Medication regimen optimization system for patients with central nervous system infections based on big data analysis

Abstract

The present invention relates to the field of medical information technology, and discloses a drug administration regimen optimization system for patients with central nervous system infections based on big data analysis, wherein: the system includes: a data acquisition module, a feature fusion module, a feature optimization module, a prediction module, a simulation data generation module, a correction module, and a drug administration module, and weighted disease data of the diseased data is obtained by performing weight mapping on the diseased data; a reconstruction error function of the multi-source fusion feature is obtained by performing low-rank approximation reconstruction on the multi-source fusion feature, and a partial derivative of each parameter in the reconstruction error function is obtained to obtain a closed-form solution of the reconstruction error function, and the parameters in the multi-source fusion feature function are updated based on the closed-form solution to obtain an updated multi-source fusion feature; the present invention can improve the accuracy of generating drug administration regimens for patients with central nervous system infections.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to a drug administration regimen optimization system for patients with central nervous system infections based on big data analysis. Background Art

[0002] Clinical treatment of central nervous system infections currently relies primarily on empirical dosing regimens, which often follow recommendations from standardized guidelines. While these regimens cover common pathogens, they neglect individual patient characteristics. For example, they utilize limited data dimensions, such as pathogen type and site of infection, while ignoring key individual characteristics such as the patient's physiological state and drug sensitivity. This lack of adaptive learning results in a disconnect between the regimen and the patient's actual needs, leading to a lack of precision in the dosing regimen. Therefore, improving the accuracy of dosing regimens generated for patients with central nervous system infections has become an unresolved issue. Summary of the Invention

[0003] The present invention provides a drug administration regimen optimization system for patients with central nervous system infections based on big data analysis, the main purpose of which is to solve the problem of how to improve the accuracy of drug administration regimen generation for patients with central nervous system infections.

[0004] To achieve the above objectives, the present invention provides a system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis, which is characterized by comprising:

[0005] Data collection module: used to collect disease data of patients with central infection;

[0006] Feature fusion module: used to perform weight mapping on the disease data to obtain weighted disease data of the disease data, perform tensor product construction on the weighted disease data to obtain a multi-source fusion feature function of the weighted disease data, and the solution of the multi-source fusion feature function is a multi-source fusion feature;

[0007] Feature optimization module: used to perform low-rank approximate reconstruction on the multi-source fusion feature to obtain a reconstruction error function of the multi-source fusion feature, calculate the partial derivative of each parameter in the reconstruction error function to obtain a closed-form solution of the reconstruction error function, and update the parameters in the multi-source fusion feature function based on the closed-form solution to obtain an updated multi-source fusion feature;

[0008] Prediction module: used to perform feature mapping on the updated multi-source fusion features to obtain the predicted treatment results of the multi-source fusion features;

[0009] Simulation data generation module: used to structure and organize the past medication data of the central nervous system infection patient and the predicted treatment results to obtain multi-dimensional and multivariate response data;

[0010] Correction module: used for correcting contradictory data in the multi-dimensional multi-element reaction data based on medical common sense;

[0011] Drug administration module: used to generate a personalized drug administration plan for the central infection patient based on the predicted treatment results and the revised multi-dimensional multi-response data.

[0012] Optionally, the disease data includes: patient information and recent medication feedback.

[0013] Optionally, the calculation formula for the multi-source fusion feature is:

[0014] ;

[0015] in: It is a multi-source fusion feature. is the weighted disease data, is the total number of data in the disease data, is the dynamic weight vector, is the patient feature dimension matrix, is the data modality dimension matrix, is the time dimension matrix, For the The disease data, is the matrix multiplication along different dimensions.

[0016] Optionally, the calculation formula of the dynamic weight vector is:

[0017] ;

[0018] in: is the dynamic weight vector, For actual treatment results, For the The disease data and actual treatment results The mutual information between For all the disease data and the actual treatment results The sum of the mutual information of For the The disease data, For the The disease data.

