A method and system for analyzing drug interaction information

Abstract

This invention relates to the field of pharmaceutical data analysis technology, specifically to a method and system for analyzing drug interaction information. The method includes the following steps: based on a pharmaceutical prescription entry system, it performs time-series correlation between patient medication records and physiological monitoring data to construct a medication behavior sequence feature group; it filters synergistic offset drug sets by judging continuous response change trends; and it outputs risk type identification data for drug combinations according to standard rules. This invention constructs a time-series mapping relationship between continuous medication data and physiological monitoring data, dynamically linking patient medication behavior with physiological responses. Parameter change trends, response amplitudes, and combination distributions are simultaneously collected and sorted. Risk features are extracted based on the synergistic offset characteristics between multiple drug combinations, enriching the data's hierarchical expressiveness, achieving multi-level interactive performance extraction and partitioning classification, enhancing the accuracy of risk type classification, and realizing continuous tracking and timely early warning of interactive risks in personalized medication processes.

Description

Technical Field

[0001] This invention relates to the field of pharmaceutical data analysis technology, and in particular to a method and system for analyzing drug interaction information. Background Technology

[0002] Pharmaceutical data analysis involves the systematic collection, integration, mining, and computational processing of drug-related information. This includes the integration and analysis of data from multiple dimensions, such as drug component data, pharmacokinetic parameters, biological target information, adverse drug reaction records, and clinical usage results. It is widely used in scenarios such as new drug screening, personalized medicine, and drug risk assessment. Traditional drug interaction information analysis methods refer to schemes that identify and analyze the potential adverse reactions or changes in efficacy that may result from the combined use of multiple drugs through literature retrieval, empirical judgment, and manual summarization. These methods typically rely on manually consulting pharmacopoeias or drug instructions for interaction descriptions and manually comparing the metabolic pathways, receptor binding characteristics, or transport mechanisms between drugs based on existing pharmacological knowledge. By analyzing known interaction information between drug combinations line by line, a drug safety analysis is completed.

[0003] Current drug interaction information analysis mainly relies on manual comparison of pharmacopoeia data and experience-based compilation, which makes it difficult to automatically identify the variable physiological responses caused by multiple drug combinations. The data collection and integration process lacks temporal sequence, making it difficult for the system to capture changes in key parameters during continuous medication for the same patient. The correlation between individual physiological responses and drug combinations is difficult to reflect dynamically, and the analysis results are often limited to static literature knowledge and single risk warnings, which cannot meet the needs of in-depth analysis of the synergistic effects of multiple drugs and continuous risk grading in the clinical setting. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for analyzing drug interaction information.

[0005] To achieve the above objectives, the present invention employs the following technical solution: a method for analyzing drug interaction information, comprising the following steps:

[0006] S1: Based on the medical prescription entry system, analyze the patient ID of the patient's medication record, determine whether the administration time and monitoring time overlap, screen the records within the monitoring interval, determine the event relationship, and reconstruct the data group to obtain the medication behavior sequence feature group;

[0007] S2: Based on the medication behavior sequence feature group, calculate the parameter change amplitude of adjacent monitoring records, compare the change direction before and after medication, analyze the correlation between parameter change amplitude and dose, adjust all parameter change amplitudes for normalization, and sort them according to numerical size or time order to obtain the dose response change sequence.

[0008] S3: Based on the dose response change sequence, compare the change trends of adjacent response sequences, analyze the trend maintenance, determine whether the response data after adjacent drug administration are continuously the same, screen drug combinations with continuous trends, and obtain a set of drugs with synergistic offset.

[0009] S4: Based on the synergistic offset drug set, calculate the combined interaction performance characteristics, analyze the drug response and dosing interval, adjust the performance content to a unified structure, compare the distribution of all combined interaction performances, and obtain the drug interaction feature set;

[0010] S5: Based on the drug interaction feature set, determine the status of the distribution segment, compare the segment interaction performance, analyze the distribution risk characteristics of drug combinations, screen combinations with the same characteristics, adjust the risk label classification, and obtain risk type label data.

[0011] The present invention improves upon this invention by including the following features: the medication behavior sequence feature set includes time tags, drug identifiers, and data association indexes; the dose response change sequence includes response change items, dose reference items, and sequence sorting tags; the co-offset drug set includes offset drug items, combination numbers, and trend indicators; the drug interaction feature set includes combination coordinates, performance intensity items, and block indexes; and the risk type identification data includes risk level, tag number, and grouping results.

[0012] The present invention is improved in that the step of obtaining the drug use behavior sequence feature group is specifically as follows:

[0013] S111: Based on the medical prescription entry system, analyze the medication records and monitoring data of physiological monitoring devices to determine whether the administration time in the medication records overlaps with the monitoring time in the monitoring data. By verifying the patient number associated with each medication record, screen the monitoring data within the corresponding time interval to obtain a time-series matching monitoring dataset.

[0014] S112: Based on the time-series matching monitoring dataset, analyze the monitoring time and patient number, screen all medication events of the same patient in overlapping time intervals, compare the time sequence of medication events, match the monitoring data before and after, and combine the associated monitoring content to obtain interval sequence pairing groups;

[0015] S113: Based on the interval sequence pairing group, analyze each group of medication events and associated monitoring data, determine the order of monitoring data, sort and configure time tags and drug identifiers according to monitoring time, and integrate the associated parameter index to obtain the medication behavior sequence feature group.

[0016] The present invention is improved in that the step of obtaining the dose response change sequence is specifically as follows:

[0017] S211: Based on the medication behavior sequence feature group, analyze the monitoring parameters corresponding to each group of medication behaviors, compare the differences of parameters in the monitoring records before and after medication, and by comparing the changes of the same parameter at adjacent monitoring time points, organize the changes of all monitoring parameters to obtain the parameter difference statistical group.

[0018] S212: Based on the parameter difference statistics group, determine the change trend of each monitoring parameter before and after medication, assign directional labels to the increasing and decreasing parameter contents, compare the dose data, and organize the dose contents and change directions to obtain a dose direction pairing set.

[0019] S213: Based on the dose direction pairing set, adjust all parameter changes to a unified standard format, arrange each group of dose and response data according to the time sequence of medication behavior, reconstruct it into a time-series data structure, and obtain the dose response change sequence.

[0020] The present invention is improved in that the step of obtaining the synergistic offset drug set is specifically as follows:

[0021] S311: Based on the dose response change sequence, analyze the change trend of response parameters corresponding to adjacent medication behaviors, compare the change direction of the same monitoring parameter under continuous medication behaviors, determine whether the trends at each time node are consistent, summarize all data segments with consistent trends, and obtain the trend continuation dataset.

[0022] S312: Based on the trend continuation dataset, screen medication events with consistent change directions, determine the performance of each drug in response trend continuity, and group medication behaviors with continuous change characteristics according to trend consistency to obtain synergistic response drug groups;

[0023] S313: Based on the synergistic response drug grouping, determine the order of drug arrangement and administration time, and combine trend consistency information to reconstruct the drug combination relationship of each group to obtain the synergistic offset drug set.

[0024] The present invention is improved in that the steps for obtaining the drug interaction feature set are specifically as follows:

[0025] S411: Based on the aforementioned synergistic offset drug set, analyze the physiological response data of each drug and the combination number corresponding to the monitoring interval, sort the parameter change amplitude by combining drug information and monitoring time, determine the change type corresponding to each drug, and obtain the combined physiological response change sequence.

