A window construction module is used to construct candidate concurrent windows for the vertical drug event set, and to construct drug activity overlap segments within the candidate concurrent windows. The output module identifies the exposure risk in the drug active overlap segment, matches specific drug events and supporting evidence to construct bridging information, and outputs it.

[0013] On the other hand, the acquisition module includes: A standardization module acquires the medical data and standardizes the medical data to obtain standardized medication records; The arrangement module establishes a longitudinal drug event trajectory for the standardized medication records according to time sequence, and generates traceability identifiers for each drug event. The longitudinal drug event trajectory is composed of an arrangement of multiple drug events; A combination module, which obtains the set of vertical drug events based on the combination of the vertical drug event trajectories.

[0014] On the other hand, the evidence slice generation module includes: The extraction module extracts drug-related prescriptions and medical histories at the time of occurrence of the drug-related event. The evidence collection module extracts current evidence slices based on the drug-related prescriptions and extracts longitudinal evidence slices based on the historical medical history. The packaging module creates the evidence slice based on the current evidence slice and the vertical evidence slice. On the other hand, the window building module includes: The window candidate module obtains the start and end times of events from the evidence slices and performs a sliding window traversal on the drug events to obtain the candidate concurrent windows. A boundary determination module extracts the event boundary for each drug event within the candidate concurrency window. An overlap determination module is used to identify active overlapping segments that overlap in time based on the event boundary.

[0015] The above technical solution has the following advantages or beneficial effects: To address the issue that existing drug prescription tracing technologies only target prescriptions and struggle to accurately pinpoint specific drug exposure events, this paper addresses the problem by processing medical data to construct a longitudinal set of drug events. When a drug-related event occurs, evidence slices are extracted based on the drug-related events, and candidate concurrent windows are constructed. Subsequently, segments with overlapping drug activity are identified to determine the exposure risk, and finally, the specific drug event and supporting evidence are located to construct bridging information for easy tracing. Attached Figure Description

[0016] Embodiments of the invention will be described more fully with reference to the accompanying drawings. However, the drawings are for illustration and explanation only and do not constitute a limitation on the scope of the invention.

[0017] Figure 1 This is an overall schematic diagram of an embodiment of the present invention; Figure 2 This is a schematic diagram of step S1 in an embodiment of the present invention; Figure 3 This is a schematic diagram of step S2 in an embodiment of the present invention; Figure 4 This is a schematic diagram of step S3 in an embodiment of the present invention; Figure 5 This is a schematic diagram of step S4 in an embodiment of the present invention; Figure 6 This is a schematic diagram of the system in an embodiment of the present invention; Figure 7 This is a schematic diagram of the acquisition module in an embodiment of the present invention; Figure 8 This is a schematic diagram of the evidence slice generation module in an embodiment of the present invention; Figure 9 This is a schematic diagram of the window construction module in an embodiment of 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] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0020] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0021] This invention includes: A method for tracing and associating medications based on a patient's longitudinal medical visit trajectory, such as... Figure 1 As shown, it includes: Step S1: Acquire the patient's medical data and establish a longitudinal pharmacological event set; Step S2: When drug-related events are obtained, evidence slices are extracted from the longitudinal drug event set based on the occurrence time of the drug-related events. Step S3: Construct candidate concurrency windows for the vertical drug event set, and construct drug activity overlap segments within the candidate concurrency windows; Step S4: Identify exposure risks in the drug active overlap segment, match specific drug events and supporting evidence to construct bridging information and output it.

[0022] Specifically, addressing the issue that existing drug prescription tracing technologies only target prescriptions and struggle to accurately pinpoint specific drug exposure events, this paper addresses the problem by processing medical data to construct a longitudinal set of drug events. When a drug-related event occurs, evidence slices are extracted based on the drug-related event, and candidate concurrent windows are constructed. Subsequently, segments with overlapping drug activity are identified to determine the exposure risk, and finally, specific drug events and supporting evidence are located to construct bridging information for easy tracing.

[0023] Specifically, in actual implementation, the above technical solutions are mainly configured as software embodiments in computer devices, such as medical workstations, to connect to the hospital's electronic medical record system or other equivalent electronic systems, to collect patients' medical data, and to screen out specific drug exposure risks when drug-related events are identified, and to screen out specific drug events from the longitudinal drug event set, as well as specific evidence from the evidence slice, to form traceability data and output it.

