Multi-modal administration of egcg for nafld treatment planning systems and methods

Abstract

The present application relates to the technical field of EGCG multi-mode administration NAFLD treatment planning, in particular to an EGCG multi-mode administration NAFLD treatment planning system and method, which comprises the following steps: firstly, analyzing the information of NAFLD patients to obtain the correlation between patient characteristics and EGCG multi-mode administration; in the analysis process, clustering is performed after gender classification to obtain a plurality of clustering sets, and the correlation of the administration mode in the clustering set is analyzed to obtain a first administration mode correlation matrix; the second administration mode correlation matrix is obtained by analyzing the pathological characteristics of the patients; thus sufficient data analysis is provided for subsequent planning; secondly, an EGCG multi-mode administration planning model composed of a plurality of sub-models is proposed; the model can obtain an initial planning scheme according to the characteristics of similar patients; finally, the present application proposes an EGCG multi-mode administration planning optimization method to obtain a plurality of final planning schemes by calculating the fitness of the patients. The above methods comprehensively realize the effect of improving the work efficiency of doctors.

Description

Technical Field

[0001] This invention relates to the field of NAFLD treatment planning technology using multimodal dosing of EGCG, specifically to a system and method for NAFLD treatment planning using multimodal dosing of EGCG. Background Technology

[0002] NAFLD stands for Non-Alcoholic Fatty Liver Disease. NAFLD refers to a range of liver diseases caused by excessive fat accumulation in the liver without excessive alcohol consumption. The range of NAFLD varies from simple fatty liver (where fat is simply present in the liver) to non-alcoholic steatohepatitis, the latter involving inflammation of the liver and damage to hepatocellular cells.

[0003] Non-alcoholic fatty liver disease (NAFLD) is commonly associated with obesity, insulin resistance, type 2 diabetes, and metabolic syndrome. Management of NAFLD typically focuses on lifestyle modifications such as weight loss, dietary adjustments, and increased physical activity, while monitoring liver function and overall health. Early diagnosis and intervention are crucial for preventing disease progression and related complications.

[0004] In the treatment of NAFLD, a multimodal administration strategy of EGCG can improve efficacy. EGCG (epigallocatechin gallate) has shown unique advantages and promise as a potential adjuvant therapy option. EGCG is an important compound found in green tea and belongs to the flavonoid catechin class. EGCG possesses antibacterial, antiviral, antioxidant, anti-atherosclerotic, antithrombotic, anti-angiogenic, anti-inflammatory, and antitumor effects, demonstrating potential efficacy in the treatment of non-alcoholic fatty liver disease (NAFLD).

[0005] Existing treatments for NAFLD using multimodal EGCG administration primarily rely on physicians developing treatment plans based on their professional knowledge, clinical experience, and the patient's medical history. However, due to limited physician resources and the increasing number of patients and complex clinical situations, physicians often struggle to develop timely treatment plans. Therefore, this invention proposes a multimodal EGCG administration system and method for NAFLD treatment planning to assist physicians and improve their work efficiency. Summary of the Invention

[0006] The purpose of this invention is to provide a multimodal dosing system and method for NAFLD treatment planning using EGCG. This is mainly achieved through the following steps: First, by collecting various information about NAFLD patients, including personal information, disease information, NAFLD treatment records, and feedback forms as reference data; second, by analyzing the collected information, this invention obtains a first dosing modality correlation matrix and a second dosing modality correlation matrix; wherein, the first dosing modality correlation matrix uses a clustering method to cluster NAFLD patients with the same characteristics, resulting in multiple cluster combinations; and the correlation between different cluster sets and EGCG dosing modalities is analyzed through the clustering results; the second... The drug pattern correlation matrix analyzes dosing patterns based on NAFLD and non-NAFLD symptom data, providing more scientific recommendations for dosing patterns. Finally, this invention proposes an EGCG multimodal dosing planning model for generating initial multimodal dosing regimens. This model consists of multiple sub-models for dosing planning, matching the information characteristics of NAFLD patients with the corresponding sub-models to find the dosing patterns among the most similar patients. Furthermore, to provide personalized EGCG multimodal dosing regimens, this invention optimizes the output of the EGCG multimodal dosing planning model, taking into account the individual circumstances of the patients to be planned.

[0007] To achieve the above objectives, the present invention provides the following technical solution: First aspect: EGCG multimodal dosing NAFLD treatment planning system, the system comprising: a data storage unit, a data acquisition unit, a data processing unit, a data analysis unit, an EGCG dosing multimodal planning unit, a dosing multimodal optimization unit, a dynamic monitoring unit, and a feedback unit; The data storage unit is used to store information on multiple patients with NAFLD. The data storage unit comprises a database management system responsible for storing and managing the patient information, including: personal information, symptom information, NAFLD treatment records, and feedback forms. The personal information includes: name, age, gender, weight, height, and medical record number. The symptom information includes both NAFLD and non-NAFLD symptom information. The NAFLD treatment records include: treatment plan, EGCG administration mode, and treatment cycle. The feedback forms contain feedback and suggestions from NAFLD patients regarding their treatment.

[0008] The data acquisition unit is used to collect information about the patients to be planned; The data processing unit is used to process patient information of NAFLD patients to obtain an NAFLD analysis dataset; wherein, the data processing unit includes: acquiring the patient information of all NAFLD patients in the data storage unit; Furthermore, the patient information is filtered to obtain key analytical information; Further, the key analysis information is cleaned to obtain the first key analysis data; Furthermore, the first key analysis data is transformed to obtain the second key analysis data; Furthermore, the second key analysis data is normalized to obtain the third key analysis data; Furthermore, feature engineering is used to process the third key analysis data to obtain the fourth key analysis data; Furthermore, multiple sets of the fourth key analysis data are integrated based on the medical record number to obtain the NAFLD analysis data set.