[0026] S412: Based on the combined physiological response change sequence, compare the change type of each group with the drug administration time label, analyze the drug administration interval of adjacent drug items, calculate the standardized results of change amplitude, change type and drug administration interval under each combination number, and uniformly map them into a continuous interval structure to obtain the interactive performance feature vector set;

[0027] S413: Based on the aforementioned interaction performance feature vector set, calculate the interaction performance vector for each combination number, analyze the difference magnitude between the vector and the mean, sum the differences, and jointly determine the consistency of the dosing interval, response change magnitude dispersion, and change type under the same combination, using the formula:

[0028] ;

[0029] The intensity index of distribution differences for each combination is obtained to obtain the drug interaction feature set, where, Indicates the first Indicators of the intensity of distributional differences among drug combinations. Indicates the first The number of normalized interaction vectors for drug combination groups. Indicates the first The first normalized interaction performance vector of drug combination item, Indicates the first The mean of the normalized interaction performance vector of the drug combination groups. Indicates the first Dosing interval index of drug combination Indicates the first The dispersion of the response variation of the drug combination. Indicates the first Consistency index of changes in drug combination types It is used to traverse the first... The indices of all interaction vector entries under the drug combination, with values ​​ranging from 1 to... .

[0030] The present invention is improved in that the steps for obtaining the risk type identification data are specifically as follows:

[0031] S511: Based on the drug interaction feature set, determine the position of each drug combination in the coordinate axis space, filter the block to which each drug combination belongs according to the block index, and classify the correspondence between drug combinations and blocks to obtain the interaction distribution classification index group.

[0032] S512: Call the block drug combination in the interactive distribution classification index group, obtain the performance intensity item and the combination coordinate component, calculate the performance intensity difference between all combinations, calculate the Euclidean distance between two combinations based on the combination coordinate component, obtain the response offset between combinations, filter the combination with the response offset less than the preset threshold, and obtain the stable set of combination distribution in the block.

[0033] S513: Based on the stable set of combined distributions within the block, determine the consistency between the block's affiliation and the standard rule risk segment, filter combinations that simultaneously meet the segment and distribution conditions, and adjust the corresponding label number and group number to obtain risk type identification data.

[0034] The present invention is improved in that the temporal overlap refers to the phenomenon that the occurrence time of a certain medication behavior and the corresponding physiological monitoring data collection time coincide on the time axis, and the monitoring interval refers to the selected time range of physiological data collection related to a certain medication behavior or a group of medication behaviors.

[0035] A drug interaction information analysis system, the system comprising:

[0036] The time-series integration module is based on the medical prescription entry system. It analyzes the patient ID in the patient's medication records, determines whether the administration time overlaps with the monitoring time, screens records within the monitoring interval, determines the event relationship, and reconstructs the data set to obtain the medication behavior sequence feature set.

[0037] The response normalization module calculates the parameter change amplitude of adjacent monitoring records based on the medication behavior sequence feature group, compares the change direction before and after medication, analyzes the correlation between parameter change amplitude and dose, adjusts all parameter change amplitudes to a uniform scale and rearranges them to obtain the dose response change sequence.

[0038] Based on the dose response change sequence, the trend recognition module compares the change trends of adjacent response sequences, analyzes the trend maintenance, determines whether the response data after adjacent drug administration are continuously the same, screens drug combinations with continuous trends, and obtains a set of drugs with synergistic offset.

[0039] Based on the synergistic offset drug set, the interaction analysis module calculates the combined interaction performance characteristics, analyzes drug response and dosing interval, adjusts the performance content to a uniform structure, compares the distribution of all combined interaction performances, and obtains the drug interaction feature set.

[0040] Based on the drug interaction feature set, the risk labeling module determines the status of distribution segments, compares the interaction performance of segments, analyzes the risk characteristics of drug combination distribution, screens combinations with the same features, adjusts the risk label classification, and obtains risk type label data.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0042] In this invention, a time-series mapping relationship is constructed between continuous medication data and physiological monitoring data to dynamically link patient medication behavior with physiological response. Parameter change trends, response amplitudes, and combination distributions are synchronously collected and sorted. Risk features are extracted based on the synergistic offset characteristics between multiple drug combinations, enriching the data's hierarchical expressiveness. This enables multi-level interactive performance extraction and partitioning classification, enhancing the accuracy of risk type classification and achieving continuous tracking and timely warning of interactive risks in the personalized medication process. Attached Figure Description

[0043] Figure 1 This is a flowchart of the main steps of the present invention;

[0044] Figure 2 This is a flowchart illustrating the process of obtaining the drug use behavior sequence feature set in this invention.

[0045] Figure 3 This is a flowchart illustrating the process of obtaining the dose response change sequence in this invention.

[0046] Figure 4 This is a flowchart illustrating the process of obtaining the synergistic offset drug set in this invention;

[0047] Figure 5 This is a flowchart illustrating the process of obtaining the drug interaction feature set in this invention.

[0048] Figure 6 This is a flowchart illustrating the process of obtaining risk type identification data in this invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0050] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0051] All user-related information involved in this invention (including but not limited to biometric information, identity verification information, behavioral data, device information, and other data that can be used for identity verification and personalized services) is collected and processed with the user's full knowledge and voluntary consent. The use of data is limited to purposes necessary for providing the technical services of this invention, and reasonable technical and management measures will be taken to ensure the security and confidentiality of users' personal information in terms of information protection and privacy.

[0052] Example

[0053] Please see Figure 1 This invention provides a technical solution, a method for analyzing drug interaction information, comprising the following steps:

[0054] S1: Based on the medical prescription entry system, analyze the patient number corresponding to the patient's medication record, determine whether there is a temporal overlap between the medication time and the physiological monitoring time, screen all records within the same monitoring interval, determine the correspondence between medication events and monitoring events, integrate the physiological monitoring content before and after medication behavior, reconstruct the monitoring data group according to the monitoring occurrence sequence, and obtain the medication behavior sequence feature group.

[0055] S2: Based on the feature group of medication behavior sequence, calculate the parameter change amplitude between adjacent monitoring records, compare the parameter change direction before and after medication, analyze the correlation between parameter amplitude change and drug dosage record, adjust all change amplitude content to a uniform scale, rearrange the drug response content according to the drug administration time sequence, and obtain the dose response change sequence.

[0056] S3: Based on the dose response change sequence, compare the change trends of adjacent drug response sequences, analyze the maintenance of continuous change trends in the time series, determine whether the physiological response data after adjacent drug administration continue to show the same change direction, screen drug combinations with continuous change trends, and determine the combination relationship according to the order of administration to obtain the synergistic offset drug set.

[0057] S4: Based on the synergistic offset drug set, calculate the interaction performance characteristics of each drug combination, analyze the physiological response data and dosing interval information of each drug in the combination, adjust the performance content of each combination to a unified comparison structure, compare the distribution of the interaction performance of all combinations, and obtain the drug interaction feature set.

[0058] S5: Based on the drug interaction feature set, determine the segment status of each drug combination distribution, compare the interaction performance of each segment, analyze the risk characteristics of the drug combination distribution under the standard rules, screen all drug combinations under the same distribution characteristics, adjust the risk label classification method of each group, and obtain risk type label data.

[0059] The medication behavior sequence feature set includes time tags, drug identifiers, and data association indexes; the dose response change sequence includes response change items, dose reference items, and sequence sorting labels; the co-offset drug set includes offset drug items, combination numbers, and trend indicators; the drug interaction feature set includes combination coordinates, performance intensity items, and block indexes; and the risk type identification data includes risk level, label number, and grouping results.

