Cerebral ischemia drug treatment data sharing analysis method based on cloud platform

Abstract

The invention discloses a cerebral ischemia drug treatment data sharing analysis method based on a cloud platform, and relates to the technical field of data sharing analysis. The method comprises the following steps: S1, carrying out normalization processing on data, and establishing a data resource library; s2, extracting a parameter set and an evaluation result, and setting a parameter association mapping rule; s3, performing bidirectional analysis on the data to generate a three-dimensional correlation parameter set; s4, desensitizing the parameter set, and generating a hierarchical shared data pool; s5, performing quality verification, and adding a sharing analysis label to the data; and S6, sharing the analysis label, generating a dynamic analysis report, and storing the dynamic analysis report to a cloud platform database. According to the method, the multi-source treatment data of the cerebral ischemia patient and the clinical research annotation data are obtained and normalized, the unified cerebral ischemia drug treatment data resource library is established, and the problems that in data management, the formats of the multi-source data are incompatible, and storage is scattered are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data sharing analysis, in particular, to a brain ischemia drug treatment data sharing analysis method based on a cloud platform. BACKGROUND

[0002] In the digital era, the amount of data is growing explosively, the business scenario is becoming complex, and the computing power demand is changing dynamically. As a computing resource delivery and management mode based on the Internet, the cloud platform integrates servers, storage, networks and software resources to provide services for individuals, enterprises and institutions in a demand-oriented and elastic expansion manner, and becomes the core infrastructure for solving traditional IT pain points and driving digital transformation.

[0003] Brain ischemia is a cerebrovascular disease with high incidence, high disability rate and high recurrence rate. The drug treatment effect is affected by multiple factors such as patient's underlying disease, drug use plan, and efficacy monitoring method. In clinical practice and scientific research, there are always core pain points such as data island, fragmented analysis and insufficient sharing security. In this case, carrying out brain ischemia drug treatment data sharing analysis has important clinical practice value, scientific research innovation significance and public health value.

[0004] However, the existing brain ischemia drug treatment data sharing analysis method based on the cloud platform lacks unified security sharing rules when used. The patient's privacy information and the core parameters required by scientific research are not classified, direct sharing may cause privacy leakage risk, and excessive desensitization may lead to loss of key data information, affecting the analysis value. At the same time, the existing method lacks a standardized verification mechanism for data quality when used, and the integrity, consistency and timeliness of data from different sources are different, which directly leads to the reduction of the reliability of the subsequent analysis results, and it is difficult to support accurate drug treatment effect evaluation. For the problems in the related art, no effective solution has been proposed so far. SUMMARY

[0005] In view of the problems in the related art, the present application proposes a brain ischemia drug treatment data sharing analysis method based on a cloud platform to overcome the above technical problems existing in the prior art.

[0006] In order to achieve the above purpose, the specific technical scheme adopted by the present application is as follows: The brain ischemia drug treatment data sharing analysis method based on the cloud platform comprises the following steps: S1, normalize the multi-source treatment data of brain ischemia patients and clinical research annotation data, and build a brain ischemia drug treatment data resource library; S2, extract the core analysis dimension parameter set and the treatment effect evaluation result from the cerebral ischemia drug treatment data resource library, and set the parameter association mapping rule of the core analysis dimension parameter set and the treatment effect evaluation result; As a preferred solution, the step of extracting the core analysis dimension parameter set and the treatment effect evaluation result from the cerebral ischemia drug treatment data resource library, and setting the parameter association mapping rule of the core analysis dimension parameter set and the treatment effect evaluation result comprises the following steps: S21, determine the dimension classification direction of the core analysis dimension parameter set, wherein the dimension classification direction comprises the drug treatment dimension, the patient basis dimension and the efficacy monitoring dimension; S22, extract the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter in the core analysis dimension parameter set based on the drug treatment dimension, the patient basis dimension and the efficacy monitoring dimension; As a preferred solution, the step of extracting the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter in the core analysis dimension parameter set based on the drug treatment dimension, the patient basis dimension and the efficacy monitoring dimension comprises the following steps: S221, determine the treatment extraction range, the basis extraction range and the monitoring extraction range of the drug treatment dimension, the patient basis dimension and the efficacy monitoring dimension, wherein the treatment extraction range comprises the drug attribute parameter and the drug use scheme parameter, the basis extraction range comprises the patient physiological characteristic parameter and the patient history characteristic parameter, and the monitoring extraction range comprises the treatment process monitoring parameter and the treatment result determination parameter; S222, extract the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter from the core analysis dimension parameter set based on the treatment extraction range, the basis extraction range and the monitoring extraction range; S223, verify and output the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter.

[0007] S23, preset the effect evaluation influence rule, analyze the dimension influence parameter of the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter and the treatment effect evaluation result based on the effect evaluation influence rule; S24, set the parameter association mapping rule of the core analysis dimension parameter set and the treatment effect evaluation result according to the dimension influence parameter.

[0008] S3, based on the parameter association mapping rule and the multivariate linear regression model, bidirectional analysis is performed on the cerebral ischemia drug treatment data to generate a three-dimensional association parameter set; As a preferred solution, the step of bidirectional analysis of the cerebral ischemia drug treatment data based on the parameter association mapping rule and the multivariate linear regression model to generate a three-dimensional association parameter set comprises the following steps: S31, determine the direction classification corresponding relationship of the two-way analysis, wherein the direction classification includes the forward analysis direction and the reverse analysis direction; S32, based on the parameter correlation mapping rule, extract the analysis variable data required for the two-way analysis of the cerebral ischemia drug treatment data, and obtain the forward correlation coefficient and the reverse correlation coefficient; As a preferred solution, the parameter correlation mapping rule based on the parameter correlation mapping rule, the analysis variable data required for the two-way analysis of the cerebral ischemia drug treatment data is extracted, and the forward correlation coefficient and the reverse correlation coefficient are obtained. The following steps are included: S321, based on the forward analysis direction, extract the forward analysis independent variable and the forward analysis dependent variable from the cerebral ischemia drug treatment data; S322, based on the reverse analysis direction, extract the reverse analysis independent variable and the reverse analysis dependent variable from the cerebral ischemia drug treatment data; S323, preset the correlation coefficient calculation rule, and based on the parameter correlation mapping rule, the correlation between the forward analysis independent variable and the forward analysis dependent variable and the reverse analysis independent variable and the reverse analysis dependent variable is calculated, and the forward correlation coefficient and the reverse correlation coefficient are obtained.

[0009] S33, the forward correlation coefficient and the reverse correlation coefficient are brought into the multivariate linear regression model to calculate the three-dimensional correlation parameter set, and the three-dimensional correlation parameter set is verified and output.

[0010] As a preferred solution, the forward correlation coefficient and the reverse correlation coefficient are brought into the multivariate linear regression model to calculate the three-dimensional correlation parameter set, and the three-dimensional correlation parameter set is verified and output. The following steps are included: S331, set the input data format requirement of the multivariate linear regression model, align the forward correlation coefficient and the reverse correlation coefficient according to the drug parameter, the patient parameter and the efficacy correlation dimension, and construct the model input matrix; S332, the model input matrix is brought into the multivariate linear regression model to calculate the three-dimensional correlation parameter set containing the drug treatment dimension parameter value, the patient basic dimension parameter value and the comprehensive correlation coefficient; As a preferred solution, the model input matrix is brought into the multivariate linear regression model to calculate the three-dimensional correlation parameter set containing the drug treatment dimension parameter value, the patient basic dimension parameter value and the comprehensive correlation coefficient. The following steps are included: S3321, determine the core calculation dimension and the calculation dimension weight of the multivariate linear regression model; S3322, extract the forward correlation coefficient and the reverse correlation coefficient from the model input matrix, and calculate the synergistic influence value of the drug treatment dimension parameter and the patient basic dimension parameter combined with the calculation dimension weight; S3323, the synergistic influence value, the forward correlation coefficient and the reverse correlation coefficient are weighted and summed to obtain the comprehensive correlation coefficient; S3324, integrating the drug treatment dimension parameter value, the patient basis dimension parameter value and the comprehensive correlation coefficient to generate a structured three-dimensional correlation parameter set.

[0011] S333, presetting a parameter set verification rule, performing compliance verification on the three-dimensional correlation parameter set based on the parameter set verification rule, if the verification is passed, outputting in a standardized format, if the verification is not passed, reextracting the analysis variable data.