[0024] The medical data includes: patient identifier, source scenario, drug name, drug code, start time, end time, route of administration, frequency, dosage, visit serial number, medical institution code, original number of days of medication, and estimated number of days of medication. This information is used to create a single vertical drug event object for each visit record.

[0025] For the same patient, the vertical drug event objects are organized into a vertical drug event set for subsequent queries.

[0026] After generating specific drug-related events, the time points of occurrence can be extracted to trace the longitudinal set of drug events. Evidence slices can be obtained by combining pre-configured diagnostic and disease course evidence data, which can be used in subsequent determination of indications and other situations.

[0027] Subsequently, candidate concurrent windows are constructed for the longitudinal drug event set to extract time periods where drug exposure risks may exist. Then, it is determined which drug events are active within the time slice to obtain the drug active overlap segment.

[0028] Finally, based on the overlapping segments of drug activity, the system reverse-engineers the vertical drug event objects from the set of vertical drug events. Combining this with a series of pre-defined rules and knowledge data, including structured rules, indication rules, duplicate dosing judgment rules, and continuous exposure rules, it performs separate discrimination to obtain specific drug events and supporting evidence. For specific location down to the medical institution and prescription, bridging information is also established for reverse tracing. The bridging record is recommended to include at least: Bridge record identifier, issue identifier, issue title, issue severity, related drug set, time overlap start point, time overlap end point, event unique key, visit unique key, visit type, visit serial number, medical institution code, drug name, and drug code.

[0029] This establishes a stable connection between the problem's outcome and the underlying events, making it easier to trace back.

[0030] In one embodiment, such as Figure 2 As shown, step S1 includes: Step S11: Obtain medical data and standardize the medical data to obtain standardized medication records; Step S12: Establish a longitudinal drug event trajectory for standardized medication records according to time sequence, and generate traceability identifiers for each drug event; A longitudinal drug event trajectory consists of an arrangement of multiple drug events. Step S13: Obtain the set of vertical drug events based on the combination of vertical drug event trajectories.

[0031] Specifically, in order to achieve a better effect in establishing a longitudinal drug event set, this embodiment first acquires medical data generated by patients at different times, different types of visits, and different institutions, including at least medication records, medical records, diagnostic or disease course evidence, laboratory tests, and basic patient information, and then performs standardized processing.

[0032] Standardization processes include unifying time representation, drug identification, route of administration, visit type, and institution of origin, enabling cross-source records to be included in the same trajectory analysis framework. For duplicate records with identical start and end times and drug identification, deduplication is performed; for duplicate prescription records from the same source, on the same date, and for the same drug, representative events with longer durations or earlier start times are retained.

[0033] For drug administration events that are not suitable for inclusion in the systemic co-exposure assessment, such as topical or inhalation drug administration events confirmed by preset rules, their original trajectory information can be retained, but they will not be included in the drug quantity calculation for systemic repeated drug use risk.

[0034] Subsequently, standardized medication records were entered as drug events and assigned traceability identifiers, while a longitudinal drug event trajectory was established according to time sequence.

[0035] The traceability identifiers include a unique key for each medical visit and a unique key for each event.

[0036] For each medication event, the system generates a unique visit key, visit_uid. The preferred rule is: 1. If a visit type and visit serial number exist, then visit_uid = visit type + visit serial number; 2. If only the patient's visit serial number exists, then use the patient's visit serial number as the visit_uid; 3. If the visit serial number is missing but the event start time exists, then visit_uid = visit type + event start time; 4. If none of the above fields are sufficient, the available source field will be retained as the downgraded visit_uid.

[0037] This design ensures that the system maintains an unbroken traceability link even when some data lacks a standard serial number.

[0038] The system further generates a unique event key, event_uid. The preferred rule is: 1. Obtain the visit_uid of the event; 2. Obtain the drug code or medical insurance code; 3. Obtain the event start time; 4. Combine the above fields into event_uid; 5. If a key field is missing, fall back to the available drug name or other event identifier.

[0039] The event_uid does not depend on the internal row number of the database, making it suitable for stable referencing in CSV export, report display, expert review, and subsequent statistics.