[0060] In S1, patient medication records refer to the actual medication information of patients registered in the hospital information system, including drug name, dosage, medication time, patient number, etc.; physiological monitoring time refers to the specific time point when relevant vital signs of patients (such as blood pressure, heart rate, liver function, etc.) are monitored and data are collected, usually from automatic records of physiological monitoring equipment; temporal overlap refers to the phenomenon that the time of occurrence of a certain medication behavior and the corresponding physiological monitoring data collection time coincide or are closely adjacent on the time axis, ensuring the temporal relevance of data pairing; monitoring interval refers to the selected time range for collecting physiological data related to a certain medication behavior (such as monitoring data within a certain time period before and after medication); medication event refers to a specific drug administration behavior, including information such as drug, dosage, administration method and specific time, which is the unit of medication record; monitoring event refers to a specific physiological parameter collection behavior, including the collection time point, parameter type (such as blood pressure, ALT, heart rate, etc.) and test results; physiological monitoring content refers to all physiological parameter data (such as continuous blood pressure records, changes in liver function indicators, etc.) collected within a specific monitoring interval and associated with medication behavior.

[0061] In S2, the parameter change range refers to the numerical change of the same physiological parameter (such as blood pressure, ALT, etc.) between two time points before and after medication, i.e., the change "before"; the parameter change direction refers to whether the parameter change range is positive or negative, reflecting whether the physiological indicator increases or decreases before and after medication; correlation refers to whether there is a numerical correlation between the parameter change range and the medication dosage, such as whether a large dose of medication corresponds to a larger change in the physiological indicator; uniform scale refers to standardizing the change ranges of different drugs and different parameters so that the effects of each drug can be compared horizontally; drug response content refers to the physiological parameter changes exhibited by the patient after medication at a specific time point and dosage.

[0062] In S3, adjacent drug response sequences refer to two or more drug administrations and their corresponding physiological response data arranged in chronological order, facilitating the observation of sequence changes; maintenance within a time series refers to multiple consecutive drug administrations and physiological parameter responses on the time axis, observing whether their changing trends are continuous and consistent; physiological response data refers to vital signs or biochemical indicators collected from patients after each drug administration, serving as an objective reflection of the drug's effect; same direction of change refers to physiological parameters showing a consistent upward or downward trend after two or more consecutive drug administrations; continuous trend of change refers to physiological parameters showing a continuous, identical, or consistent state of change in multiple drug administration-response event sequences; and combination relationship refers to identifying which drugs work together to create the same physiological trend by analyzing the sequential administration of drugs and their corresponding response relationships.

[0063] In S4, the interaction performance characteristics refer to the specific ways in which the drug combination affects the patient's physiological response, such as the magnitude and type of parameter changes, reflecting the actual effect of the interaction between drugs; the unified comparison structure refers to adjusting the response characteristics of different drug combinations to a format or standard that can be directly compared horizontally, so that the interaction effects of different combinations are comparable; the distribution refers to the overall arrangement and classification status in the numerical space after statistically analyzing and sorting the interaction characteristics of all drug combinations.

[0064] In S5, "segment status" refers to the specific distribution of all drug combination interaction characteristics after they are classified into different risk intervals or categories according to preset standards; "interaction performance content" refers to the description of the interaction performance of each drug combination in the standardized matrix, including but not limited to its specific impact on physiological parameters; "standard rules" refers to the judgment criteria, classification basis, or expert consensus rules used to determine the risk level of drug combinations, which can be national standards, hospital-defined standards, etc.; "risk characteristics" refers to the labels or status of drug combinations in terms of medication safety and interaction risks determined through the above distribution and comparison analyses; "same distribution characteristics" refers to drug combinations having similar distribution characteristics in interaction performance, i.e., belonging to the same category, the same interval, or the same group; "classification method" refers to the standards and operating procedures for classifying drug combinations into different risk categories, i.e., how to map the results obtained from the analysis into risk labels.

[0065] Please see Figure 2 The specific steps for obtaining the feature set of medication behavior sequence are as follows:

[0066] S111: Based on the medical prescription entry system, analyze the medication records and monitoring data of physiological monitoring devices to determine whether the administration time in the medication records overlaps with the monitoring time in the monitoring data. By verifying the patient number associated with each medication record, screen the monitoring data within the corresponding time interval to obtain a time-series matching monitoring dataset.

[0067] The "Patient ID" and "Administration Time" fields are extracted. Using "Patient ID" as the key field, all monitoring records for the same patient are screened from the physiological monitoring data table. Each monitoring record includes "Monitoring Time," "Monitoring Item," and "Monitoring Value" fields. During the operation, a time window of 60 minutes before and after "Administration Time" is set as the time axis center. That is, the start time is "Administration Time minus 60 minutes," and the end time is "Administration Time plus 60 minutes," used to determine whether there is a "time overlap." The monitoring data of each patient is traversed one by one. In each loop, the time difference between "Monitoring Time" and "Administration Time" is calculated. For example, if the administration time of a record is 08:00 and the monitoring time is 07:10, the time difference is 50 minutes, which falls within the time window and is considered matching data. Conversely, if the monitoring time is 06:30... If the time difference is 0 and 90 minutes, it is excluded and not included in the matching set. The above judgment process is performed for each medication record, and all monitoring records that meet the conditions are collected into a group. The medication record ID is used as the primary key to form a one-to-many mapping structure with the matched monitoring records. For example, if patient C takes medication at 08:00 on June 1, and the monitoring data is recorded at 07:10, 07:45, 08:20, 09:00, and 10:30, the first four data points all fall within the set time window, thus forming the time-series matching monitoring dataset for this medication record. The record structure is as follows: medication record ID is RX1208, and the corresponding monitoring time points are [07:10, 07:45, 08:20, 09:00]. This process is repeated for all medication records of this patient, and the time-series matching monitoring dataset corresponding to all medication records is output.

[0068] S112: Based on the time-series matching monitoring dataset, analyze the monitoring time and patient number, screen all medication events of the same patient in overlapping time intervals, compare the time sequence of medication events, match the monitoring data before and after, and combine the associated monitoring content to obtain interval sequence pairing groups;

[0069] In each time-series matching monitoring dataset, the patient ID and monitoring time period are extracted. For example, if the monitoring time period is from 07:10 to 09:00, all medication events for that patient ID in the medication record table are retrieved again. The "administration time" field of each medication record is compared with the start and end times of the aforementioned monitoring time period to determine if there is any overlap or nesting relationship. That is, if a medication time falls between 07:00 and 09:00, it is considered that the medication event has an intersection with the monitoring time period and is retained. At the same time, the "administration time" and "administration dosage" of the medication event are recorded. All medication records that meet the conditions are sorted by time. The sorting method is to calculate all medication time values ​​in minutes and then sort them in ascending order. For example, if there are three medication time points at 06:50, 07:30, and 08:40, the sorted order will be 06:50, 07:30, and 08:40 respectively. The records are labeled as the 1st, 2nd, and 3rd medication administrations. After sorting, the original monitoring record list is returned to, and all monitoring records are sorted in ascending order by time. Then, each monitoring data point is determined to have occurred before or after a corresponding medication administration. The specific determination method is as follows: if a monitoring record time is less than a certain medication administration time, it is marked as "pre-monitoring"; if it is greater than that medication administration time but less than the next medication administration time, it is marked as "post-monitoring". For example, a monitoring record at 07:10 belongs to pre-monitoring if the medication administration time is 07:30, and belongs to post-monitoring if the monitoring time is 08:00. In this way, pre- and post-monitoring pairs are constructed for each medication administration, forming "interval sequence pairing groups". For example, if the medication event is 07:30, the pre-monitoring time is 07:10, and the post-monitoring time is 08:00, and the monitoring values, such as heart rates of 72 and 85 respectively, this constitutes a complete pairing.