[0019] Optionally, the calculation formula of the reconstruction error function is:

[0020] ;

[0021] in: It is a multi-source fusion feature. is the dynamic weight vector, is the Frobenius norm, is the reconstruction error function, for The disease data, is the matrix multiplication along different dimensions, is the projection matrix set, is the patient feature dimension matrix, is the data modality dimension matrix, is the time dimension matrix, For the The disease data, is the weighted disease data, is the total number of data of the disease data, is the reconstruction matrix of the projection matrix set, are the reconstruction matrices corresponding to the patient feature dimension matrix, data modality dimension matrix, and time dimension matrix, respectively. is the regularization coefficient.

[0022] Optionally, calculating the partial derivative of each parameter in the reconstruction error function includes:

[0023] Step 1: fixing the reconstruction matrix of the projection matrix set, calculating the closed form solution of the current projection matrix set, and updating the projection matrix set based on the closed form solution of the projection matrix set to obtain an updated projection matrix set;

[0024] Step 2: Update the multi-source fusion feature based on the updated projection matrix set to obtain an updated multi-source fusion feature, wherein the calculation formula of the updated multi-source fusion feature is:

[0025] ;

[0026] in: is the updated multi-source fusion feature, is the dynamic weight vector, is the weighted disease data, is the matrix multiplication along different dimensions, The updated patient feature dimension matrix, data modality dimension matrix and time dimension matrix;

[0027] Step 3: Fixing the updated projection matrix set, updating the reconstruction matrix of the projection matrix set, calculating the closed form solution of the reconstruction matrix of the projection matrix set, and updating the reconstruction matrix of the projection matrix set based on the closed form solution of the reconstruction matrix of the projection matrix set to obtain an updated reconstruction matrix of the projection matrix set;

[0028] Step 4: performing feature update on the reconstruction error function based on the reconstruction matrix of the updated projection matrix set to obtain an updated reconstruction error function.

[0029] Optionally, the calculation formula for predicting treatment results is:

[0030] ;

[0031] in: To predict treatment outcomes, is the activation function, is the weight training matrix, is the bias term, is the updated multi-source fusion feature.

[0032] Optionally, the past medication data of the central nervous system infection patient and the predicted treatment results are structured and organized to obtain multi-dimensional multivariate response data:

[0033] Step 1: Structuring the past medication data of the central nervous system infection patient and the predicted treatment results to obtain the logical rules of the central nervous system infection patient;

[0034] Step 2: Gradual expansion of the key variables for predicting treatment outcomes based on medical logic constraints to obtain key variable ranges of the key variables;

[0035] Step 3: Based on the logic rules and the range of the key variables, generate a basic medication regimen set for the central nervous system infection patient through an orthogonal combination strategy;

[0036] Step 4: Cross-match the drug variables in the basic medication regimen set with the predicted treatment results to obtain the treatment logic of the predicted treatment results, and generate multi-dimensional multivariate response data including medication regimen characteristics, efficacy prediction values ​​and side effect probabilities based on the treatment logic.

[0037] Optionally, the correcting of contradictory data in the multi-dimensional multivariate response data based on medical common sense includes:

[0038] Based on the medical common sense and the variable consistency principle, the contradictory data in the multi-dimensional multivariate response data are corrected.

[0039] Optionally, the steps for generating the personalized medication regimen are as follows:

[0040] Based on the predicted treatment outcome and the revised multi-dimensional multivariate response data, a core data set for treatment decision-making of the patient with central nervous system infection is obtained;

[0041] Based on the treatment decision core data set, a personalized dosing regimen is generated for the central infection patient. Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The present invention uses the disease data of patients with central infection and the constructed multi-source fusion feature function to integrate the patient characteristics, initial time and initial data of patients with central infection Multi-source fusion features are obtained by dynamically correlating with actual treatment results, overcoming the technical problems of data fragmentation, large individual differences, and difficult dynamic response in the treatment of central nervous system infections. At the same time, by fully integrating patient-specific data, dosage selection is made more accurate, reducing the problem of large fluctuations in efficacy, providing mathematical and algorithmic support for the generation of precise antimicrobial drug regimens, and improving the accuracy of drug administration operations for patients with central nervous system infections.