[0012] S4, presetting a data sharing security rule, performing desensitization processing on the three-dimensional correlation parameter set based on the data sharing security rule, and generating a hierarchical shared data pool; As a preferred solution, the preset data sharing security rule, based on the data sharing security rule, performs desensitization processing on the three-dimensional correlation parameter set, and generates a hierarchical shared data pool, comprising the following steps: S41, setting a field extraction rule, and extracting sensitive information fields and non-sensitive information fields in the three-dimensional correlation parameter set based on the field extraction rule; S42, presetting a data sharing security rule and a user hierarchical classification rule, and performing desensitization processing on the sensitive information fields in the three-dimensional correlation parameter set based on the data sharing security rule; S43, combining the desensitization-processed sensitive information fields and non-sensitive information fields based on the user hierarchical classification rule to generate a hierarchical shared data pool.

[0013] As a preferred solution, the combination of the desensitization-processed sensitive information fields and non-sensitive information fields based on the user hierarchical classification rule to generate a hierarchical shared data pool comprises the following steps: S431, specifying the user types in the user hierarchical classification rule, wherein the user types include scientific research users, clinical users and management users; S432, presetting data access permission rules for scientific research users, clinical users and management users respectively, wherein the data access permission rules include user field combination ranges and data use permissions; S433, combining the desensitization-processed sensitive information fields and non-sensitive information fields according to the user field combination ranges and data use permissions to form user type exclusive shared data sets; S434, storing the user type exclusive shared data sets in a standardized format to a cloud platform to generate a hierarchical shared data pool, and configuring hierarchical independent access interfaces for the hierarchical shared data pool.

[0014] S5, presetting a data quality evaluation rule, performing quality verification on the hierarchical shared data in the hierarchical shared data pool through the data quality evaluation rule, and adding shared analysis tags to the quality-verified hierarchical shared data; As a preferred solution, the preset data quality evaluation rule, quality checking of the hierarchical shared data in the hierarchical shared data pool by the data quality evaluation rule, and adding a shared analysis tag to the hierarchical shared data after quality checking include the following steps: S51, determine the data quality evaluation dimension rule, wherein the data quality evaluation dimension rule includes a data integrity rule, a data consistency rule and a data timeliness rule; S52, extract hierarchical shared data in the hierarchical shared data pool based on the data integrity rule, the data consistency rule and the data timeliness rule, wherein the hierarchical shared data includes an integrity check field, a consistency check field and a timeliness check field; S53, preset data quality evaluation rule, check the integrity check field, the consistency check field and the timeliness check field based on the data quality evaluation rule; S54, add a shared analysis tag according to the checked integrity check field, consistency check field and timeliness check field, and integrate and output.

[0015] S6, based on the shared analysis tag of the hierarchical shared data, generate a brain ischemia drug treatment effect dynamic analysis report, and store the brain ischemia drug treatment effect dynamic analysis report to a cloud platform database.

[0016] The beneficial effects of the present application are: 1, the present application acquires multi-source treatment data and clinical research annotation data of brain ischemia patients and performs normalization processing, builds a unified brain ischemia drug treatment data resource library, avoids the problem of incompatible multi-source data format and scattered storage in traditional data management, reduces the repeated workload of data cleaning and format conversion. At the same time, by determining the extraction range of the three dimensions of drug treatment, patient basis and efficacy monitoring, the core analysis parameters are highly matched with the clinical logic of brain ischemia drug treatment, invalid data interference is avoided, the efficiency of data preprocessing and parameter extraction is improved, and by constructing three-dimensional data quality evaluation rules, the data in the hierarchical shared data pool is checked, and the data that passes the check is added with a shared analysis tag, which effectively eliminates the problem of low-quality data mixing, avoids analysis deviation or application failure caused by data quality problems.

[0017] 2. This invention achieves bidirectional data analysis by pre-setting effect evaluation rules and combining them with a multivariate linear regression model. On the one hand, it clarifies the positive and negative analysis directions and variable correspondences, and accurately quantifies the correlation strength between core parameters and treatment effects through correlation coefficient calculation rules. On the other hand, by constructing a model input matrix and allocating calculation dimension weights, it transforms the positive and negative correlation coefficients into a three-dimensional correlation parameter set containing drug parameters, patient parameters, and comprehensive correlation coefficients. At the same time, it introduces a compliance verification mechanism to avoid analytical bias, improves the accuracy of the correlation between cerebral ischemia drug treatment data and efficacy results, and makes the analysis conclusions more clinically valuable and scientifically credible. Furthermore, by pre-setting data sharing security rules and user classification rules, it performs sensitive and non-sensitive field splitting and de-identification processing on the three-dimensional correlation parameter set, and then configures differentiated field combination ranges and usage permissions for three types of users: research, clinical, and management, forming a dedicated shared dataset with an independent access interface, avoiding the problem of insufficient sharing value caused by excessive restrictions. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a cloud-based method for sharing and analyzing data on cerebral ischemia drug treatment, according to an embodiment of the present invention. Detailed Implementation

[0020] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0021] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0022] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the cloud-based data sharing and analysis method for cerebral ischemia drug treatment according to an embodiment of the present invention includes the following steps: S1. Obtain multi-source treatment data and clinical research labeled data of patients with cerebral ischemia, normalize them, and build a data resource library of drug treatment for cerebral ischemia. Specifically, the treatment data such as patient electronic medical records, medication records, imaging examination results, and neurological function scores are collected from a clinical system, and the labeled data (including efficacy judgment criteria and drug dose gradient labeling) in the clinical research project are collected synchronously, and then normalization processing is carried out: unifying data formats (such as converting drug doses in different units to mg / time and unifying dates to a standard format), removing repeated and missing value exceeding threshold data through data cleaning, using standardization algorithms (such as Z-score standardization) to normalize different orders of magnitude parameters (such as age, liver and kidney function indicators) to a unified interval, quantifying and grading semantic ambiguous data (such as good efficacy) according to preset standards, and realizing semantic unification of data. Finally, a cloud architecture data resource library is built, the normalized data is stored according to the drug treatment dimension-patient basic dimension-therapeutic effect monitoring dimension, a data index and access interface is designed, and a data backup and security protection mechanism is configured.

[0023] S2, extracting a core analysis dimension parameter set and a treatment effect evaluation result from the cerebral ischemia drug treatment data resource library, and setting a parameter association mapping rule between the core analysis dimension parameter set and the treatment effect evaluation result; In the embodiment of the application, the step of extracting the core analysis dimension parameter set and the treatment effect evaluation result from the cerebral ischemia drug treatment data resource library, and setting the parameter association mapping rule between the core analysis dimension parameter set and the treatment effect evaluation result comprises the following steps: S21, determining the dimension classification direction of the core analysis dimension parameter set, wherein the dimension classification direction comprises a drug treatment dimension, a patient basic dimension, and a therapeutic effect monitoring dimension; Specifically, the dimension classification direction of the core analysis dimension parameter set is determined, the classification boundaries are determined focusing on the core logic and analysis requirements of cerebral ischemia drug treatment, focusing on the three key links of drug, patient, and efficacy, the drug treatment dimension focuses on the properties and use scheme of the drug, covering the drug generic name, dosage form, drug dose, drug frequency, treatment course, and types of combined drugs, and the core reflects the key elements of treatment intervention; the patient basic dimension focuses on individual difference influencing factors, including age, gender, physiological indicators (height, weight, liver and kidney function), underlying diseases (hypertension, diabetes, etc.), and medical history characteristics, reflecting the potential tolerance and response differences of patients to treatment; the therapeutic effect monitoring dimension focuses on the dynamic evaluation of treatment effect, including neurological function deficit scores before and after treatment, imaging lesion changes, adverse reaction occurrence, and stroke recurrence records, directly reflecting the actual effectiveness of treatment intervention. The three types of dimensions are independent of each other and closely related, and by determining the core coverage range and parameter attributes of each dimension, a logical and comprehensive classification system is formed.