[0070] S113: Based on interval sequence pairing groups, analyze each group of medication events and associated monitoring data, determine the order of monitoring data, configure time tags and drug identifiers according to monitoring time, and integrate associated parameter indexes to obtain medication behavior sequence feature groups;

[0071] Using the monitoring time field as the sorting criterion, and employing an ascending order, all monitoring records in the group are numbered sequentially according to their actual time. For example, two monitoring data points with times of 07:10 and 08:00 are labeled T1 and T2, respectively, and subsequent records are automatically numbered in chronological order. Next, the drug information is extracted from the corresponding medication records in this group, and the generic name or system code of the drug is used to construct a drug identification field. For example, "metoprolol" is represented by the system code MTL001. Then, the parameter name fields from the monitoring records, such as heart rate, systolic blood pressure, and diastolic blood pressure, are extracted, and a unified parameter index is established. For example, heart rate is set as HR01, systolic blood pressure as BP01, and diastolic blood pressure as BP02, etc. The correspondence between the measured value of each parameter and the time point is recorded. The resulting structure is: Medication ID + Time Tag + The structured data formed by combining drug identifier, parameter identifier, and parameter value, such as RX1208-T1-MTL001-HR01-72 and RX1208-T2-MTL001-HR01-85, is called the "medication behavior sequence feature group." In a practical example, if a patient takes medication at 07:30, and the heart rate is 72 at time T1 and 85 at time T2, and the blood pressure changes from 120 / 80 to 135 / 90, the sequence will record four data items: heart rate T1=72, heart rate T2=85, systolic blood pressure T1=120, systolic blood pressure T2=135, diastolic blood pressure T1=80, and diastolic blood pressure T2=90. This sequence structure comprehensively expresses the changes in monitoring values ​​before and after medication, while maintaining the integrity of the time sequence structure, and can be used for trend judgment and response analysis later.

[0072] Please see Figure 3 The specific steps for obtaining the dose response change sequence are as follows:

[0073] S211: Based on the feature group of medication behavior sequence, analyze the monitoring parameters corresponding to each group of medication behavior, compare the differences of parameters in the monitoring records before and after medication, and compile the changes of the same parameter at adjacent monitoring time points by comparing them one by one to obtain the parameter difference statistical group.

[0074] Extract the corresponding monitoring time points before and after each group of records. Align the data by parameter type at each time point to ensure that both sides belong to the same parameter dimension. Then, iterate through each monitoring parameter item, performing a difference calculation between the previous and subsequent monitoring values. The calculation method is to subtract the previous value from the subsequent value. For example, if the heart rate before medication was 72 beats / min and after medication it was 84 beats / min, the difference is 12 beats / min, which is the magnitude of the heart rate change. If the ALT before medication was 45 U / L and after medication it was 38 U / L, the change is -7 U / L, indicating a downward trend. After all parameter difference calculations are completed, record all parameter changes involved in the calculation for each medication event, labeling the corresponding parameter name, change value, and direction of change. The direction determination criteria are: a change value greater than 0 is defined as increasing, less than 0 is defined as decreasing, and equal to 0 is considered no change and not included in the change statistics. For each data set, all parameters are categorized into increasing and decreasing groups according to the above criteria. For example, if a data set shows a systolic blood pressure increase from 125 mmHg to 138 mmHg and a diastolic blood pressure increase from 82 mmHg to 86 mmHg, both are categorized into the increasing group. If the heart rate decreases from 90 to 85, it is categorized into the decreasing group. During the classification process, the change value of each parameter is recorded simultaneously for subsequent statistics. For example, the change range is set to four levels: 0-5, 6-15, 16-30, and 31 and above. Each parameter change value falls into the corresponding range and is marked with a change level number for unified classification and statistics. After all parameters are recorded, the changes in parameters involved in the medication event are compiled into a structured output. The structure includes parameter name, previous value, subsequent value, change value, change direction, and change level, forming a parameter difference statistical group.

[0075] S212: Based on the parameter difference statistical group, determine the change trend of each monitoring parameter before and after medication, assign directional labels to the increasing and decreasing parameter contents, compare the dose data, and organize the dose contents and change directions to obtain the dose direction pairing set;

[0076] Based on the recorded differences and direction indicators of each monitored parameter, the direction of change field of each record is called, and a judgment operation is performed. If the value is positive, it is assigned "increasing"; if it is negative, it is assigned "decreasing," representing increasing and decreasing trends, respectively. This is recorded in the direction indicator field. Next, the dose data recorded in each group of medication events is extracted, and the dose field is compared with the parameter change direction field. A combination operation is performed to pair each medication dose value with all the parameter direction indicators it causes to form a new structured item. For example, in a set of data, the dose is 50mg, and the corresponding parameters are heart rate (increasing), systolic blood pressure (increasing), and ALT (decreasing). This is recorded as three pairs of data: 50mg - heart rate increase, 50mg - systolic blood pressure increase, and 50mg - ALT decrease. To make the direction pairing content distinguishable, in the dose data... The dose intensity range is set at the level, divided into three levels: low, medium, and high based on the actual dose distribution. For example, 0-49mg is set as low dose, 50-99mg as medium dose, and 100mg and above as high dose. A dose level field is added to the paired data, that is, each pair of paired data records the actual dose value and its corresponding dose level. For example, heart rate increase -50mg - medium dose, systolic blood pressure increase -50mg - medium dose. If there is an event with a dose of 30mg that causes a decrease in ALT value of 6U / L, it is marked as ALT decrease -30mg - low dose. This lays the foundation for subsequent unified analysis of the correspondence between dose and response. After the above process is completed for all medication records, a dose direction pairing set is compiled. Each record in this set is a combination structure of a certain dose and the corresponding change direction of multiple parameters in a medication event.

[0077] S213: Based on the dose-direction pairing set, adjust all parameter changes to a unified standard format, arrange each group of dose and response data according to the time sequence of medication behavior, reconstruct it into a time-series data structure, and obtain the dose-response change sequence.

[0078] The parameter variation values ​​are converted to a uniform format. Unit and absolute differences between different parameter types are normalized. A proportional mapping method is used to convert the variation range of each parameter into a standard value range. For example, the upper and lower limits of each parameter variation value are set according to the maximum and minimum values ​​of its corresponding parameter in historical data. Through interval mapping, each variation value is compressed to a standard scale between 0 and 1. For instance, a heart rate change from 72 to 84, with a variation range of 12 beats / min, and assuming the maximum heart rate variation is 30 beats / min, then the standardized value is 12 / 30 = 0.4. This process is performed on all parameter variation ranges to form a uniform scale value. After processing, the dose values, direction of change, and standardized variation range in each data set are organized into a structured record. The system calls the medication time field recorded in the medication event, sorts all medication records in ascending order of timestamp, and reassembles the sorted dose-response pairs from earliest to latest. For example, the 06:30 event is 40mg - heart rate decrease of 0.2, the 08:00 event is 50mg - heart rate increase of 0.4, and the 10:15 event is 100mg - heart rate increase of 0.8. The record order is [40 - decrease of 0.2], [50 - increase of 0.4], [100 - increase of 0.8]. This order is integrated into a continuous data sequence, and the output structure is: time point, dose value, dose level, parameter name, direction of change, and standardized amplitude. The above process is repeated for all parameters and all medication records to reconstruct a time-series data structure set, forming a dose response change sequence.