[0040] Meanwhile, in the process of adding longitudinal drug event trajectories, the system also expands them into review items as intermediate items for review, including normalized route of administration, whether it is for external use, whether it is a solvent, current evidence slice, longitudinal indication evidence slice, continuous diagnosis set, basic drug information, and special population information.

[0041] After all events are organized in chronological order, a patient-level longitudinal drug event trajectory is formed.

[0042] Finally, a set of longitudinal drug event trajectories is established based on all longitudinal drug event trajectories.

[0043] In one embodiment, such as Figure 3 As shown, step S2 includes: Step S21: Extract the drug-related prescriptions and medical history for the time points of occurrence of drug-related events; Step S22: Extract the current evidence slice based on the drug-related prescription, and extract the longitudinal evidence slice based on the historical medical history; Step S23: Create an evidence slice based on the current evidence slice and the longitudinal evidence slice.

[0044] Specifically, for drug-related events, the system extracts drug-related prescriptions and historical medical history. The drug-related prescriptions are used to directly link to a single patient visit record to create a current evidence slice. The current evidence slice covers evidence that can directly explain the medication use, such as the chief complaint, diagnosis, past medical history, present medical history, and treatment plan from visits near the anchor point.

[0045] Simultaneously, the system extracts longitudinal evidence slices within a longer review window. These longitudinal evidence slices cover historical chief complaints, historical diagnoses, present illness history, ongoing diagnostic sets, and past treatment information, used to determine whether the current prescription is supported by the existing disease course or long-term treatment goals.

[0046] In subsequent processing, the current evidence slice and the longitudinal evidence slice are matched separately. This stratification method can utilize historical disease information without directly equating weakly correlated historical diagnoses with strong supporting evidence at the current point in time.

[0047] This mechanism addresses common misjudgments in single prescription review. By stratifying current evidence slices with longitudinal evidence slices, the system can utilize longitudinal disease course evidence while preserving differences in the strength of evidence.

[0048] In one embodiment, such as Figure 4 As shown, step S3 includes: Step S31: Obtain the start and end times of events from the evidence slices, and use the sliding window method to traverse the drug events to obtain candidate concurrent windows; Step S32: Extract the event boundary for each drug event in the candidate concurrency window; Step S33: Investigate active overlapping segments that overlap in time based on event boundaries.

[0049] Specifically, to achieve better identification of active overlapping segments, the system first sorts patient medication events by start time, end time, drug code, and drug name, and then performs deduplication. Deduplication includes at least: deduplicating events with completely identical start and end times and drug codes; retaining representative events with longer durations or earlier starts for duplicate prescription records from the same source, on the same day, and with the same drug code; and excluding events unsuitable for systemic co-exposure assessment, such as topical or inhaled administration events. The system groups medication events with temporal continuity or overlap into the same candidate concurrency window as a coarse-grained container.

[0050] Subsequently, within a candidate concurrency window, the system extracts the start date and the day after the end date of all drug events as time boundary points and sorts them by boundary points. A candidate time slice is formed between adjacent boundary points. For each candidate time slice, the system determines which drug events are active within that time slice. If there are fewer than two active drug events, the time slice does not constitute a co-exposure segment; if there are two or more active drug events, an active overlap segment is formed.

[0051] If the active drug sets of adjacent time slices are completely identical, they are merged into the same active overlap segment. Each active overlap segment must contain at least: overlap segment start date, overlap segment end date, active drug event index set, overlap days and corresponding drug name, medical insurance code or pharmacological concept set.

[0052] To avoid misjudging duplicate records of the same drug as multiple drug overlap, the system further performs drug identity merging within the active segment, including: prioritizing merging the same drug by medical insurance code or drug code; merging by combining drug name, standardized route of administration and drug identity key when necessary; retaining only one representative event for multiple events with the same identity in the calculation of "number of different drugs"; and excluding pre-defined exempted routes of administration such as external use and inhalation from the systemic duplicate drug risk assessment.

[0053] After the above process, only the merged active overlapping segments are retained for subsequent identification.

[0054] In one embodiment, such as Figure 5 As shown, step S4 includes: Step S41: Identify problems by combining active overlapping segments with evidence fragments to obtain a set of problem objects; Step S42: Perform reverse matching between the problem object set and the vertical drug event set to obtain specific drug events; Step S43: Obtain supporting evidence based on specific drug events and construct bridging information output.