[0043] 2. The present invention obtains a closed-form solution by taking partial derivatives of each parameter in the reconstructed error function and updates the parameter matrix in the multi-source fusion feature through the closed-form solution. The predicted treatment result is obtained by feature mapping the updated multi-source fusion feature. When the input is a multi-dimensional fusion feature, the dynamic weight vector across dimensions can be automatically learned through linear transformation based on mutual information. At the same time, the feature dimension matrix, data modality dimension matrix and time dimension matrix of the central infection patient are automatically updated through the closed-form solution to ensure that the dosing regimen can be dynamically adjusted as the pathogen mutates, so as to achieve accurate optimization of the dosing regimen based on the individual characteristics of the patient, effectively control the treatment risk, quickly respond to complex scenarios, and significantly improve the accuracy of central infection treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A schematic flow chart of a system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to one embodiment of the present invention. DETAILED DESCRIPTION

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments belong to some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0046] The terms used in the embodiments of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The singular forms "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise, and "a plurality" generally includes at least two.

[0047] As used herein, the words “if” or “when” may be interpreted as “at the time of” or “when” or “in response to determining” or “in response to detecting,” depending on the context. Similarly, the phrases “if it is determined” or “if (stated condition or event) is detected” may be interpreted as “when it is determined” or “in response to the determination” or “when detecting (stated condition or event)” or “in response to detecting (stated condition or event),” depending on the context.

[0048] The embodiment of the present application provides a drug administration regimen optimization system for patients with central infection based on big data analysis. The execution subject of the drug administration regimen optimization system for patients with central infection based on big data analysis includes but is not limited to at least one of the electronic devices such as the server, terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the drug administration regimen optimization system for patients with central infection based on big data analysis can be executed by software or hardware installed on a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0049] like Figure 1 As shown, it is a flow chart of the system for optimizing the dosing regimen for patients with central infection based on big data analysis of the present invention. The system 100 for optimizing the dosing regimen for patients with central infection based on big data analysis of the present invention can be installed in an electronic device. According to the functions implemented, the system 100 for optimizing the dosing regimen for patients with central infection based on big data analysis may include a data acquisition module 101, a feature fusion module 102, a feature optimization module 103, a prediction module 104, a simulation data generation module 105, a correction module 106, and a dosing module 107. The module described in the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by a processor of an electronic device and can perform fixed functions, which are stored in the memory of an electronic device.

[0050] In this embodiment, the functions of each module / unit are as follows:

[0051] The data collection module 101 is used to collect disease data of patients with central nervous system infection.

[0052] In the embodiment of the present invention, the disease data includes: patient information and recent medication feedback.

[0053] Specifically, patient information is the basic data used for disease analysis and personalized modeling, including but not limited to: demographic information, clinical status data, laboratory test indicators, and vital signs monitoring data.

[0054] In detail, recent medication feedback reflects the patient's response to the current treatment plan and is the key basis for dynamically optimizing the medication regimen. It includes but is not limited to: medication records, treatment effect indicators, and adverse reaction monitoring.

[0055] Specifically: demographic information includes: age, gender, weight, height, genetic background; clinical status data includes: past medical history, current infection type, severity of illness; laboratory test indicators include: blood routine, cerebrospinal fluid biochemistry, pathogen detection; vital signs monitoring data include: body temperature, blood pressure, heart rate, and blood oxygen saturation.

[0056] Specifically: Medication records include: drug name, dosage, frequency of administration, and course of treatment; treatment effect indicators include: improvement of clinical symptoms, microbiological clearance rate, and changes in inflammatory markers; adverse reaction monitoring includes: drug toxicity-related data; treatment compliance data: whether the patient takes the medication regularly as prescribed by the doctor.