[0079] Please see Figure 4 The specific steps for obtaining the synergistic offset drug set are as follows:

[0080] S311: Based on the dose response change sequence, analyze the change trend of response parameters corresponding to adjacent medication behaviors, compare the change direction of the same monitoring parameter under continuous medication behaviors, determine whether the trends at each time point are consistent, summarize all data segments with consistent trends, and obtain the trend continuation dataset.

[0081] Sort the medication events in ascending order by timestamp field to create a time series index. Then, for each adjacent medication behavior in the series, retrieve the corresponding standardized physiological parameter change records and perform item-by-item matching by parameter name to ensure that the comparison operation is performed on the same parameter type. For example, extract the direction of heart rate change in events A and B. If both events A and B show an increasing direction, they are considered to be in the same direction and assigned a value of "1". If the directions are opposite, a value of "0" is assigned. Perform the above judgment process in all adjacent medication events. For each pair of consecutive medication behaviors, iterate through all the same parameters and perform direction consistency judgment. For example, if events A and B both contain three parameters: heart rate, diastolic blood pressure, and ALT, compare the changes in the directions of these three parameters. If two or more of the three parameters are consistent, the medication behavior is considered to be consistent in trend. The trend consistency threshold is set at 66%, which means that the number of trend-consistent items in the same parameter sequence divided by the total number of parameters is ≥0.66. In practice, if four of the five parameters of a pair of consecutive events are consistent in direction, the consistency ratio is 0.8, which meets the threshold standard and is recorded as a trend-maintaining segment. The start and end events of this segment are marked as the start and end points of a trend segment. The same operation is continued to form multiple trend segments. Each segment contains data structures such as the numbers of two or more consecutive events, parameter consistency records, and start and end timestamps. All trend-consistent event segments are output in structured form to form a trend continuation dataset.

[0082] S312: Based on the trend continuation dataset, screen medication events with consistent change direction, determine the performance of each drug in response trend continuity, and group medication behaviors with continuous change characteristics according to trend consistency to obtain synergistic response drug groups;

[0083] Extract the medication event number and the name of the medication involved in each trend segment. Establish a one-to-one mapping table between medications and events for all medication events. Then, count the number of trend segments in which each medication appears in the mapping table. The judgment criterion is: if a medication appears in three or more trend segments and the direction of change is always consistent, it is considered stable in terms of trend continuity; conversely, if a medication's direction of change is inconsistent in a trend segment, it is recorded as unstable. The stability threshold is set to the proportion of events with consistent direction in a trend segment ≥ 70%, which serves as the grouping criterion. Screen all medication events that meet the stability requirement and group them by medication name, for example, aspirin-related events A1, A2, A3, A4, A5, A6, A7, A8, A9 ... 2. If A3 all appear in the increasing trend segment, they are grouped into the same response group. Furthermore, if propranolol appears in events B1, B2, and B3 in the decreasing trend segment, with two in the decreasing direction and one in the increasing direction, the consistency is only 66.7%, which does not meet the set threshold and is therefore removed from the group. Each response drug group includes the drug name, corresponding event number, trend direction label, and frequency of occurrence. The output structure is: {Drug name: Aspirin, Trend direction: Increasing, Event number: [A1, A2, A3]}, {Drug name: Losartan, Trend direction: Decreasing, Event number: [C4, C5, C6]}, etc., forming a collaborative response drug group.

[0084] S313: Based on the synergistic response drug grouping, determine the order of each drug and the order of administration, combine trend consistency information, reconstruct the drug combination relationship of each group, and obtain the synergistic offset drug set;

[0085] Extract the timestamp field of all events in each group, sort each event by time from earliest to latest, and record the order in which drugs appear in each group throughout the trend period. For example, aspirin appears at 08:00 in event A1, 10:00 in A2, and 12:00 in A3. After sorting, mark their usage order as 1st, 2nd, and 3rd for subsequent combination structure determination. Then, extract the drug name, dosage value, and response direction for each event in each group to construct a combination matrix. Each row in the matrix represents an event combination, and each column is the response direction label corresponding to the drug. Finally, perform a trend consistency field check to confirm all drugs in the group. If the direction remains consistent throughout the segment, the group of drugs is considered a synergistic response drug combination. Record information such as drug name, dosage, time sequence index, and direction marker. Finally, sort all combinations by the earliest event time in the combination and generate a structure list of synergistic offset drug sets. Example of recorded structure: Combination ID: G001, Members: [Aspirin - 08:00 - Rise, Losartan - 08:20 - Rise], Combination ID: G002, Members: [Benapril - 07:10 - Fall, Metoprolol - 07:50 - Fall]. Output the structure set of all drug combinations that meet the synergistic response characteristics.

[0086] Please see Figure 5 The specific steps for obtaining the drug interaction feature set are as follows:

[0087] S411: Based on the synergistic offset drug set, analyze the physiological response data of each drug and the combination number corresponding to the monitoring interval, sort the parameter change amplitude by combining drug information and monitoring time, determine the change type of each drug, and obtain the combination physiological response change sequence.

[0088] A bidirectional index mapping table between combination numbers and drug names is established. Then, all corresponding drug items are extracted for each combination number. The administration time field of each drug is called in the medication behavior sequence to determine the monitoring time interval before and after administration. Physiological response parameters recorded within this monitoring interval, such as heart rate, systolic blood pressure, diastolic blood pressure, ALT, and AST, are retrieved one by one. The values ​​at the corresponding monitoring time points are extracted, and the difference between the values ​​before and after medication is used as the parameter change amplitude. The difference calculation method is set as "post-measured value minus pre-measured value". The change values ​​of each monitoring parameter corresponding to each drug item are calculated. After obtaining all change amplitudes, a drug-parameter association table is established according to parameter name. Then, the change values ​​of each parameter are sorted in descending order, using the absolute value as the primary key. For the same parameter, the larger the change amplitude, the higher the sorting position. The numerical interval division standard is set as: change amplitude... An absolute value less than 5 indicates a slight change, 5 to 15 indicates a moderate change, and greater than 15 indicates a severe change. After sorting, the magnitude of change for each parameter is graded and recorded as "L1" (slight), "L2" (moderate), and "L3" (severe). The parameter change grading caused by each drug in its combination is compared with the drug information to determine the type of change of each drug in a specific physiological response within the combination. The change type is determined by the direction of parameter increase or decrease and the change level. The rule is: a positive change value and a level of L3 are marked as "strong increase", a negative change value and a level of L3 are marked as "strong decrease", and other cases are assigned values ​​such as "moderate increase", "moderate decrease", "slight increase", and "slight decrease" according to the corresponding rules. All drug items are grouped according to the combination number. Within each group, the drug name, monitoring parameter, change value, change direction, change level, and change type label are recorded to generate structured sequence data. The output is a combined physiological response change sequence.