[0024] S22, extracting the drug treatment dimension parameter, the patient basic dimension parameter, and the therapeutic effect monitoring dimension parameter in the core analysis dimension parameter set based on the drug treatment dimension, the patient basic dimension, and the therapeutic effect monitoring dimension. In the embodiments of the present application, the extraction of the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter in the core analysis dimension parameter set based on the drug treatment dimension, the patient basis dimension and the efficacy monitoring dimension includes the following steps: S221, determining the treatment extraction range, the basis extraction range and the monitoring extraction range of the drug treatment dimension, the patient basis dimension and the efficacy monitoring dimension, wherein the treatment extraction range includes the drug attribute parameter and the drug use scheme parameter, the basis extraction range includes the patient physiological characteristic parameter and the patient history characteristic parameter, and the monitoring extraction range includes the treatment process monitoring parameter and the treatment result determination parameter; Specifically, according to the logical refinement boundary of attribute-scheme, physiology-history, process-outcome, the treatment extraction range focuses on the core elements of drug intervention, the drug attribute parameter covers inherent characteristics such as drug generic name, dosage form, pharmacological classification and adverse reaction type; the drug use scheme parameter includes dynamic use information such as drug dosage, medication frequency, treatment duration, combination of combined medication and administration route. The basis extraction range focuses on individual differences of patients, the physiological characteristic parameter includes quantifiable physiological indicators such as age, gender, height, weight, liver and kidney function, blood pressure and blood glucose; the history characteristic parameter includes history of basic diseases such as hypertension and diabetes, number of stroke attacks, past medical history and allergy history.

[0025] The monitoring extraction range focuses on the whole process of efficacy evaluation, the treatment process monitoring parameter covers the neurological function score during treatment, vital sign changes, drug adverse reaction occurrence time and degree; the treatment result determination parameter includes core evaluation indexes such as post-treatment lesion imaging changes, final efficacy classification (significant improvement / moderate improvement / no improvement), recurrence, etc. By clearly defining the specific content of each sub-range, it is ensured that the extracted parameters are comprehensive and targeted, and accurate basis is provided for the construction of the core analysis dimension parameter set in the subsequent step.

[0026] S222, extracting the drug treatment dimension parameter, the patient basis dimension parameter and the efficacy monitoring dimension parameter from the core analysis dimension parameter set based on the treatment extraction range, the basis extraction range and the monitoring extraction range; Specifically, first, the extraction range is compared with the treatment, and the drug attribute related fields (such as drug generic name, dosage form, pharmacological classification) and use scheme fields (such as drug dosage, drug frequency, treatment course, combination of combined drug) are screened from the core analysis dimension parameter set to integrate into the drug treatment dimension parameter; second, according to the basic extraction range, the patient physiological characteristic fields (age, sex, height, weight, liver and kidney function indicators, etc.) and medical history characteristic fields (basic disease history, previous stroke history, allergy history, etc.) are extracted to form the patient basic dimension parameter; finally, according to the monitoring extraction range, the treatment process monitoring fields (NIHSS score, vital signs, adverse reaction records during treatment) and treatment result determination fields (changes in imaging lesions after treatment, efficacy classification, recurrence) are screened to form the efficacy monitoring dimension parameter, and during the extraction process, the field definitions of each range need to be strictly matched, redundant and invalid data are excluded, and format checking (such as unit unification, numerical specification) and integrity checking (core parameters without missing) are performed to ensure that the three types of parameters extracted accurately correspond to the dimension requirements, providing standardized data support for subsequent association analysis and efficacy evaluation.

[0027] S223, verifying and outputting the drug treatment dimension parameter, the patient basic dimension parameter and the efficacy monitoring dimension parameter.

[0028] Specifically, first, format checking is performed to unify the parameter format standard (such as drug dosage unit is mg, date is YYYY-MM-DD), ensure that the numerical type parameters (such as age, blood pressure) have no abnormal values, and the text type parameters (such as drug name, disease history) have standard semantics, second, integrity checking is carried out, the missing rate of core parameters (such as the missing rate of key indicators such as drug dosage, NIHSS score should be ≤5%) is calculated, the missing data is labeled with reasons and supplemented by interpolation method. Finally, logical consistency verification is implemented to check the correlation between parameters (such as combination of combined drugs needs to match the corresponding drug frequency, and efficacy classification needs to be consistent with lesion change trend), eliminate contradictory data, and after verification, integrate parameters according to the three-dimensional structure of drug treatment-patient basis-efficacy monitoring, generate standardized parameter set and output, and record verification report (including checking results, abnormal data processing method) at the same time, to provide high-quality and reliable parameter basis for subsequent association mapping rule setting.

[0029] S23, presetting effect evaluation influence rules, and analyzing the dimension influence parameters of the drug treatment dimension parameter, the patient basic dimension parameter and the efficacy monitoring dimension parameter and the treatment effect evaluation result based on the effect evaluation influence rules; Specifically, the rule framework is determined to establish a three-level influence model according to the drug intervention intensity-patient tolerance-therapeutic response, and the correlation threshold of each dimension parameter and the efficacy result is set (such as the expected range of NIHSS score improvement corresponding to the increase of 10 mg of drug dosage).

[0030] Based on the rule analysis dimension impact parameter, the drug treatment dimension parameter (dose, course of treatment, etc.) is correlated with the efficacy result, and the parameters with significant influence (such as the dose of thrombolytic drug and the 24-hour neurological function improvement rate) are screened out, and then the patient's basic dimension parameters (age, underlying disease, etc.) are analyzed. The adjustment effect of individual difference on the efficacy (such as the attenuation coefficient of the same dose of drug for elderly patients on the efficacy), combined with the efficacy monitoring dimension parameters (process indicators and result indicators), verify the consistency of dynamic monitoring data and the final efficacy (such as the prediction weight of 7-day NIHSS score change on 90-day prognosis).

[0031] Through multi-dimensional cross analysis, the influence strength of each parameter on the efficacy result (such as the standardized regression coefficient) is quantified, and the core dimension impact parameters such as drug dose, number of underlying diseases, and 24-hour lesion change after treatment are extracted.

[0032] S24, according to the dimension impact parameter, set the parameter correlation mapping rule of the core analysis dimension parameter set and the treatment effect evaluation result.

[0033] Specifically, based on the quantification results of the dimension impact parameters (such as the standardized regression coefficient of each parameter on the efficacy), the high impact parameters (such as the dose of thrombolytic drug and the 24-hour NIHSS score change after treatment) are classified according to the influence intensity and set as the first-level correlation item, the medium impact parameters (such as the course of treatment and the number of underlying diseases) are set as the second-level correlation item, and the low impact parameters (such as patient weight and drug frequency) are set as the third-level correlation item.

[0034] Set mapping conditions for each level of correlation item: the first-level correlation item adopts strong mapping rule (such as dose ≥ 50 mg and NIHSS score decrease ≥ 4 points, corresponding to significant effective efficacy); the second-level correlation item adopts conditional mapping rule (such as course of treatment ≥ 14 days and no history of diabetes, which improves the efficacy judgment level); the third-level correlation item adopts auxiliary mapping rule (such as abnormal body mass index, which modifies the efficacy evaluation result by ± 5%), and sets the interaction mapping logic between parameters (such as combined drug, the synergistic influence coefficient of dose and course of treatment), establishes a rule dynamic adjustment mechanism: regularly incorporate new dimension impact parameters (such as new biomarkers), optimize the mapping threshold through clinical data backtracking verification, and ensure that the rule covers the core correlation.

[0035] S3, based on the parameter correlation mapping rule and the multivariate linear regression model, bidirectional analysis is performed on the cerebral ischemia drug treatment data to generate a three-dimensional correlation parameter set; In the embodiment of the application, the bidirectional analysis of the cerebral ischemia drug treatment data based on the parameter correlation mapping rule and the multivariate linear regression model generates a three-dimensional correlation parameter set, which includes the following steps: S31, determine the direction classification and variable correspondence of bidirectional analysis, wherein the direction classification includes a forward analysis direction and a reverse analysis direction; Specifically, the direction classification and variable correspondence of bidirectional analysis are determined, and a clear analysis framework and variable mapping system are constructed around the core logic of parameter-efficacy and efficacy-parameter. The forward analysis direction focuses on the influence of therapeutic intervention-patient characteristics on efficacy, and the core goal is to identify key parameters driving treatment effect. The variable correspondence is that the drug treatment dimension parameters (dose, course of treatment, combination therapy, etc.) and the patient's basic dimension parameters (age, underlying disease, liver and kidney function, etc.) are used as independent variables, the treatment effect evaluation results (significant improvement, moderate improvement, no improvement) and the core indicators of efficacy monitoring dimension (such as NIHSS score after treatment, lesion shrinkage ratio) are used as dependent variables, and a mapping relationship of intervention variables-efficacy variables is formed.