[0057] Furthermore, blood routine tests include but are not limited to: white blood cell count and neutrophil ratio; cerebrospinal fluid biochemistry includes but is not limited to: glucose and protein content; past medical history includes but is not limited to: immunodeficiency and chronic diseases; current infection type includes but is not limited to: bacterial meningitis and viral intracranial infection; disease severity includes but is not limited to: Glasgow Coma Scale.

[0058] Furthermore, clinical symptom improvement includes, but is not limited to, fever reduction, recovery of consciousness, microbiological clearance rate, negative cerebrospinal fluid culture, and changes in inflammatory markers. Drug toxicity-related data include, but are not limited to, fluctuations in renal function indicators accompanied by rash or elevated liver enzymes. Whether the patient takes medication regularly as prescribed includes, but is not limited to, records in smart medicine boxes or nurse medication logs.

[0059] The feature fusion module 102 is used to perform weight mapping on the diseased data to obtain weighted diseased data of the diseased data, perform tensor product construction on the weighted diseased data to obtain a multi-source fusion feature function of the weighted diseased data, and the solution of the multi-source fusion feature function is a multi-source fusion feature.

[0060] In this embodiment of the present invention, the calculation formula of the multi-source fusion feature is:

[0061] ;

[0062] in: It is a multi-source fusion feature. is the weighted disease data, is the total number of data in the disease data, is the dynamic weight vector, For actual treatment results, For the The disease data and actual treatment results The mutual information between For all the disease data and the actual treatment results The sum of the mutual information of is the patient feature dimension matrix, is the data modality dimension matrix, is the time dimension matrix, For the The disease data, For the The disease data, is the matrix multiplication along different dimensions.

[0063] In detail, To quantify the The relative importance of the disease data to the actual treatment outcome, In order to perform linear transformations along the patient feature dimension, data modality dimension, and time dimension in sequence, the original data are projected into the joint feature space to enhance interpretability.

[0064] In detail, disease data usually contains multi-source heterogeneous information, and the contribution of different data to treatment outcomes varies significantly. This embodiment calculates a dynamic weight vector through mutual information. Its mathematical essence is to quantify the dependency between the i-th disease data and the actual treatment outcome. The higher the mutual information value, the stronger the correlation between the data and the treatment outcome, and the greater its weight. This design can automatically filter out redundant or irrelevant information, focus on key features, and solve the problems of "information noise" and "curse of dimensionality" in multi-source data.

[0065] In detail, disease data naturally has multidimensional attributes, including patient feature dimension, data modality dimension and time dimension. The present invention uses tensor product operation to perform nonlinear transformation on the weighted disease data in three-dimensional space, generating a multi-source fusion feature function containing cross-dimensional interaction information, which solves the problem that traditional splicing or simple weighted fusion methods are difficult to capture high-order interaction relationships across dimensions.

[0066] Specifically, the weights are dynamically adjusted through mutual information to avoid the subjectivity of manually labeling the importance of features. This is especially suitable for scenarios where the label "treatment results" is scarce in clinical data. It can automatically mine the key data most relevant to the therapeutic effect, assign low weights to noise data with low correlation with the treatment results, reduce the interference of irrelevant information on the model, and improve the stability of subsequent classification or prediction tasks.

[0067] Specifically, the tensor product can capture nonlinear dependencies between different dimensions, such as "the correlation between the changing trend of a certain modality data in a time series and patient characteristics", significantly enhance the semantic expression ability of features, and support the flexible combination of multi-dimensional data. Regardless of how the data modality or time granularity changes, unified processing can be achieved through the adaptation of the corresponding dimensional matrix, thereby improving the versatility of the method.

[0068] Furthermore, in practical applications, taking a treatment effect prediction task as an example, data input is first performed, including m=50 disease data in 3 modalities, and the actual treatment result is the efficacy score (0-100 points). Through mutual information analysis, it is found that the mutual information value between "blood sugar fluctuation frequency" and the actual treatment result is the highest, and the dynamic weight vector reaches 0.25, while the mutual information of "irrelevant medical history" is close to 0, and the dynamic weight vector approaches 0, verifying the noise filtering effect. After tensor product construction, the accuracy of multi-source fusion features in the prediction model is improved compared with traditional methods, indicating the key role of multidimensional interactive features in efficacy prediction.