[0055] Specifically, in order to achieve better problem identification results, in this embodiment, the active overlapping segments are first combined with evidence fragments to identify problems in order to obtain a set of problem objects.

[0056] During this process, the system uses a modular approach to identify drug-related issues, including risks of overlapping medications, long-term exposure risks, whether the evidence supports the indication, side effects, and pharmaceutical issues.

[0057] For overlapping risks, the system classifies and determines overlapping issues based on the duration of the active overlapping period, drug identity, pharmacological concept, route of administration, and preset exception rules. When the number of overlapping days reaches the standard reporting threshold, the system outputs a standard overlapping risk. When the number of overlapping days is within the short-term overlap range, a medium-risk warning is only output if a specific pharmacological concept or a specific drug focus set is matched; no issue is output if the short-term focus configuration is not matched. When the allowed combination, route of administration exception, or maximum overlapping days exemption rule is matched, the system can skip issue generation or adjust the issue severity. The relevant drug set and trigger time window are retained when an issue is generated as the basis for subsequent reverse matching.

[0058] To address long-term exposure risks, the system converts longitudinal drug events into exposure entries and matches them with target continuous exposure rules based on pharmacological concepts or drug name keywords. The matched exposure entries are then arranged chronologically, and adjacent entries are concatenated into a single continuous exposure segment if the interval between them does not exceed a preset maximum number of days. If the interval exceeds the maximum number of days, the exposure is split into different segments to distinguish between continuous treatment and restarts after interruption. Finally, the start time, end time, and duration of each continuous exposure segment are calculated, and continuous exposure issues of varying severity are output based on a duration threshold. This mechanism can identify long-term continuous exposures with different drug names but belonging to the same pharmacological concept.

[0059] To address the question of whether the evidence supports the indication, the current evidence slice is first matched with the structured indication rules for the drug.

[0060] When the current evidence directly supports the purpose of medication, it is deemed to directly support it.

[0061] If the current evidence does not match, but the longitudinal evidence slices or continuous diagnostic set match the corresponding indication rules, it is judged as weak support in the longitudinal course of the disease. Weak support results are not directly output as obvious off-label issues, but can be used for reporting explanations or retrospective review.

[0062] Only if neither current nor longitudinal evidence supports the indication, and there are no pre-defined weak supporting circumstances or conditions for manual review, will the system identify it as an indication evidence gap, off-label risk, or subject to review. This process reduces the likelihood of misjudging reasonable chronic disease prescription renewals as off-label, while preserving the differences in the strength of evidence.

[0063] To address side effect issues, suspected side effect evidence, such as symptoms, chief complaints, present medical history, adverse reaction descriptions, and abnormal laboratory results, is extracted from patients' longitudinal medical history. Suspected related drugs and their exposure time ranges are also extracted from a set of drug events. Subsequently, matching is performed according to pre-defined drug-side effect knowledge entries or keyword rules, and the matching results are output as problem objects or objects to be reviewed. The key to this process is tracing suspected side effect issues back to specific drug exposure events, rather than predicting the probability of adverse reactions occurring.

[0064] For pharmaceutical issues, a modular approach is used to identify multiple drug-related problems. The modules cover contraindication conflicts, off-label use, dosage discrepancies, duplicate dosing, continuous exposure, interactions, and specific risk signals. Each module uses the same longitudinal trajectory, review entries, and evidence slices under a unified orchestration entry point to perform identification, and outputs problem results with a uniform structure. If an anomaly occurs in a module, the system isolates the anomaly while retaining procedural information, ensuring that other modules can continue to execute.

[0065] In the specific identification process, the above identification uses two sets of track models. First, the first track is used to identify active overlapping segments in advance. When a preset condition is triggered, the second track is used for identification.

[0066] The first track is a high-speed basic review track based on hard-coded pharmaceutical rules, which directly utilizes longitudinal trajectories, active overlapping segments, continuous exposure segments, and evidence slices to complete batch screening.

[0067] The second track is triggered when there are knowledge gaps in the first track, insufficient explanation of complex disease courses, individualization issues with multiple pieces of evidence across time for patients, or when a high-level semantic review of the first track results is required.

[0068] When the second track is triggered, the system does not directly input all the original patient data. Instead, the trajectory layer and the first track first trim candidate related drugs, candidate time windows, candidate visit sets, candidate longitudinal evidence slices, and candidate bridging associations. Then, the trimmed structured context is input into the trained model for individualized analysis.