[0036] The reverse analysis direction focuses on the key influencing parameters and optimization direction based on the efficacy difference, and the core goal is to locate the parameter adjustment space. The variable correspondence is that the treatment effect evaluation results (such as no improvement, recurrence, and other adverse outcomes) are used as independent variables, the drug treatment dimension parameters (dose rationality, drug regimen adaptability), patient basic dimension parameters (high-risk history, abnormal physiological indicators), and efficacy monitoring dimension dynamic parameters (adverse reactions during treatment, indicator change trend) are used as dependent variables, a mapping relationship of efficacy variables-intervention optimization variables is constructed, and the analysis objectives and variable correspondence rules of the two directions are determined to ensure that bidirectional analysis is independent and complementary.

[0037] S32, based on the parameter association mapping rule, extract the analysis variable data required for bidirectional analysis of cerebral ischemia drug treatment data, and obtain forward association coefficients and reverse association coefficients; In the embodiment of the application, based on the parameter association mapping rule, the analysis variable data required for bidirectional analysis of cerebral ischemia drug treatment data is extracted, and forward association coefficients and reverse association coefficients are obtained, including the following steps: S321, based on the forward analysis direction, extract forward analysis independent variables and forward analysis dependent variables from the cerebral ischemia drug treatment data; Specifically, the core goal of forward analysis is determined, and the driving factors of drug treatment and patient's basic characteristics on efficacy are identified, so the independent variables focus on the core parameters that can be intervened, and the dependent variables anchor the efficacy evaluation results.

[0038] When extracting independent variables for positive analysis, key parameters were screened from the perspective of drug treatment: including interventional indicators such as dosage (e.g., rt-PA 0.9 mg / kg), frequency of administration (once daily), duration of treatment (14 days), and combination therapy regimen (e.g., aspirin plus clopidogrel); and characteristic parameters that significantly affect efficacy were selected from the perspective of patient baseline: such as age (stratified ≥65 years / <65 years), underlying diseases (duration of hypertension / diabetes), baseline NIHSS score (≥10 points / <10 points), and liver and kidney function indicators (serum creatinine value), to ensure that the independent variables cover the core elements of treatment intervention and patient characteristics.

[0039] When extracting the dependent variable for positive analysis, focus on the core assessment results of the efficacy monitoring dimension, including hard indicators such as changes in NIHSS score at 7 / 90 days after treatment (e.g., a decrease of ≥4 points), reduction in brain imaging lesion volume (≥50%), efficacy grade (significant improvement / ineffectiveness), and incidence of adverse events (e.g., symptomatic bleeding), to ensure that the dependent variable can directly quantify the treatment effect.

[0040] During the extraction process, data type validation (numerical / categorical matching) and time node alignment (such as the correspondence between medication regimen and efficacy evaluation time window) are required to remove irrelevant or abnormal data, ultimately forming a standardized variable set of drug parameters, patient characteristics, and efficacy results.

[0041] S322. Extracting inverse analysis independent and inverse analysis dependent variables from cerebral ischemia drug treatment data based on the inverse analysis direction; Specifically, the goal of reverse analysis is to trace back from the efficacy results (especially poor or unsatisfactory efficacy) to explore the optimization potential in the drug treatment plan and the patient's basic characteristics. Therefore, the independent variable is anchored to the efficacy assessment results, and the dependent variable focuses on the core parameters that need to be investigated.

[0042] When extracting independent variables for reverse analysis, the core screening criterion is the evaluation results of treatment efficacy: priority is given to non-ideal outcomes such as no improvement in efficacy, relapse of the disease, or serious adverse reactions (such as symptomatic intracranial hemorrhage) as core independent variables. At the same time, moderate improvement and significant improvement outcomes are included as controls. By comparing different efficacy groups, key influencing factors are highlighted. Independent variables need to be quantitatively classified (e.g., no improvement = 0, moderate improvement = 1, significant improvement = 2) to ensure that they meet the data format requirements for subsequent analysis.

[0043] When extracting the dependent variable for inverse analysis, we focused on parameters that might lead to differences in efficacy: from the perspective of drug treatment, we screened intervention-related parameters such as dosage rationality (e.g., whether it is lower than the recommended dose), medication regimen suitability (e.g., whether there are contraindications to combination therapy), and treatment course integrity (whether the medication was stopped early); from the perspective of patient baseline, we selected individual-related parameters such as high-risk characteristics (e.g., advanced age ≥75 years, comorbid diabetic nephropathy, baseline NIHSS score ≥15 points) and abnormal values ​​of physiological indicators (e.g., the amplitude of blood pressure fluctuations during treatment); at the same time, we included dynamic abnormal indicators from the efficacy monitoring dimension (e.g., no decrease in NIHSS score after 3 days of treatment), forming a multi-dimensional dependent variable set.

[0044] During the extraction process, it is necessary to ensure the temporal correlation between efficacy results and parameters (such as medication records before adverse reactions occur) and data integrity (excluding samples with missing key parameters), ultimately forming a standardized variable set of efficacy outcome-intervention parameters.

[0045] S323. Preset the correlation coefficient calculation rules, and calculate the correlation between the positive analysis independent variable and the positive analysis dependent variable and the negative analysis independent variable and the negative analysis dependent variable based on the parameter correlation mapping rules to obtain the positive correlation coefficient and the negative correlation coefficient.

[0046] Specifically, the appropriate method should be selected based on the type of variable. If the independent and dependent variables are numerical (such as changes in drug dosage and NIHSS score), the Pearson correlation coefficient should be used to calculate the linear association strength. If they are categorical variables (such as efficacy grade and underlying disease type), the Spearman rank correlation coefficient or chi-square test should be used. At the same time, a threshold for determining the association strength should be set (such as |r|≥0.3 for moderate association and |r|≥0.5 for strong association), and the criteria for outlier removal should be clearly defined (such as data that deviate from the mean by 3 times the standard deviation).

[0047] When calculating the positive correlation coefficient, the drug / patient parameter-efficacy correlation hierarchy in the parameter correlation mapping rule is used as the basis: the independent variables (drug dosage, underlying disease, etc.) of the positive analysis are matched one by one with the dependent variables (efficacy score, lesion change, etc.), and the pairwise correlation values ​​are calculated by the preset correlation coefficient method. High correlation parameter pairs that meet the mapping rule are selected (such as the strong correlation coefficient between thrombolytic dose and 24-hour efficacy improvement), and the influence direction (positive correlation / negative correlation) of each coefficient is marked to form a positive correlation coefficient matrix.

[0048] In calculating the reverse correlation coefficient, according to the backtracking logic of the mapping rule of efficacy-parameter, the reverse analysis independent variable (efficacy outcome, such as no improvement, adverse reaction) is matched with the dependent variable (dose rationality, high-risk characteristics, etc.), and the correlation method suitable for classification variables (such as Logistic regression coefficient conversion) is used to quantify the correlation strength of different efficacy outcomes and each parameter, and the strong correlation coefficient (such as the negative correlation coefficient between not reaching the recommended dose and no improvement of efficacy) between non-ideal efficacy (such as recurrence) and the parameter is extracted, and a reverse correlation coefficient matrix is generated. In the calculation process, the consistency of the coefficient and the parameter association mapping rule needs to be checked (such as the strong correlation coefficient needs to match the first-level correlation item in the rule), and the variable extraction and calculation logic needs to be rechecked for contradictory results.

[0049] S33, the forward correlation coefficient and the reverse correlation coefficient are brought into a multivariate linear regression model to calculate a three-dimensional correlation parameter set, and the three-dimensional correlation parameter set is verified and output.

[0050] In the embodiments of the application, the step of bringing the forward correlation coefficient and the reverse correlation coefficient into a multivariate linear regression model to calculate a three-dimensional correlation parameter set, and verifying and outputting the three-dimensional correlation parameter set comprises the following steps: S331, set the input data format requirement of the multivariate linear regression model, align the forward correlation coefficient and the reverse correlation coefficient according to the drug parameter, the patient parameter and the efficacy correlation dimension, and construct the model input matrix; Specifically, the input format requirement is determined: the numerical type is unified to floating point type (retaining 4 decimal places), the missing values are filled with mean value (the missing rate of core parameters is less than or equal to 3%), and the abnormal values are removed by IQR method; in order to avoid the influence of dimension, the correlation coefficient is normalized (mapped to the interval [0, 1]), and each coefficient is labeled with the corresponding parameter name (such as drug dose-therapeutic correlation coefficient) and dimension label (drug / patient / efficacy), so as to ensure that the data is traceable.