[0089] S412: Based on the combined physiological response change sequence, compare the change type of each group with the dosing time label, analyze the dosing interval of adjacent drug items, calculate the standardized results of change amplitude, change type and dosing interval under each combination number, and uniformly map them into a continuous interval structure to obtain the interaction performance feature vector set;

[0090] Extract all drug items under each combination number, call the dosing time label field, and sort the drug items in ascending order by time. Then, extract the dosing time values ​​of adjacent drug items within each group and perform a difference calculation, setting the calculation method to subtract the dosing time of the preceding item from the dosing time of the subsequent item, to obtain the dosing interval between drug items, recorded in minutes. This value is used to reflect the tightness of the dosing time in the drug combination. After obtaining the interval time, combine the change amplitude value and change type of each drug item, and perform a unified scaling operation on the three data. Set the scaling range to 0 to 1, and set the baseline maximum value of each parameter as the standardization upper limit. The baseline value settings are as follows: the maximum parameter change amplitude is set to 30, the maximum change type level is set to 3 (strong increase / strong decrease is 3, moderate increase / moderate decrease is 2, slight increase / slight decrease is 1), and the upper limit of the dosing interval is set to 1. For 80 minutes, all values ​​are standardized by dividing the current value by the upper limit. For example, if the variation of a drug is 12, its standardized range is 12 ÷ 30 = 0.4. If the variation type is medium to high level 2, the standardized value is 2 ÷ 3 ≈ 0.67. If the dosing interval is 45 minutes, the standardized value is 45 ÷ 180 = 0.25. After completing the standardization of the three indicators for all drug items, the standardized value set of each drug item in each group is merged into a continuous interval structure to form a vector representation. Each vector structure contains parameter name, standardized variation range, standardized type level, and standardized dosing interval. Examples of structures include: [BP systolic blood pressure, 0.4, 0.67, 0.25], [HR, 0.2, 0.33, 0.5], etc. The vectors of all drug items in each group are numbered and summarized, and the output is an interactive performance feature vector set.

[0091] S413: Based on the interaction performance feature vector set, calculate the interaction performance vector for each combination number, analyze the difference magnitude between the vector and the mean, sum the differences of each item, and jointly judge the consistency of the dosing interval, the dispersion of response change magnitude, and the change type under the same combination, using the formula:

[0092] ;

[0093] The intensity index of distribution differences for each combination is obtained to obtain the drug interaction feature set, where, Indicates the first Indicators of the intensity of distributional differences among drug combinations. Indicates the first The number of normalized interaction vectors for drug combination groups. Indicates the first The first normalized interaction performance vector of drug combination item, Indicates the first The mean of the normalized interaction performance vector of the drug combination groups. Indicates the first Dosing interval index of drug combination Indicates the first The dispersion of the response variation of the drug combination. Indicates the first Consistency index of changes in drug combination types It is used to traverse the first... The indices of all interaction vector entries under the drug combination, with values ​​ranging from 1 to... , and This is an adjustment coefficient used to address the issue of inconsistent dimensions. It is the weight of the dosing interval index. These are the difference weights for controlling the dispersion of response variation and the consistency index;

[0094] The intensity index of distribution difference is for the first A quantitative characterization of the discreteness and distribution fluctuation intensity of the interaction characteristics of drug combinations in a standardized multidimensional vector space; molecular part (root mean square deviation): measures the... The overall deviation of each item in the set of interaction performance feature vectors from its mean describes the consistency or dispersion within this set of data. The denominator (normalization constraint) combines the dosing interval index, the dispersion of response change amplitude, and the consistency index of change type of the combination as a means of adjusting and normalizing the fluctuation of the numerator distribution, so that the index can reflect the comparability between different combinations under a unified comparison structure. The index reflects the fluctuation and consistency characteristics of the interaction effect distribution of a drug combination in multiple sets of drug response data. The larger the value, the more dispersed and more volatile the interaction performance characteristics of the drug combination are within the interval; the smaller the value, the more concentrated and consistent the performance is.

[0095] For combined numbering The three drug response data belonging to this combination were extracted sequentially: Item 1: blood pressure change of 14 mmHg, heart rate change of 6 bpm, and ALT change of 12 U / L; Item 2: blood pressure change of 11 mmHg, heart rate change of 9 bpm, and ALT change of 14 U / L; Item 3: blood pressure change of 13 mmHg, heart rate change of 5 bpm, and ALT change of 10 U / L. Max-min normalization was used to standardize all monitoring parameters, setting the maximum blood pressure change to 15 mmHg, heart rate to 10 bpm, and ALT to 20 U / L. The normalized results of the original data were: Item 1: 0.933, 0.6, 0.6; Item 2: 0.733, 0.9, 0.7; Item 3: 0.867, 0.5, 0.5. The average of the three normalized dimensions of each item was calculated to construct a normalized interaction performance vector scalar, which is as follows: , , The mean vector of this group is calculated as follows:

[0096] ;

[0097] Further calculation of the squared deviation of each vector from the mean is as follows:

[0098] ;

[0099] ;

[0100] ;

[0101] The numerator is calculated as follows:

[0102] ;

[0103] Let the normalized dosing interval index of this combination be... The dispersion of the response change amplitude is The consistency index of change type is , , Then the denominator is Substitute into the formula:

[0104] ;

[0105] Set distribution difference intensity index The complete range of values ​​is Based on the statistical analysis of the distribution patterns of historical drug combination interactions, this interval was divided into three continuous and non-overlapping segments:

[0106] like If the value is high, it is divided into a highly concentrated segment. The interaction vectors of drug combinations within this segment are closely distributed around the mean, indicating that the parameters within the combination fluctuate very little and the change types are highly consistent.

[0107] like If the deviation is within the range of moderate offset, it indicates that there is a certain degree of response difference within the combination, and the parameter changes and fluctuations begin to show an inconsistent trend.

[0108] like If the range is 0, it is divided into a highly discrete segment. This segment reflects the dramatic changes and significant fluctuations in the internal response of the drug combination, exhibiting significant parameter instability and type conflict.

[0109] The result of the current combination number Meet the conditions The result indicates that the drug combination falls into a highly concentrated segment, and its interaction performance is highly concentrated under a unified normalization structure. The internal parameter variation is low and the proportion of convergent response types is high. In the subsequent construction of the drug interaction feature set, this value is directly used to calibrate the low-offset classification of the combination in the matrix, and serves as the discrimination input for clustering or risk assessment in the feature set, thereby forming a clear horizontal difference division with other segment combinations.

[0110] Please see Figure 6 The specific steps for obtaining risk type identification data are as follows:

[0111] S511: Based on the drug interaction feature set, determine the position of each drug combination in the coordinate axis space, filter the block to which each drug combination belongs according to the block index, and classify the correspondence between drug combinations and blocks to obtain the interaction distribution classification index group.

[0112] The system retrieves the standardized dose value, standardized response amplitude, and standardized dosing interval recorded in each data set as coordinate inputs in three-dimensional space, assigning them values ​​to the X, Y, and Z axes respectively. This constructs the position points of the combination items in the spatial coordinate system. For example, if a combination item has a standardized dose of 0.6, a standardized response amplitude of 0.3, and a standardized dosing interval of 0.2, its coordinate position in three-dimensional space is (0.6, 0.3, 0.2). The same coordinate generation operation is performed on all combination items to batch construct a set of combination item points. Subsequently, the three-dimensional space is divided into several equally spaced blocks, with each axis direction segmented in units of 0.25. The resulting divisions are 0–0.25, 0.26–0.50, 0.51–0.75, and 0.76–1.00 segments. A block judgment operation is performed on the coordinate values ​​of each combination item, retrieving its X, Y, and Z values. The process involves determining the segment number each item falls into, and combining the segment numbers of the three axes to form the block index number of that combination. For example, if X falls into 0.5–0.75, it is number 3; if Y falls into 0–0.25, it is number 1; and if Z falls into 0.26–0.50, it is number 2. The combined block index is then “3-1-2”. Each block index is unique. The corresponding block index number is recorded for all combination items. A mapping table between combination numbers and block indices is then established, and all combination items falling into the same block index are grouped into one category, completing the block classification operation. In the output structure, each record contains the combination number, its position value in the coordinate axis space, the block number, and the number of the combination group to which it belongs. An example structure is as follows: Combination ID=G008, Position=(0.6, 0.15, 0.35), Block Index=3-1-2, Group Number=C03. This integration yields the interactive distribution classification index group.