[0069] When both tracks have outputs, the system merges the results: consistent results are confirmed, conflicting results are adjudicated according to preset strategies, and rule conclusions and model conclusions are displayed in layers; if the model fails to process the data, it reverts to the rule track results to ensure the continuous availability of the batch review process.

[0070] Before the results generated by the above modules are entered into the final output, the system performs deduplication aggregation, suppression rule processing, and severity merging. For duplicate issues generated by the same triggering fact, the final issue that can represent that fact is retained; for results with different severity, the final severity is retained according to a preset strategy. During the aggregation process, procedural evidence related to the source of the issue, the triggering time, and relevant drugs is retained for subsequent reverse matching and expert review.

[0071] After identifying the problem, the system extracts the relevant drug set and triggering time window from the problem results and iterates through all longitudinal drug events for the patient. The system filters events whose drug identity belongs to the relevant drug set; when the problem has a time window, it further requires that the event time intersects with that window.

[0072] The matching results are deduplicated based on event-level stability identifiers, forming a set of candidate events that trigger the issue. If the issue result does not carry a specific time window, it falls back to matching only based on the relevant drug set. This reverse matching reduces the issue from the drug name level back to the specific event level.

[0073] When the problem involves only a single drug, or when the relevant drug information is insufficient to constitute a valid multi-drug association, the system can skip generating invalid multi-drug bridging records to avoid forming misleading associations.

[0074] Finally, for ease of traceability, the system generates a bridging relationship for each question and each matched drug event. The bridging relationship semantically connects the question result, related drugs, triggering events, source of medical attention, institution, and temporal relationship; when the same question involves multiple triggering events, multiple bridging relationships are generated. This bridging table can be used for patient-level interpretation or directly exported to CSV or other structured output formats.

[0075] As an optional implementation, a detailed table of questions and medication bridging can be further derived. This table supports the following analyses: which medications constitute a question, whether these medications come from the same visit, whether these medications come from different institutions, and the distribution of a certain type of question across visit or institution trajectories, thereby significantly enhancing the interpretability and statistical capabilities of the questions. A drug traceability and association system based on the patient's longitudinal medical visit trajectory, used to implement the above-mentioned drug traceability and association method; like Figure 6 As shown, the drug traceability system includes: Acquisition Module 1 acquires the patient's medical data and establishes a longitudinal set of pharmacological events; Evidence slice generation module 2: When a drug-related event is obtained, evidence slice generation module 2 traces and extracts evidence slices from the vertical drug event set according to the occurrence time of the drug-related event. Window construction module 3 constructs candidate concurrent windows for the vertical drug event set, and constructs drug active overlapping segments within the candidate concurrent windows; Output module 4 identifies exposure risks in the drug activity overlap segment, matches specific drug events and supporting evidence to construct bridging information, and outputs it.

[0076] In one embodiment, such as Figure 7 As shown, the acquisition module 1 includes: Standardization module 11 acquires medical data and standardizes the medical data to obtain standardized medication records; Arrangement module 12 establishes a longitudinal drug event trajectory for standardized medication records according to time sequence, and generates traceability identifiers for each drug event. A longitudinal drug event trajectory consists of an arrangement of multiple drug events. Combination module 13 obtains a set of vertical drug events based on the combination of vertical drug event trajectories.

[0077] In one embodiment, such as Figure 8 As shown, the evidence slice generation module 2 includes: Extraction module 21 extracts drug-related prescriptions and historical medical history based on the time point of occurrence of drug-related events. Evidence collection module 22 extracts current evidence slices based on drug-related prescriptions and extracts longitudinal evidence slices based on historical medical history. Packaging module 23, the packaging module 23 creates evidence slices based on the current evidence slices and the longitudinal evidence slices. In one embodiment, such as Figure 9 As shown, window building module 3 includes: Window candidate module 31 obtains the start and end times of events from evidence slices and performs sliding window traversal on drug events to obtain candidate concurrent windows; Boundary determination module 32 extracts the event boundary for each drug event in the candidate concurrency window; The overlap determination module 33 checks active overlapping segments that overlap in time based on the event boundary.