[0051] In the data alignment stage, taking the three-dimensional framework of drug parameter-patient parameter-therapeutic correlation as the framework: first, classify the forward and reverse correlation coefficients according to the dimension label, and match the parameter names under the same dimension (such as the forward coefficient and the reverse coefficient corresponding to the drug dose are grouped together); then, check the consistency of the time dimension (such as the correlation coefficient of a drug in a certain course needs to correspond to the efficacy data in the same period), supplement the parameter attribute description (such as the combination of drugs needs to be labeled), and eliminate the cross-dimension data misplacement problem.

[0052] When constructing the model input matrix, the rows are samples and the columns are parameter-coefficient combinations: the row dimension is a complete treatment data sample for each cerebral ischemia patient, and the column dimension is arranged in the order of drug parameter correlation coefficient (positive plus negative) - patient parameter correlation coefficient (positive plus negative) - efficacy correlation coefficient (positive plus negative), such as rt-PA dose positive coefficient, age negative coefficient, NIHSS score efficacy correlation coefficient, etc.; the matrix cells are filled with the corresponding normalized correlation coefficients, and the column name, dimension label and data source (positive / negative analysis) are marked at the top of the matrix to form a structured input matrix.

[0053] S332. Substitute the model input matrix into the multivariate linear regression model to calculate the three-dimensional correlation parameter set, which includes the drug treatment dimension parameter value, the patient's basic dimension parameter value, and the comprehensive correlation coefficient. In this embodiment of the application, the step of substituting the model input matrix into a multivariate linear regression model to calculate a three-dimensional correlation parameter set containing drug treatment dimension parameter values, patient baseline dimension parameter values, and comprehensive correlation coefficients includes the following steps: S3321. Clarify the core calculation dimensions and weights of the calculation dimensions in the multivariate linear regression model; Specifically, the determination of core calculation dimensions is based on parameter correlation mapping rules and bidirectional correlation coefficient analysis: priority is given to including dimensions strongly correlated with efficacy, namely drug treatment dimensions (including parameters such as drug dosage, course of treatment, and combination therapy, corresponding to core intervention items with a positive correlation coefficient |r|≥0.5) and patient baseline dimensions (including parameters such as age, underlying diseases, and liver and kidney function, corresponding to high-risk items in the reverse correlation coefficient that are significantly associated with adverse efficacy). At the same time, efficacy-related dimensions (such as changes in NIHSS scores before and after treatment, lesion improvement rate, etc., as intermediate correlation items connecting the first two dimensions and efficacy results) are also included to form a three-dimensional calculation framework of drug-patient-efficacy, ensuring that the dimensions cover the core links of treatment intervention, individual differences, and efficacy response.

[0054] The setting of dimensional weights needs to be achieved through a multi-dimensional quantification method: First, based on the strength of the bidirectional correlation coefficient, positive correlation coefficients (e.g., r=0.6 for drug dosage and efficacy) and negative correlation coefficients (e.g., r=-0.4 for advanced age and no improvement in efficacy) are converted into initial weights (positive values ​​for positive correlations and absolute values ​​for negative correlations, normalized according to proportion); second, adjustments are made in conjunction with clinical expert consensus, increasing the weight proportion for dimensions with significant clinical significance but small absolute values ​​of coefficients (e.g., the time window for thrombolytic therapy); finally, optimization is achieved through cross-validation, incorporating the initial weights into the model test set, and adjusting the weights according to the goodness of fit (R²). 2 Adjust the weight allocation of each dimension (e.g., increase the weight of the drug dimension from 0.4 to 0.5 to improve the model's R-value). 2To 0.75 or more), ensuring that the weight meets the data correlation rules and conforms to the clinical actual logic. Finally, the weight proportion of each dimension is determined (such as 0.5 for the drug dimension, 0.3 for the patient dimension, and 0.2 for the efficacy correlation dimension), forming a standardized weight system that can be directly input into the model, providing a basis for subsequent synergistic effect value calculation.

[0055] S3322, extracting the positive correlation coefficient and the negative correlation coefficient from the model input matrix, and calculating the synergistic effect value of the drug treatment dimension parameter and the patient basic dimension parameter in combination with the calculation dimension weight; Specifically, the correlation coefficient is extracted: according to the parameter type-label of the input matrix column dimension-correlation direction (such as drug dose-positive age-negative), the correlation coefficients of the drug treatment dimension and the patient basic dimension are screened, the positive correlation coefficients (such as rt-PA dose positive coefficient r=0.62) of drug dose, treatment period, etc. are extracted for the drug dimension, the negative correlation coefficients (such as ≥75 years old negative coefficient |r|=0.48) of age, basic disease, etc. are extracted for the patient dimension, and the intermediate coefficients of the efficacy correlation dimension are excluded to ensure that the extraction object focuses on the two core interventions and individual dimensions.

[0056] Then, the weight matching is performed: the extracted correlation coefficient is matched with the preset calculation dimension weight-the drug treatment dimension weight (such as 0.5) matches all drug-related correlation coefficients, and the patient basic dimension weight (such as 0.3) matches all patient-related correlation coefficients. The single-dimension weighted coefficient is calculated by the coefficient x dimension weight (such as drug dose weighted coefficient=0.62x0.5=0.31, and elderly weighted coefficient=0.48x0.3=0.144), highlighting the influence priority of different dimensions.

[0057] Finally, the synergistic effect value is calculated: for each group of drug-patient parameter combination (such as rt-PA dose+age), the interaction correlation degree (such as r interaction =0.35) of the two types of weighted coefficients is calculated by Pearson correlation analysis first, and then the formula (drug weighted coefficient+patient weighted coefficient) x interaction correlation degree is used to calculate the synergistic value (such as (0.31+0.144) x 0.35≈0.159); after repeating the calculation for all parameter combinations, the mean or maximum value is taken as the final synergistic effect value, quantifying the combined effect strength of drug and patient dimension parameters.

[0058] S3323, the synergistic effect value is weighted and summed with the positive correlation coefficient and the negative correlation coefficient to obtain the comprehensive correlation coefficient; Specifically, a reasonable weight allocation scheme is preset: combined with the clinical significance and the pre-analysis conclusion, a higher weight (such as 40%) is given to the synergistic influence value, because it reflects the interaction between the drug and the patient parameters and is the core of multivariate analysis; the positive correlation coefficient weight is the second (such as 35%), focusing on the direct driving effect of therapeutic intervention on efficacy; the reverse correlation coefficient weight is slightly lower (such as 25%), focusing on the parameter influence of efficacy backtracking, and the sum of the weights of the three types is 100%, and can be fine-tuned according to the model fitting needs (such as when the clinical attention is more on the interaction effect, the synergistic influence value weight can be increased to 45%).

[0059] Subsequently, quantitative fusion calculation is carried out: first, the synergistic influence value, the positive correlation coefficient and the reverse correlation coefficient are normalized (mapped to the [0, 1] interval) to eliminate dimensional differences; then the weighted scores of the three types of indicators are calculated according to the preset weights, that is, synergistic influence value x synergistic weight + positive correlation coefficient x positive weight + reverse correlation coefficient x reverse weight. For example, the synergistic influence value of a group of parameters is 0.6, the positive correlation coefficient is 0.7, and the reverse correlation coefficient is 0.5, and the weight is calculated according to 40%, 35% and 25%, and the comprehensive correlation coefficient = 0.6x0.4 + 0.7x0.35 + 0.5x0.25 = 0.24 + 0.245 + 0.125 = 0.61.

[0060] Finally, the result is checked: on the one hand, it is checked whether the comprehensive correlation coefficient is in the reasonable interval [0, 1], and the abnormal value (such as the result exceeding 1 or being negative) is eliminated and the calculation process is adjusted; on the other hand, the parameter correlation mapping rule is checked to confirm that the size of the comprehensive correlation coefficient matches the correlation strength judgment standard (such as 0.61 corresponding to moderate to strong correlation), so as to ensure that the result not only conforms to the data logic, but also matches the clinical cognition of the parameter-therapeutic effect correlation.

[0061] S3324, integrate the drug treatment dimension parameter value, the patient basic dimension parameter value and the comprehensive correlation coefficient to generate a structured three-dimensional correlation parameter set.