[0113] S512: Call the block drug combinations within the interactive distribution classification index group, obtain the performance intensity item and combination coordinate components, calculate the performance intensity difference between all combinations, and calculate the Euclidean distance between two combinations based on the combination coordinate components, using the formula:

[0114] ;

[0115] Obtain the response offsets between combinations, filter combinations with similar response offsets to obtain a stable set of combination distributions within the block. Indicates drug combination Combined with drugs The response offset between Indicates drug combination The corresponding performance intensity item, Indicates drug combination The difference between the corresponding performance intensity items reflects the differences in interaction performance intensity between different combinations. Indicates drug combination In the Coordinate components in a combined coordinate dimension Indicates drug combination Coordinate components in the same dimension Indicates the number of dimensions in the combined coordinate space;

[0116] Response offset ( ) refers to: different drug combinations ( and In similar or adjacent coordinate spaces, the normalized difference in interaction intensity; response offset reflects the magnitude of change in the interaction characteristics of drug combinations, while also considering the relative distance between the two combinations in the multidimensional feature space (coordinate space); molecular part This represents the absolute difference in performance intensity between the two combinations; the denominator is... This indicates the similarity between two combinations in the multidimensional coordinate space. The closer the distance, the smaller the denominator, and the farther the distance, the larger the denominator. Overall, the larger the response offset value, the more significant the difference in the interaction performance intensity of drug combinations that are spatially close. The smaller the value, the closer the spatially adjacent combinations are in terms of performance intensity.

[0117] Obtain the performance intensity item corresponding to each combination. and and combined coordinate components and To unify the operational relationships of terms with different dimensions, linear normalization (Min-Max normalization) was applied to standardize the intensity and each coordinate component. The original intensity of combination z was 85, and the original intensity of combination j was 60. After normalization based on the global minimum of 50 and maximum of 100, the following values ​​were obtained: , The original coordinates of combination z are 120, 72, 38, and the coordinates of combination j are 140, 90, 40. The corresponding global minimum and maximum values ​​before normalization are (100, 200), (50, 100), and (30, 60), respectively. The coordinates after normalization are... , , , , , Substituting into the formula, we get:

[0118] ;

[0119] This result indicates that when drug combinations With combination Response offset When the value falls within the preset stable offset range This interval is defined as a "segment with similar responses and spatial proximity," used to identify pairs of combinations that are highly consistent in both interaction behavior and spatial distribution. If the calculation result falls within this interval... If it falls into the "moderate offset segment," it indicates that the combination has a moderate degree of deviation in response or space, while when When this occurs, it is classified as a "high offset segment," indicating a significant separation in the combination in terms of intensity or spatial coordinates. The results obtained in this study... If the value is within the stable offset range and meets the judgment condition of the current screening response offset, it will be included as a valid basis for the consistency of response trends between combination z and j, and will be included in the set of combination pairs in the stable set of combination distribution within the block.

[0120] S513: Based on the stable set of combined distribution within the block, determine the consistency between the block's affiliation and the risk segment of the standard rules, filter combinations that simultaneously meet the segment and distribution conditions, and adjust the corresponding label number and group number to obtain risk type identification data;

[0121] The process calls each block number and the total number of combinations within that block as input. It then determines whether each block meets the stability requirements in terms of distribution characteristics. The stability criteria are: the number of combinations within the block is no less than 5, and the standard deviation of the standardized response amplitude is less than 0.15. Specifically, the mean of the response amplitude field for all combinations within the block is calculated, then the sum of the squares of the differences between each item and the mean is divided by the number of combinations. The square root of the variance is then taken to obtain the standard deviation. If the result is not greater than 0.15, it is considered part of the response characteristic distribution set and is recorded as a stable block. All block numbers that meet this stability condition are recorded as a stable distribution block set. Next, the standard rule risk segment division table is called. This table defines several risk block number ranges. For example, the high-risk segment number range includes all combinations where the X-axis number is 4 and the Y-axis number is 3 or 4 in the block index. Specifically, the indices are "4-3-x" and "...". The process is as follows: 4-4-x, where x is an arbitrary number value. A comparison operation is performed, matching each block number in all interactive distribution classification index groups with the set of standard rule risk segment numbers. Combinations that simultaneously meet stability requirements and overlap with risk rules are selected and considered drug combinations falling into key high-risk interactive segments. The label and group numbers of these combinations are then adjusted. First, a new label number is assigned based on the risk segment they belong to; for example, a high-risk segment is uniformly assigned "R3", a medium-risk segment "R2", and a low-risk segment "R1". Then, group numbers are uniformly assigned to combinations within the same block; for example, all combinations with combination numbers "G005", "G006", and "G012" within block index "4-3-2" are adjusted to "Group_432". The output structure is: Combination ID, Original Label Number, New Label Number, Original Group Number, New Group Number, yielding risk type identification data.

[0122] A drug interaction information analysis system, the system comprising:

[0123] The time-series integration module is based on the medical prescription entry system. It analyzes the patient ID in the patient's medication records, determines whether the administration time overlaps with the monitoring time, screens records within the monitoring interval, determines the event relationship, and reconstructs the data set to obtain the medication behavior sequence feature set.

[0124] The response normalization module calculates the parameter change amplitude of adjacent monitoring records based on the feature group of medication behavior sequence, compares the change direction before and after medication, analyzes the correlation between parameter change amplitude and dose, adjusts all parameter change amplitudes to a uniform scale and rearranges them to obtain the dose response change sequence.

[0125] The trend identification module compares the trend of adjacent response sequences based on dose response change sequences, analyzes trend maintenance, determines whether the response data after adjacent drug administration are continuously the same, screens drug combinations with continuous trends, and obtains a set of drugs with synergistic shifts.

[0126] The interaction analysis module is based on the synergistic offset drug set, calculates the combined interaction performance characteristics, analyzes drug response and dosing interval, adjusts the performance content to a uniform structure, compares the distribution of all combined interaction performances, and obtains the drug interaction feature set.

[0127] The risk labeling module is based on the drug interaction feature set to determine the status of distribution segments, compare the interaction performance of segments, analyze the risk characteristics of drug combination distribution, screen combinations with the same features, adjust the risk label classification, and obtain risk type label data.

[0128] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for analyzing drug interaction information, characterized in that, Includes the following steps: S1: Based on the medical prescription entry system, analyze the patient ID of the patient's medication record, determine whether the administration time and monitoring time overlap, screen the records within the monitoring interval, determine the event relationship, and reconstruct the data group to obtain the medication behavior sequence feature group; S2: Based on the medication behavior sequence feature group, calculate the parameter change amplitude of adjacent monitoring records, compare the change direction before and after medication, analyze the correlation between parameter change amplitude and dose, adjust all parameter change amplitudes for normalization, and sort them according to numerical size or time order to obtain the dose response change sequence. S3: Based on the dose response change sequence, compare the change trends of adjacent response sequences, analyze the trend maintenance, determine whether the response data after adjacent drug administration are continuously the same, screen drug combinations with continuous trends, and obtain a set of drugs with synergistic offset. S4: Based on the synergistic offset drug set, calculate the combined interaction performance characteristics, analyze the drug response and dosing interval, adjust the performance content to a unified structure, compare the distribution of all combined interaction performances, and obtain the drug interaction feature set; S5: Based on the drug interaction feature set, determine the status of the distribution segment, compare the segment interaction performance, analyze the distribution risk characteristics of drug combinations, screen combinations with the same characteristics, adjust the risk label classification, and obtain risk type label data.