[0069] The feature optimization module 103 is used to perform low-rank approximate reconstruction on the multi-source fusion feature to obtain a reconstruction error function of the multi-source fusion feature, calculate the partial derivative of each parameter in the reconstruction error function to obtain a closed-form solution of the reconstruction error function, and update the parameters in the multi-source fusion feature function based on the closed-form solution to obtain an updated multi-source fusion feature.

[0070] In this embodiment of the present invention, the calculation formula of the reconstruction error function is:

[0071] ;

[0072] in: It is a multi-source fusion feature. is the dynamic weight vector, is the Frobenius norm, is the reconstruction error function, for The disease data, is the matrix multiplication along different dimensions, is the projection matrix set, is the patient feature dimension matrix, is the data modality dimension matrix, is the time dimension matrix, For the The disease data, is the weighted disease data, is the total number of data of the disease data, is the reconstruction matrix of the projection matrix set, are the reconstruction matrices corresponding to the patient feature dimension matrix, data modality dimension matrix, and time dimension matrix, respectively. is the regularization coefficient.

[0073] In detail, This is to constrain the weight of the projection matrix to avoid overfitting due to its excessive value. Multi-source fusion features are reconstructed across dimensions through projection matrices to generate low-rank approximate features, eliminate redundant information and retain core interaction relationships.

[0074] In this embodiment of the present invention, calculating the partial derivative of each parameter in the reconstruction error function includes:

[0075] Step 1: fixing the reconstruction matrix of the projection matrix set, calculating the closed form solution of the current projection matrix set, and updating the projection matrix set based on the closed form solution of the projection matrix set to obtain an updated projection matrix set;

[0076] Step 2: Update the multi-source fusion feature based on the updated projection matrix set to obtain an updated multi-source fusion feature, wherein the calculation formula of the updated multi-source fusion feature is:

[0077] ;

[0078] in: is the updated multi-source fusion feature, is the dynamic weight vector, is the weighted disease data, is the matrix multiplication along different dimensions, The updated patient feature dimension matrix, data modality dimension matrix and time dimension matrix;

[0079] Step 3: Fixing the updated projection matrix set, updating the reconstruction matrix of the projection matrix set, calculating the closed form solution of the reconstruction matrix of the projection matrix set, and updating the reconstruction matrix of the projection matrix set based on the closed form solution of the reconstruction matrix of the projection matrix set to obtain an updated reconstruction matrix of the projection matrix set;

[0080] Step 4: performing feature update on the reconstruction error function based on the reconstruction matrix of the updated projection matrix set to obtain an updated reconstruction error function.

[0081] In detail, although the multi-source fusion features have integrated multi-dimensional interactive information, there is still dimensional redundancy, such as repeated features of different modal data and noise fluctuations in time series. This embodiment introduces a reconstruction error function through the Frobenius norm constraint. By minimizing the reconstruction error, the model learns the low-dimensional manifold structure that can best retain the original information and eliminate noise and redundancy.

[0082] Specifically, by minimizing the Frobenius norm distance between the original weighted disease data and the low-rank reconstructed data, the core feature components are forced to be retained. For example, short-term fluctuation noise is filtered out in the time dimension, long-term disease trends are retained, and repeated signals from different detection methods are eliminated in the modal dimension to improve feature purity. After low-rank approximation, the dimension of high-dimensional medical data can be reduced to 10%-20% of the original space, significantly reducing the computational cost of subsequent modeling while avoiding the risk of overfitting.

[0083] In detail, the present invention adopts a step-by-step closed-form solution method. This alternating optimization strategy decomposes the high-dimensional joint optimization problem into several low-dimensional sub-problems, greatly reducing the computational complexity while avoiding falling into local optimality.