[0062] Specifically, the integration framework is determined, taking the patient individual as the basic unit, an independent data entry is established for each patient to ensure that the parameter value and the comprehensive correlation coefficient correspond one by one, and then the data is filled according to the dimension classification: the specific parameter value is filled in the drug treatment dimension, such as the drug generic name, the drug dose (mg / time), the drug frequency and the treatment course; the age, gender, basic disease (such as hypertension / diabetes) and baseline physiological index (such as blood pressure, liver and kidney function value) are supplemented in the patient basic dimension; the corresponding calculation result is labeled in the comprehensive correlation coefficient dimension, and the correlation strength grade (such as 0.61-moderate to strong correlation) is also noted.

[0063] After the structured processing, the data is standardized in table or JSON format, each column is explicitly labeled with dimension category, parameter name and data unit, and each row corresponds to the complete data of one patient; data verification items are set to check the logical consistency of the same patient's drug parameter value, basic parameter value and comprehensive correlation coefficient (such as a reasonable correlation coefficient corresponding to a high dose of drug), eliminate contradictory data, and form a standardized three-dimensional correlation parameter set.

[0064] S333, preset parameter set verification rules, and the three-dimensional correlation parameter set is verified for compliance based on the parameter set verification rules, if the verification is passed, it is output in a standardized format, if the verification is not passed, the analysis variable data is extracted again.

[0065] Specifically, the preset verification rules are: in terms of integrity, the missing rate of drug treatment and patient basic dimension core parameters (such as dose, age, efficacy score) is required to be less than or equal to 3%; in terms of logic, the matching of parameter value and comprehensive correlation coefficient is checked (such as a positive correlation coefficient corresponding to a high dose of drug needs to be greater than or equal to 0.4), and there is no contradictory data (such as a correlation coefficient less than 0.3 corresponding to significant efficacy); in terms of standardization, the data format (such as unit, code) is ensured to meet the preset specification (such as date in YYYY-MM-DD format).

[0066] Based on the rule verification, the integrity and standardization problems are first screened by an automatic tool, and then the logical contradictory data is manually reviewed. If the verification is passed, it is output in a structured format of patient ID-drug parameter-patient parameter-comprehensive correlation coefficient, and a verification report is generated to mark the qualified items; if it is not passed, the problem source is located: if the variable extraction is not complete, the key parameters are supplemented from the multi-source data, if the correlation coefficient is abnormal due to calculation error, the correlation coefficient calculation link is corrected, after the variable data is extracted and analyzed again, the compliance verification is performed again, until the parameter set meets all the rule requirements, to ensure the accuracy and usability of the output data.

[0067] S4, preset data sharing security rules, and the three-dimensional correlation parameter set is desensitized based on the data sharing security rules, and a hierarchical shared data pool is generated; In the embodiment of the application, the preset data sharing security rules, based on the data sharing security rules, desensitize the three-dimensional correlation parameter set, and generate a hierarchical shared data pool, including the following steps: S41, set field extraction rules, and extract sensitive information fields and non-sensitive information fields in the three-dimensional correlation parameter set based on the field extraction rules; Specifically, the extraction rule is set: standards are formulated from three dimensions of data type, privacy risk and analysis necessity, sensitive information field focuses on identifiable patient identity or privacy disclosure content, and the rule is clear that the field contains personal identification (such as name, ID number, mobile phone number) and private medical records (such as detailed description of past medical history); the rule of non-sensitive information field is that it is only used for analysis and has no privacy disclosure risk, including desensitized drug parameters (such as dose, course of treatment, no patient identification), standardized efficacy data (such as NIHSS score, lesion change ratio), and anonymized patient characteristics (such as age stratification, basic disease classification code).

[0068] During extraction, the three-dimensional associated parameter set is scanned according to the rule first: sensitive fields are located by keyword matching (such as name and ID card), and identification information with fixed format (such as 18-digit ID number) is extracted by regular expression; non-sensitive fields are filtered according to the analysis purpose, and the core data supporting efficacy analysis in the dimensions of drug treatment, patient basis and comprehensive correlation coefficient (such as drug generic name, age interval, correlation strength value) are retained; after extraction, sensitive fields are stored in encrypted form (such as using AES algorithm), and non-sensitive fields are formatted according to analysis requirements, ensuring that the classification extraction meets the requirements of privacy protection and does not affect the subsequent data sharing and analysis use.

[0069] S42, preset data sharing security rules and user hierarchical classification rules, and based on the data sharing security rules, perform desensitization processing on the sensitive information fields in the three-dimensional associated parameter set; Specifically, two types of rules are preset first: the data sharing security rule clearly defines the sensitive information processing standard, such as personal identification (name, ID number) which needs to be completely desensitized, and private medical data (detailed medical history) which needs to be partially concealed; the user hierarchical classification rule divides users according to permissions, such as researchers who only access non-sensitive data, and administrators who can view complete data after desensitization.

[0070] Based on the security rule processing sensitive fields, a differentiated desensitization strategy is adopted: for identification information such as name and mobile phone number, replace it with an anonymous code (such as patient A); for ID number, keep the first 6 digits of administrative region code and the last 4 digits, and shield the middle part with *; for detailed medical history, simplify it to disease classification (such as hypertension level 3) through generalization, and delete the specific symptom description.

[0071] After desensitization, data usability needs to be verified to ensure that key analysis parameters (such as drug dose and efficacy score) are not affected, and desensitization logs are recorded to mark the processed fields, methods and time. Combined with the user hierarchical rule, configure corresponding data views for different users to achieve desensitization without losing use, which meets the requirements of privacy protection and guarantees the value of data sharing.

[0072] S43, based on the user hierarchical classification rule, combine the sensitive information fields and non-sensitive information fields after desensitization to generate hierarchical shared data pool.

[0073] In the embodiments of the present application, the combining of the sensitive information field and the non-sensitive information field after the desensitization processing based on the user hierarchical classification rule to generate the hierarchical shared data pool comprises the following steps: S431, determining the user type in the user hierarchical classification rule, wherein the user type comprises a scientific research user, a clinical user and a management user; S432, presetting data access permission rules for the scientific research user, the clinical user and the management user, wherein the data access permission rules comprise a user field combination range and a data use permission; Specifically, the user field combination range of the scientific research user is limited to non-sensitive analysis data, including anonymized drug treatment parameters (dose, course of treatment), patient basic characteristics (age stratification, basic disease code) and comprehensive correlation coefficients, and all personal identification fields are excluded; the data use permission only opens query and statistical analysis functions, prohibits downloading of original data, and requires signing of a data use agreement to ensure that it is used for scientific research and does not disclose privacy.

[0074] The user field combination range of the clinical user covers patient desensitization identification (such as visit ID), complete treatment parameters (including specific medication plan), real-time efficacy monitoring data (such as NIHSS score change), but hides core privacy fields such as ID card number and mobile phone number; the data use permission opens query and associated patient historical diagnosis and treatment record functions, allows export for clinical case analysis, but prohibits use for non-diagnosis and treatment purposes.

[0075] The user field combination range of the management user includes full-amount desensitization data (including information after desensitization of sensitive fields) and data quality statistical fields (such as parameter missing rate); the data use permission opens data monitoring and permission management functions, and views user access logs, but only for data management and security audit, strictly prohibits use for scientific research or clinical diagnosis and treatment, and ensures accurate matching of permissions and responsibilities.

[0076] S433, combining the sensitive information field and the non-sensitive information field after the desensitization processing according to the user field combination range and the data use permission to form a user type exclusive shared data set; Specifically, according to the preset user field combination range, corresponding fields are selected from the desensitized three-dimensional correlation parameter set: for the scientific research user, non-sensitive analysis fields are selected, such as anonymized drug parameters (dose, course of treatment), patient basic characteristics (age stratification, disease code) and comprehensive correlation coefficients, and all personal identification related fields are excluded; for the clinical user, diagnosis and treatment related fields containing desensitization identification (visit ID) are selected, including complete medication plan and efficacy monitoring data, and core privacy fields are hidden; for the management user, full-amount desensitization fields (including information after desensitization of sensitive fields) and data quality statistical fields are selected.

[0077] Then combine the data with the use permission limit field function: the scientific research user dataset only retains the fields required for query and statistical analysis, and shields the download-related data interface associated fields; the clinical user dataset retains the historical record associated fields required for diagnosis and treatment matching, and limits the export function of non-diagnosis and treatment fields; the management user dataset retains the permission management and log query associated fields, and shields the data modification related fields, and then encapsulates the dataset according to the user type, adopts the standardized format (such as CSV, JSON), marks the field purpose and use limit, and ensures that each exclusive dataset only contains the field combination within the user permission range.