2. The drug interaction information analysis method according to claim 1, characterized in that, The medication behavior sequence feature set includes time tags, drug identifiers, and data association indexes; the dose response change sequence includes response change items, dose reference items, and sequence sorting labels; the co-offset drug set includes offset drug items, combination numbers, and trend indicators; the drug interaction feature set includes combination coordinates, performance intensity items, and block indexes; and the risk type identification data includes risk level, label number, and grouping results.

3. The drug interaction information analysis method according to claim 2, characterized in that, The specific steps for obtaining the medication behavior sequence feature set are as follows: S111: Based on the medical prescription entry system, analyze the medication records and monitoring data of physiological monitoring devices to determine whether the administration time in the medication records overlaps with the monitoring time in the monitoring data. By verifying the patient number associated with each medication record, screen the monitoring data within the corresponding time interval to obtain a time-series matching monitoring dataset. S112: Based on the time-series matching monitoring dataset, analyze the monitoring time and patient number, screen all medication events of the same patient in overlapping time intervals, compare the time sequence of medication events, match the monitoring data before and after, and combine the associated monitoring content to obtain interval sequence pairing groups; S113: Based on the interval sequence pairing group, analyze each group of medication events and associated monitoring data, determine the order of monitoring data, sort and configure time tags and drug identifiers according to monitoring time, and integrate the associated parameter index to obtain the medication behavior sequence feature group.

4. The drug interaction information analysis method according to claim 3, characterized in that, The specific steps for obtaining the dose response change sequence are as follows: S211: Based on the medication behavior sequence feature group, analyze the monitoring parameters corresponding to each group of medication behaviors, compare the differences of parameters in the monitoring records before and after medication, and by comparing the changes of the same parameter at adjacent monitoring time points, organize the changes of all monitoring parameters to obtain the parameter difference statistical group. S212: Based on the parameter difference statistics group, determine the change trend of each monitoring parameter before and after medication, assign directional labels to the increasing and decreasing parameter contents, compare the dose data, and organize the dose contents and change directions to obtain a dose direction pairing set. S213: Based on the dose direction pairing set, adjust all parameter changes to a unified standard format, arrange each group of dose and response data according to the time sequence of medication behavior, reconstruct it into a time-series data structure, and obtain the dose response change sequence.

5. The drug interaction information analysis method according to claim 4, characterized in that, The specific steps for obtaining the synergistic offset drug set are as follows: S311: Based on the dose response change sequence, analyze the change trend of response parameters corresponding to adjacent medication behaviors, compare the change direction of the same monitoring parameter under continuous medication behaviors, determine whether the trends at each time node are consistent, summarize all data segments with consistent trends, and obtain the trend continuation dataset. S312: Based on the trend continuation dataset, screen medication events with consistent change directions, determine the performance of each drug in response trend continuity, and group medication behaviors with continuous change characteristics according to trend consistency to obtain synergistic response drug groups; S313: Based on the synergistic response drug grouping, determine the order of drug arrangement and administration time, and combine trend consistency information to reconstruct the drug combination relationship of each group to obtain the synergistic offset drug set.

6. The drug interaction information analysis method according to claim 5, characterized in that, The specific steps for obtaining the drug interaction feature set are as follows: S411: Based on the aforementioned synergistic offset drug set, analyze the physiological response data of each drug and the combination number corresponding to the monitoring interval, sort the parameter change amplitude by combining drug information and monitoring time, determine the change type corresponding to each drug, and obtain the combined physiological response change sequence. S412: Based on the combined physiological response change sequence, compare the change type of each group with the drug administration time label, analyze the drug administration interval of adjacent drug items, calculate the standardized results of change amplitude, change type and drug administration interval under each combination number, and uniformly map them into a continuous interval structure to obtain the interactive performance feature vector set; S413: Based on the aforementioned interaction performance feature vector set, calculate the interaction performance vector for each combination number, analyze the difference magnitude between the vector and the mean, sum the differences, and jointly determine the consistency of the dosing interval, response change magnitude dispersion, and change type under the same combination, using the formula: ； The intensity index of distribution differences for each combination is obtained to obtain the drug interaction feature set, where, Indicates the first Indicators of the intensity of distributional differences among drug combinations. Indicates the first The number of normalized interaction vectors for drug combination groups. Indicates the first The first normalized interaction performance vector of drug combination item, Indicates the first The mean of the normalized interaction performance vector of the drug combination groups. Indicates the first Dosing interval index of drug combination Indicates the first The dispersion of the response variation of the drug combination. Indicates the first Consistency index of changes in drug combination types It is used to traverse the first... The indices of all interaction vector entries under the drug combination, with values ​​ranging from 1 to... , and This is an adjustment coefficient used to address the issue of inconsistent dimensions.

7. The drug interaction information analysis method according to claim 6, characterized in that, The specific steps for obtaining the risk type identification data are as follows: S511: Based on the drug interaction feature set, determine the position of each drug combination in the coordinate axis space, filter the block to which each drug combination belongs according to the block index, and classify the correspondence between drug combinations and blocks to obtain the interaction distribution classification index group. S512: Call the block drug combination in the interactive distribution classification index group, obtain the performance intensity item and the combination coordinate component, calculate the performance intensity difference between all combinations, calculate the Euclidean distance between two combinations based on the combination coordinate component, obtain the response offset between combinations, filter the combination with the response offset less than the preset threshold, and obtain the stable set of combination distribution in the block. S513: Based on the stable set of combined distributions within the block, determine the consistency between the block's affiliation and the standard rule risk segment, filter combinations that simultaneously meet the segment and distribution conditions, and adjust the corresponding label number and group number to obtain risk type identification data.

8. The method for analyzing drug interaction information according to claim 7, characterized in that, The time overlap refers to the phenomenon that the time of occurrence of a certain medication behavior coincides with the time of collection of corresponding physiological monitoring data on the time axis. The monitoring interval refers to the selected time range for collecting physiological data related to a certain medication behavior or a group of medication behaviors.

9. A drug interaction information analysis system, characterized in that, The system is used to implement the drug interaction information analysis method according to any one of claims 1-8, and the system comprises: The time-series integration module is based on the medical prescription entry system. It analyzes the patient ID in the patient's medication records, determines whether the administration time overlaps with the monitoring time, screens records within the monitoring interval, determines the event relationship, and reconstructs the data set to obtain the medication behavior sequence feature set. The response normalization module calculates the parameter change amplitude of adjacent monitoring records based on the medication behavior sequence feature group, compares the change direction before and after medication, analyzes the correlation between parameter change amplitude and dose, adjusts all parameter change amplitudes to a uniform scale and rearranges them to obtain the dose response change sequence. Based on the dose response change sequence, the trend recognition module compares the change trends of adjacent response sequences, analyzes the trend maintenance, determines whether the response data after adjacent drug administration are continuously the same, screens drug combinations with continuous trends, and obtains a set of drugs with synergistic offset. Based on the synergistic offset drug set, the interaction analysis module calculates the combined interaction performance characteristics, analyzes drug response and dosing interval, adjusts the performance content to a uniform structure, compares the distribution of all combined interaction performances, and obtains the drug interaction feature set. Based on the drug interaction feature set, the risk labeling module determines the status of distribution segments, compares the interaction performance of segments, analyzes the risk characteristics of drug combination distribution, screens combinations with the same features, adjusts the risk label classification, and obtains risk type label data.

Images