[0084] Furthermore, the projection matrix set and the reconstruction matrix of the projection matrix set are fixed in steps to avoid parameter oscillation problems in joint optimization. This is particularly suitable for non-stationary distributions commonly found in medical data, such as differences in characteristic distributions among different patient groups, ensuring the robustness of the optimization process to changes in data distribution.

[0085] In detail, the regularization term suppresses model overfitting by penalizing excessively large matrix parameters. In medical data scenarios, the sample size is usually limited and the noise distribution is complex. Regularization can effectively improve the model's generalization ability on unknown data and ensure the stability of feature optimization results.

[0086] The prediction module 104 is configured to perform feature mapping on the updated multi-source fusion features to obtain a predicted treatment result based on the multi-source fusion features.

[0087] In an embodiment of the present invention, the calculation formula for predicting treatment results is:

[0088] ;

[0089] in: To predict treatment outcomes, is the activation function, is the weight training matrix, is the bias term, is the updated multi-source fusion feature.

[0090] Detailed, , ,in: is the set of real numbers, is the dimension of the updated multi-source fusion feature.

[0091] Specifically, in medical scenarios, treatment outcomes are often affected by the interaction of multiple factors, such as the nonlinear process of drug metabolism dynamics and the dose-efficacy curve deviation caused by individual differences among patients. The introduction of activation functions enables the model to fit nonlinear mappings. By flexibly selecting activation functions and output dimensions, the module can seamlessly adapt to various clinical tasks such as classification and regression.

[0092] Specifically, by flexibly selecting activation functions, it can effectively respond to clinical needs such as classification and regression, and capture the complex relationship between features and efficacy. Secondly, the parameters in the calculation formula for predicting treatment results have clear physical meanings and low computational complexity, taking into account the dual requirements of medical scenarios for model transparency and real-time performance, and realizing joint training with the feature fusion module to maximize the synergistic effect of feature space and prediction targets, and improve the robustness of the model in real clinical data.

[0093] The simulation data generation module 105 is used to structure the past medication data of the central nervous system infection patient and the predicted treatment results to obtain multi-dimensional and multi-faceted response data.

[0094] In an embodiment of the present invention, the past medication data of the central nervous system infection patient and the predicted treatment results are structured and organized to obtain multi-dimensional multivariate response data:

[0095] Step 1: Structuring the past medication data of the central nervous system infection patient and the predicted treatment results to obtain the logical rules of the central nervous system infection patient;

[0096] Step 2: Gradual expansion of the key variables for predicting treatment outcomes based on medical logic constraints to obtain key variable ranges of the key variables;

[0097] Step 3: Based on the logic rules and the range of the key variables, generate a basic medication regimen set for the central nervous system infection patient through an orthogonal combination strategy;

[0098] Step 4: Cross-match the drug variables in the basic medication regimen set with the predicted treatment results to obtain the treatment logic of the predicted treatment results, and generate multi-dimensional multivariate response data including medication regimen characteristics, efficacy prediction values ​​and side effect probabilities based on the treatment logic.

[0099] In detail, the past medication data of patients with central nervous system infections, such as antibiotic type, dosage, and duration of medication, have complex correlations with predicted treatment outcomes, such as efficacy scores and the probability of side effects, and need to be structured through step one: converting unstructured data, such as electronic medical record text, into structured fields, such as drug name, route of administration, and laboratory indicators, establishing a mapping relationship of "medication regimen → treatment outcome", and extracting medical logic rules, such as "when β-lactam antibiotics need to be used in combination with glucocorticoids, the dosage needs to be adjusted according to the cerebrospinal fluid protein content" to ensure that the generated data meets clinical diagnosis and treatment standards.

[0100] Specifically, through gradient expansion: for continuous variables, such as creatinine clearance, it is divided into multiple intervals according to medical guidelines, such as <30mL / min, 30-60mL / min, and >60mL / min; for discrete variables, such as pathogen types, full enumeration expansion is performed, such as Streptococcus pneumoniae, Neisseria meningitidis, Listeria monocytogenes, etc., to construct a variable space covering different disease severity and pathogen characteristics, and solve the problem of insufficient rare case samples in real data.