[0078] S434, store the user type exclusive shared dataset in a standardized format to the cloud platform, generate a hierarchical shared data pool, and configure a hierarchical independent access interface for the hierarchical shared data pool.

[0079] Specifically, first, process each user type dataset in a standardized format: uniformly adopt CSV or Parquet format, standardize field naming (such as drug dose_mg and age stratification_code), data type (numeric / character type) and coding rules (such as disease code using ICD-10 standard), and ensure the consistency of different dataset formats for unified management of the cloud platform.

[0080] Subsequently, build a hierarchical shared data pool on the cloud platform: divide the data pool into three levels according to the user type, store the scientific research user exclusive dataset in the scientific research level pool, store the clinical user dataset in the clinical level pool, and store the management user dataset in the management level pool; configure independent storage partitions for each data pool, set up a data isolation mechanism (such as a virtual private cloud partition), prevent cross-access of different levels of data, and at the same time, enable the cloud platform data backup function to ensure data security.

[0081] Finally, configure an independent access interface for the hierarchical data pool: develop a dedicated API interface for each data pool, the interface parameters match the corresponding dataset fields (such as the scientific research level interface only opening non-sensitive field query parameters); set up interface access control (such as IP white list, Token authentication), the scientific research level interface only allows scientific research user IP to call, the clinical level interface needs to be bound to the clinical user diagnosis and treatment system account, and the management level interface needs to be authenticated by the administrator; at the same time, embed the data use log recording function in the interface to monitor the access behavior in real time, and ensure that the hierarchical data pool is safely and controllably accessed by the corresponding user.

[0082] S5, preset data quality evaluation rules, and quality check the hierarchical shared data in the hierarchical shared data pool through the data quality evaluation rules, and add shared analysis tags to the hierarchical shared data after quality check; In the embodiments of the present application, the preset data quality evaluation rule is used to perform quality checking on the hierarchical shared data in the hierarchical shared data pool through the data quality evaluation rule, and a shared analysis tag is added to the hierarchical shared data after quality checking, including the following steps: S51, determining the data quality evaluation dimension rule, wherein the data quality evaluation dimension rule includes data integrity rule, data consistency rule and data timeliness rule; Specifically, the data integrity rule focuses on data without missing, and clearly stipulates that the missing rate of core parameters (such as drug dose, NIHSS score, patient age) should be ≤3%, the missing rate of secondary parameters should be ≤5%, and missing data should be marked with reasons (such as patient refusal to provide test equipment failure) and the method of supplement should be unified (such as numerical type with mean filling, and division type with mode filling).

[0083] The data consistency rule ensures that the data is not contradictory, including format consistency (such as date unified as YYYY-MM-DD, drug dose unit unified as mg), logical consistency (such as significant efficacy corresponding to comprehensive correlation coefficient ≥0.5, combination drug corresponding to drug frequency), cross-data source consistency (such as the diagnosis and treatment data and efficacy monitoring data of the same patient need to be time-aligned and information-matched), and contradictory data needs to set priority determination standard (such as correcting analysis data according to clinical original record).

[0084] The data timeliness rule guarantees the timeliness of the data, stipulates that the interval between data collection and entry should be ≤24 hours (such as efficacy monitoring data should be entered within 1 day after detection is completed), the historical data update period should be ≤3 months (such as parameter standards based on new clinical guidelines need to be timely synchronized to the data set), and the data timestamp (collection time, update time) is marked, and the overdue data that is not updated needs to be marked as to be verified, to ensure that the data can reflect the current treatment and analysis requirements.

[0085] S52, extracting hierarchical shared data in the hierarchical shared data pool based on the data integrity rule, the data consistency rule and the data timeliness rule, wherein the hierarchical shared data includes integrity checking field, consistency checking field and timeliness checking field; Specifically, according to the data integrity rule, the integrity checking field is screened from the hierarchical shared data pool: focusing on core parameters (such as drug dose, patient age, efficacy score), extracting fields with missing risk (such as drug dose and baseline NIHSS score), marking field missing rate statistics, and ensuring that all required parameters required by the rule are covered.

[0086] Then according to the data consistency rule, the consistency verification field is screened: the format sensitive field (such as the drug use date needs to comply with the YYYY-MM-DD format, the drug unit needs to be unified as mg), the logical correlation field (such as the therapeutic effect classification needs to match the comprehensive correlation coefficient, the combined drug needs to correspond to the drug frequency), and the cross-data source correlation field (such as the patient's visit ID needs to be consistent in the diagnosis and treatment data and the therapeutic effect data), to ensure that the field can verify the format and logical consistency required by the rule.

[0087] Finally, according to the data timeliness rule, the timeliness verification field is screened: the field with timestamp (such as data collection time and therapeutic effect update time), the field that needs to be updated regularly (such as the clinical guideline version associated parameter), the statistical items such as the interval between field collection and entry, and the historical data update cycle are marked, to ensure that the timeliness judgment index required by the rule is covered. According to the different needs of research, clinical and management level data pool, the priority of each verification field is adjusted (such as the clinical pool focuses on timeliness field, and the research pool focuses on completeness and consistency field), to form a hierarchical quality verification field set.

[0088] S53, preset data quality evaluation rules, based on the data quality evaluation rules, the completeness verification field, the consistency verification field and the timeliness verification field are verified; Specifically, the completeness rule sets the core field missing rate threshold (such as ≤3% is qualified), the secondary field ≤5%, and the missing reason mark completeness ≥90%; The consistency rule specifies the format verification standard (such as the date format error rate ≤1%), the logical correlation error rate (such as the number of items that the therapeutic effect does not match the correlation coefficient ≤2%), and the cross-source data matching rate (such as ≥95%); The timeliness rule stipulates that the collection and entry delay ≤24 hours reaches the standard rate (≥98%), and the historical data update overdue items ≤3%.

[0089] During verification, the missing rate and reason mark completeness of the completeness field are counted and compared with the threshold to determine the eligibility; the error format of the consistency field is screened by the format verification tool, the associated contradictory items are checked by the logical algorithm, and the cross-source matching deviation is calculated; the timestamp of the timeliness field is extracted, and the delay time and overdue update proportion are counted. After verification, a quality report is generated, which marks the unqualified items of each field (such as the drug dose field with a missing rate of 7%), and divides the quality level according to the rule (excellent / good / poor), to provide accurate direction for data optimization.

[0090] S54, according to the completeness verification field, the consistency verification field and the timeliness verification field after verification, a shared analysis tag is added and integrated and output.

[0091] S6, based on the shared analysis tag of the hierarchical shared data, a brain ischemia drug treatment effect dynamic analysis report is generated, and the brain ischemia drug treatment effect dynamic analysis report is stored to the cloud platform database.

[0092] Specifically, based on shared analysis tags (such as drug efficacy comparison, patient stratification efficacy and time trend analysis), corresponding dimension data are selected from the hierarchical shared data pool: drug type-efficacy correlation coefficient data are extracted to support drug comparison analysis, age stratification-efficacy result data is selected for patient stratification analysis, and time stamp-efficacy change data is integrated to achieve trend tracking, ensuring that the data and tag analysis needs are accurately matched.

[0093] Next, a report is generated according to the dynamic analysis framework: it mainly uses visual charts (line charts to show the trend of efficacy over time, bar charts to compare the effects of different drugs, and heat maps to show the stratified efficacy of patients), combined with quantitative conclusions (such as drug A having a 12% higher efficacy improvement rate than drug B in 90 days), highlighting the dynamic changes in data; the report structure is divided into three parts: core conclusions, data support, and analysis methods, ensuring clear logic and intuitive conclusions.