[0078] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for drug traceability and association based on a patient's longitudinal medical visit trajectory, characterized in that, include: Step S1: Acquire the patient's medical data and establish a longitudinal pharmacological event set; Step S2: When a drug-related event is obtained, trace and extract evidence slices from the longitudinal drug event set according to the occurrence time of the drug-related event; Step S3: Construct candidate concurrency windows for the vertical drug event set, and construct drug activity overlap segments within the candidate concurrency windows; Step S4: Identify the exposure risk in the active overlapping segment of the drug and match specific drug events and supporting evidence to construct bridging information and output it.

2. The drug traceability and association method according to claim 1, characterized in that, Step S1 includes: Step S11: Obtain the medical data and standardize the medical data to obtain standardized medication records; Step S12: Establish a longitudinal drug event trajectory for the standardized medication records according to time sequence, and generate traceability identifiers for each drug event; The longitudinal drug event trajectory is composed of an arrangement of multiple drug events; Step S13: Obtain the longitudinal drug event set based on the longitudinal drug event trajectory combination.

3. The drug traceability and association method according to claim 2, characterized in that, Step S2 includes: Step S21: For the time point of occurrence of the drug-related event, extract the drug-related prescription and historical medical history respectively; Step S22: Extract the current evidence slice based on the drug-related prescription, and extract the longitudinal evidence slice based on the historical medical history; Step S23: Establish the evidence slice based on the current evidence slice and the longitudinal evidence slice.

4. The drug traceability and association method according to claim 2, characterized in that, Step S3 includes: Step S31: Obtain the start and end times of the events from the evidence slices, and perform a sliding window traversal on the drug events to obtain the candidate concurrent windows; Step S32: Extract the event boundary of each drug event in the candidate concurrency window; Step S33: Investigate the active overlapping segments that overlap in time according to the event boundaries.

5. The drug traceability and association method according to claim 1, characterized in that, Step S4 includes: Step S41: Combine the active overlapping segments with the evidence fragments to identify problems and obtain a set of problem objects; Step S42: Perform reverse matching between the set of problem objects and the set of vertical drug events to obtain the specific drug event; Step S43: Obtain the supporting evidence based on the specific drug event matching, and construct the bridging information output.

6. The drug traceability and association method according to claim 5, characterized in that, In step S41, the active overlapping segment is pre-identified using the first track, and when a preset condition is triggered, the second track is used for identification.

7. A drug traceability and association system based on a patient's longitudinal medical visit trajectory, characterized in that, Used to implement the drug traceability and association method as described in any one of claims 1-6; The drug traceability system includes: The acquisition module acquires the patient's medical data and establishes a longitudinal set of pharmacological events; The evidence slice generation module, when acquiring a drug-related event, traces and extracts evidence slices from the longitudinal drug event set based on the occurrence time of the drug-related event. A window construction module is used to construct candidate concurrent windows for the vertical drug event set, and to construct drug activity overlap segments within the candidate concurrent windows. The output module identifies the exposure risk in the drug active overlap segment, matches specific drug events and supporting evidence to construct bridging information, and outputs it.

8. The drug traceability and association system according to claim 7, characterized in that, The acquisition module includes: A standardization module acquires the medical data and standardizes the medical data to obtain standardized medication records; The arrangement module establishes a longitudinal drug event trajectory for the standardized medication records according to time sequence, and generates traceability identifiers for each drug event. The longitudinal drug event trajectory is composed of an arrangement of multiple drug events; A combination module, which obtains the set of vertical drug events based on the combination of the vertical drug event trajectories.

9. The drug traceability and association system according to claim 8, characterized in that, The evidence slice generation module includes: The extraction module extracts drug-related prescriptions and medical histories at the time of occurrence of the drug-related event. The evidence collection module extracts current evidence slices based on the drug-related prescriptions and extracts longitudinal evidence slices based on the historical medical history. A packaging module, which creates the evidence slice based on the current evidence slice and the longitudinal evidence slice.

10. The drug traceability and association system according to claim 8, characterized in that, The window construction module includes: The window candidate module obtains the start and end times of events from the evidence slices and performs a sliding window traversal on the drug events to obtain the candidate concurrent windows. A boundary determination module extracts the event boundary for each drug event within the candidate concurrency window. An overlap determination module is used to identify active overlapping segments that overlap in time based on the event boundary.

Images