[0101] In detail, the orthogonal experimental design concept is used to decompose the medication regimen into factors such as drug type, dosage, course of treatment, and combination medication. Several levels are set for each factor to generate a basic medication regimen set through an orthogonal table to ensure that the factor combinations are evenly distributed and non-redundant, covering the entire factor space at the minimum computational cost, and reducing the amount of calculation compared to the full factor combination.

[0102] In detail, a mapping relationship of "medication regimen → efficacy → side effects" is established through cross-matching. The treatment logic output by the prediction module is used to assign efficacy prediction values ​​and side effect probabilities to each basic regimen. The characteristics of the medication regimen, such as drug interaction index, dosing frequency, efficacy indicators, and safety indicators, are integrated to form multivariate reaction data with multi-dimensional labels to meet the multi-objective requirements of subsequent model training and clinical decision analysis.

[0103] The correction module 106 is configured to correct contradictory data in the multi-dimensional multi-factor response data based on medical common sense.

[0104] In an embodiment of the present invention, contradictory data in the multi-dimensional multivariate response data are corrected based on the medical common sense and the variable consistency principle.

[0105] In detail, by eliminating or correcting contradictory data, the model can avoid learning incorrect associations such as ineffective drugs and improved efficacy, and significantly improve the prediction accuracy and robustness of subsequent models. Experimental data show that in the task of predicting the efficacy of bacterial meningitis, the corrected data can improve the accuracy of the model. Secondly, the distribution of the corrected data is closer to the real clinical scenario, which effectively alleviates the problem of insufficient model generalization ability caused by data generation bias, especially in rare diseases or special population scenarios. It can improve the reliability of the model's prediction of edge cases and reduce the risk of misjudgment in clinical decision-making.

[0106] The medication module 107 is used to generate a personalized medication regimen for the central infection patient based on the predicted treatment results and the corrected multi-dimensional multi-response data.

[0107] In an embodiment of the present invention, a core data set for treatment decision-making of the patient with central nervous system infection is obtained based on the predicted treatment outcome and the corrected multi-dimensional multivariate response data;

[0108] A personalized dosing regimen for the central infection patient is generated based on the treatment decision core data set.

[0109] In detail, the beneficial effects of generating personalized medication regimens are achieved through the triple mechanisms of data integration, intelligent optimization, and rule constraints: first, dynamic responses based on individual patient characteristics break through the limitations of empirical treatment and make medication regimens highly compatible with the condition; second, automated decision-making significantly shortens diagnosis and treatment time, adapts to emergency and large-scale clinical scenarios, and helps improve medical efficiency.

[0110] In the several embodiments provided by the present invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the module division is merely a logical function division, and other division methods may be used in actual implementation.

[0111] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0112] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0113] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0114] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A drug administration regimen optimization system for patients with central nervous system infections based on big data analysis, characterized in that: include: Data collection module: used to collect disease data of patients with central nervous system infection, including patient information and recent medication feedback; Feature fusion module: used to perform weight mapping on the diseased data to obtain weighted diseased data of the diseased data, perform tensor product construction on the weighted diseased data to obtain a multi-source fusion feature function of the weighted diseased data. The solution of the multi-source fusion feature function is a multi-source fusion feature. The calculation formula of the multi-source fusion feature is: ； in: It is a multi-source fusion feature. is the weighted disease data, is the total number of data in the disease data, is the dynamic weight vector, is the patient feature dimension matrix, is the data modality dimension matrix, is the time dimension matrix, For the The disease data, is the matrix multiplication along different dimensions; The calculation formula of the dynamic weight vector is: ； in: For actual treatment results, For the The disease data and actual treatment results The mutual information between For all the disease data and the actual treatment results The sum of the mutual information of For the Individuals with the disease; Feature optimization module: used to perform low-rank approximate reconstruction on the multi-source fusion feature to obtain a reconstruction error function of the multi-source fusion feature, calculate the partial derivative of each parameter in the reconstruction error function to obtain a closed-form solution of the reconstruction error function, and update the parameters in the multi-source fusion feature function based on the closed-form solution to obtain an updated multi-source fusion feature; Prediction module: used to perform feature mapping on the updated multi-source fusion features to obtain the predicted treatment results of the multi-source fusion features; Simulation data generation module: used to structure and organize the past medication data of the central nervous system infection patient and the predicted treatment results to obtain multi-dimensional and multivariate response data; Correction module: used for correcting contradictory data in the multi-dimensional multi-element reaction data based on medical common sense; Drug administration module: used to generate a personalized drug administration plan for the central infection patient based on the predicted treatment results and the revised multi-dimensional multi-response data.