[0094] Reports are stored in the cloud platform database: they are categorized and archived by report type, generation time, and corresponding data pool, stored in an encrypted format, and the original data index used in the report generation is synchronously associated to facilitate subsequent traceability and updates; database access permissions are set to allow only management users and authorized analysis users to view and update reports, ensuring report storage security and traceability, and providing dynamic data support for subsequent efficacy analysis and decision-making.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cloud-based method for sharing and analyzing data on drug therapy for cerebral ischemia, characterized in that, Includes the following steps: S1. Obtain multi-source treatment data and clinical research labeled data of patients with cerebral ischemia, normalize them, and build a data resource library of drug treatment for cerebral ischemia. S2. Extract the core analysis dimension parameter set and treatment effect evaluation results from the cerebral ischemia drug treatment data resource library, and set the parameter association mapping rules between the core analysis dimension parameter set and the treatment effect evaluation results; S3. Based on parameter correlation mapping rules and multivariate linear regression models, bidirectional analysis is performed on cerebral ischemia drug treatment data to generate a three-dimensional correlation parameter set. S4. Preset data sharing security rules, perform desensitization processing on the three-dimensional association parameter set based on the data sharing security rules, and generate a hierarchical shared data pool; S5. Preset data quality assessment rules, verify the quality of the hierarchical shared data in the hierarchical shared data pool through the data quality assessment rules, and add sharing analysis tags to the hierarchical shared data after quality verification. S6. Based on the shared analysis tags of hierarchical shared data, generate a dynamic analysis report on the therapeutic effect of cerebral ischemia drugs, and store the dynamic analysis report on the therapeutic effect of cerebral ischemia drugs in the cloud platform database.

2. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 1, characterized in that, The process of extracting the core analysis dimension parameter set and treatment effect evaluation results from the cerebral ischemia drug treatment data resource library, and setting parameter association mapping rules between the core analysis dimension parameter set and the treatment effect evaluation results includes the following steps: S21. Clarify the dimensional classification direction of the core analysis dimension parameter set, including the drug treatment dimension, patient baseline dimension, and efficacy monitoring dimension; S22. Extract the drug treatment dimension parameters, patient baseline dimension parameters, and efficacy monitoring dimension parameters from the core analysis dimension parameter set based on the drug treatment dimension, patient baseline dimension, and efficacy monitoring dimension. S23. Preset the effect evaluation influence rules, and analyze the influence parameters of drug treatment dimension parameters, patient baseline dimension parameters, and efficacy monitoring dimension parameters on the treatment effect evaluation results based on the effect evaluation influence rules; S24. Based on the parameter settings of the core analysis dimension parameter set, establish the parameter association mapping rules between the treatment effect evaluation results and the dimensional influence parameter settings.

3. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 1, characterized in that, The method for bidirectional analysis of cerebral ischemia drug treatment data based on parameter correlation mapping rules and multivariate linear regression models to generate a three-dimensional correlation parameter set includes the following steps: S31. Clarify the relationship between the directional classification of two-way analysis and the variables, where the directional classification includes forward analysis direction and reverse analysis direction; S32. Based on the parameter association mapping rule, extract the analytical variable data required for bidirectional analysis of cerebral ischemia drug treatment data, and obtain the positive correlation coefficient and the negative correlation coefficient. S33. Substitute the positive and negative correlation coefficients into the multivariate linear regression model to calculate the three-dimensional correlation parameter set, and verify and output the three-dimensional correlation parameter set.

4. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 1, characterized in that, The preset data sharing security rules, based on which the three-dimensional correlation parameter set is desensitized and a hierarchical shared data pool is generated, include the following steps: S41. Set field extraction rules, and extract sensitive information fields and non-sensitive information fields from the three-dimensional association parameter set based on the field extraction rules; S42. Preset data sharing security rules and user classification rules, and perform desensitization processing on sensitive information fields in the three-dimensional association parameter set based on the data sharing security rules; S43. Based on user classification rules, combine the desensitized sensitive information fields with non-sensitive information fields to generate a hierarchical shared data pool.

5. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 1, characterized in that, The preset data quality assessment rules involve verifying the quality of the hierarchical shared data in the hierarchical shared data pool using these rules, and then adding shared analysis tags to the verified hierarchical shared data. This process includes the following steps: S51. Define the rules for data quality assessment dimensions, which include data integrity rules, data consistency rules, and data timeliness rules. S52. Extract hierarchical shared data from the hierarchical shared data pool based on data integrity rules, data consistency rules, and data timeliness rules. The hierarchical shared data includes integrity verification fields, consistency verification fields, and timeliness verification fields. S53. Preset data quality assessment rules, and perform verification on integrity verification fields, consistency verification fields and timeliness verification fields based on the data quality assessment rules; S54. Add shared analysis tags to the verified integrity verification field, consistency verification field, and timeliness verification field, and then integrate and output them.

6. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 2, characterized in that, The extraction of core analysis dimension parameters, including drug treatment dimension parameters, patient baseline parameters, and efficacy monitoring parameters, from the core analysis dimension based on drug treatment, patient baseline, and efficacy monitoring dimensions includes the following steps: S221. Clarify the treatment extraction scope, basic extraction scope, and monitoring extraction scope for the drug treatment dimension, patient baseline dimension, and efficacy monitoring dimension. The treatment extraction scope includes drug attribute parameters and drug usage protocol parameters. The basic extraction scope includes patient physiological characteristic parameters and patient medical history characteristic parameters. The monitoring extraction scope includes treatment process monitoring parameters and treatment result judgment parameters. S222. Based on the treatment extraction range, the basic extraction range, and the monitoring extraction range, extract drug treatment dimension parameters, patient basic dimension parameters, and efficacy monitoring dimension parameters from the core analysis dimension parameter set. S223. Verify and output the parameters for drug treatment, patient baseline, and efficacy monitoring.

7. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 3, characterized in that, The step of extracting the analytical variable data required for bidirectional analysis of cerebral ischemia drug treatment data based on parameter association mapping rules, and obtaining the positive and negative correlation coefficients, includes the following steps: S321. Extracting positive analysis independent variables and positive analysis dependent variables from cerebral ischemia drug treatment data based on the positive analysis direction; S322. Extracting inverse analysis independent and inverse analysis dependent variables from cerebral ischemia drug treatment data based on the inverse analysis direction; S323. Preset the correlation coefficient calculation rules, and calculate the correlation between the positive analysis independent variable and the positive analysis dependent variable and the negative analysis independent variable and the negative analysis dependent variable based on the parameter correlation mapping rules to obtain the positive correlation coefficient and the negative correlation coefficient.

8. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 3, characterized in that, The process of inputting the positive and negative correlation coefficients into a multivariate linear regression model to calculate the three-dimensional correlation parameter set, and then validating and outputting the three-dimensional correlation parameter set, includes the following steps: S331. Set the input data format requirements for the multivariate linear regression model, align the positive and negative correlation coefficients according to the dimensions of drug parameters, patient parameters, and efficacy correlation, and construct the model input matrix. S332. Substitute the model input matrix into the multivariate linear regression model to calculate the three-dimensional correlation parameter set, which includes the drug treatment dimension parameter value, the patient's basic dimension parameter value, and the comprehensive correlation coefficient. S333, Preset parameter set verification rules: Perform compliance verification on the three-dimensional associated parameter set based on the parameter set verification rules. If the verification passes, output in a standardized format. If the verification fails, re-extract the analysis variable data.

9. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 4, characterized in that, The process of combining desensitized sensitive information fields with non-sensitive information fields based on user classification rules to generate a hierarchical shared data pool includes the following steps: S431. Clarify the user types in the user classification rules, including research users, clinical users, and management users; S432. Preset data access permission rules for research users, clinical users and management users respectively. The data access permission rules include the user field combination range and data usage permissions. S433. According to the user field combination range and data usage permissions, combine the desensitized sensitive information fields with non-sensitive information fields to form a user-type exclusive shared dataset. S434. Store the user-type-specific shared dataset in a standardized format to the cloud platform, generate a hierarchical shared data pool, and configure hierarchical independent access interfaces for the hierarchical shared data pool.

10. The method for sharing and analyzing data on cerebral ischemia drug therapy based on a cloud platform according to claim 8, characterized in that, The step of inputting the model input matrix into a multivariate linear regression model to calculate a three-dimensional correlation parameter set containing drug treatment dimension parameter values, patient baseline dimension parameter values, and comprehensive correlation coefficients includes the following steps: S3321. Clarify the core calculation dimensions and weights of the calculation dimensions in the multivariate linear regression model; S3322. Extract the positive correlation coefficient and the negative correlation coefficient from the model input matrix, and calculate the synergistic influence value between the drug treatment dimension parameters and the patient's basic dimension parameters by combining the calculation dimension weights. S3323. The synergistic impact value is weighted and summed with the positive correlation coefficient and the negative correlation coefficient to obtain the comprehensive correlation coefficient; S3324. Integrate drug treatment dimension parameter values, patient baseline dimension parameter values, and comprehensive correlation coefficients to generate a structured three-dimensional correlation parameter set.

Images