2. The system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to claim 1, characterized in that: The calculation formula of the reconstruction error function is: ； in: is the Frobenius norm, is the reconstruction error function, is the projection matrix set, is the reconstruction matrix of the projection matrix set, are the reconstruction matrices corresponding to the patient feature dimension matrix, data modality dimension matrix, and time dimension matrix, respectively. is the regularization coefficient.

3. The system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to claim 2, characterized in that: The step of obtaining a partial derivative of each parameter in the reconstruction error function comprises: Step 1: fixing the reconstruction matrix of the projection matrix set, calculating the closed form solution of the current projection matrix set, and updating the projection matrix set based on the closed form solution of the projection matrix set to obtain an updated projection matrix set; Step 2: Update the multi-source fusion feature based on the updated projection matrix set to obtain an updated multi-source fusion feature, wherein the calculation formula of the updated multi-source fusion feature is: ； in: is the updated multi-source fusion feature, The updated patient feature dimension matrix, data modality dimension matrix and time dimension matrix; Step 3: Fixing the updated projection matrix set, updating the reconstruction matrix of the projection matrix set, calculating the closed form solution of the reconstruction matrix of the projection matrix set, and updating the reconstruction matrix of the projection matrix set based on the closed form solution of the reconstruction matrix of the projection matrix set to obtain an updated reconstruction matrix of the projection matrix set; Step 4: performing feature update on the reconstruction error function based on the reconstruction matrix of the updated projection matrix set to obtain an updated reconstruction error function.

4. The system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to claim 3, characterized in that: The calculation formula for predicting treatment results is: ； in: To predict treatment outcomes, is the activation function, is the weight training matrix and , is the bias term and , is the set of real numbers, is the dimension of the updated multi-source fusion feature.

5. The system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to claim 1, characterized in that: The past medication data of the central nervous system infection patients and the predicted treatment results are structured and sorted to obtain multi-dimensional multivariate response data: Step 1: Structuring the past medication data of the central nervous system infection patient and the predicted treatment results to obtain the logical rules of the central nervous system infection patient; Step 2: Gradual expansion of the key variables for predicting treatment outcomes based on medical logic constraints to obtain key variable ranges of the key variables; Step 3: Based on the logic rules and the range of the key variables, generate a basic medication regimen set for the central nervous system infection patient through an orthogonal combination strategy; Step 4: Cross-match the drug variables in the basic medication regimen set with the predicted treatment results to obtain the treatment logic of the predicted treatment results, and generate multi-dimensional multivariate response data including medication regimen characteristics, efficacy prediction values ​​and side effect probabilities based on the treatment logic.

6. The system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to claim 1, characterized in that: The correcting of contradictory data in the multi-dimensional multi-factor response data based on medical common sense includes: The contradictory data in the multidimensional multivariate response data are corrected based on medical common sense and the principle of variable consistency.

7. The system for optimizing drug administration regimens for patients with central nervous system infections based on big data analysis according to claim 1, characterized in that: The steps for generating the personalized medication regimen are as follows: Based on the predicted treatment results and the revised multi-dimensional multivariate response data, a core data set for treatment decision-making of the patient with central nervous system infection is obtained; A personalized dosing regimen for the central infection patient is generated based on the treatment decision core data set.

Images