[0009] The data analysis unit is used to analyze the patient information in the data storage unit; wherein, the data analysis unit includes: a first analysis module and a second analysis module; The first analysis module includes: acquiring the NAFLD analysis data set; The NAFLD analysis dataset is classified into a first dataset and a second dataset based on the gender field attribute. Furthermore, the data in the first and second datasets are clustered according to the field attributes of age, weight, and height, respectively, to obtain P male first clusters and Q female first clusters; wherein, the male first clusters and the female first clusters contain data on the age, weight, height, and corresponding EGCG administration patterns of NAFLD patients; Furthermore, an EGCG dosing pattern association analysis was performed on the first male cluster and the first female cluster to obtain male dosing association matrices and female dosing association matrices. Furthermore, the male dosing correlation matrix and the female dosing correlation matrix are jointly analyzed to obtain the first dosing pattern correlation matrix; The second analysis module includes: acquiring NAFLD symptom data, non-NAFLD symptom data, and EGCG administration pattern data from the NAFLD analysis dataset; Furthermore, the NAFLD symptom data and the non-NAFLD symptom data are used as input datasets, and the EGCG dosing pattern data is used as output datasets; Furthermore, the input dataset and the output dataset are substituted into the relational analysis model to obtain the second drug administration pattern correlation matrix.

[0010] The EGCG multimodal dosing planning unit is used to perform initial multimodal dosing planning for patients to be planned. The EGCG multimodal dosing planning unit uses the EGCG multimodal dosing planning model to perform initial dosing pattern planning for the patients to be planned; wherein the EGCG multimodal dosing planning model includes multiple modal dosing planning sub-models, and the specific process includes: The NAFLD analysis dataset is used as the training set; wherein, the personal information and disease information processed by the data processing unit are used as input samples during training; the NAFLD treatment records processed by the data processing unit are used as real samples during training; and the feedback table processed by the data processing unit is used as model optimization information during training. Furthermore, the input sample is received by the input layer, and the features of the input sample are extracted to obtain the drug administration pattern training features; Further, the drug administration pattern training features are input into the matching layer to obtain a drug administration model planning sub-model; wherein, the matching layer is used to match the drug administration pattern training features with clustering features to select a drug administration model planning sub-model that meets the requirements; the drug administration planning sub-model is a first drug administration pattern correlation matrix and a second drug administration pattern correlation matrix, and is a sub-model constructed by using the XGBoost regression model for different clusters; Further, the EGCG dosing pattern value is output by the dosing model planning sub-model; the dosing pattern planning error between the EGCG dosing pattern value and the real sample during training is calculated; Furthermore, the drug administration pattern planning error and the model optimization information are used to optimize the drug administration pattern planning sub-model; Furthermore, when the EGCG multimodal drug delivery planning model reaches the training stopping condition, the trained EGCG multimodal drug delivery planning model is obtained. Furthermore, the data acquisition unit acquires the input data of the patient to be planned, and after processing by the data processing unit, it is input into the trained EGCG multimodal drug delivery planning model to obtain the initial planning scheme for EGCG multimodal drug delivery.

[0011] The dynamic monitoring unit is used to monitor the clinical indicators of the patients to be planned in real time. The multi-modal dosing optimization unit is used to optimize the initial dosing regimen for the patient to be planned; wherein, the optimization process of the multi-modal dosing optimization unit includes: Obtain the initial multimodal dosing plan for EGCG for the patients to be planned, and identify key factors; Furthermore, an influence parameter is assigned to each of the key factors; Furthermore, the EGCG is classified using a multimodal drug delivery initial planning scheme to obtain multiple classification sets; Furthermore, by cross-combining multiple classification sets, multiple different subsets of initial drug delivery planning schemes are obtained; Further, the suitability of each subset of the initial dosing planning schemes to the patient to be planned is calculated; Furthermore, the subset of the initial dosing plans whose fitness exceeds the fitness threshold is taken as the final multimodal dosing plan for EGCG; The feedback unit is used to provide feedback to the doctor on multiple final EGCG multimodal dosing plans.

[0012] The second aspect: EGCG multimodal dosing for NAFLD treatment planning includes: Patient information for multiple NAFLD patients was retrieved from the database management system. This patient information includes: personal information, disease information, NAFLD treatment records, and a feedback form. The personal information includes: name, age, gender, weight, height, and medical record number. The disease information includes both NAFLD and non-NAFLD disease information. The NAFLD treatment records include: treatment plan, EGCG administration mode, and treatment cycle. The feedback form contains feedback and suggestions from NAFLD patients regarding their treatment.

[0013] Further, the patient information is processed to obtain a NAFLD analysis dataset; wherein the patient information processing includes: filtering the patient information to obtain key analysis information; cleaning the key analysis information to obtain first key analysis data; transforming the first key analysis data to obtain second key analysis data; normalizing the second key analysis data to obtain third key analysis data; processing the third key analysis data using feature engineering to obtain fourth key analysis data; and integrating multiple fourth key analysis data according to the medical record number to obtain the NAFLD analysis dataset.

[0014] Furthermore, a dosing pattern correlation analysis is performed on the data in the NAFLD analysis dataset to obtain a first dosing pattern correlation matrix and a second dosing pattern correlation matrix; The analysis process of the first drug administration mode correlation matrix includes: The NAFLD analysis dataset is classified into a first dataset and a second dataset based on the gender field attribute. Based on the age, weight, and height fields, the data in the first and second datasets are clustered to obtain P male first clusters and Q female first clusters; wherein, the male first clusters and the female first clusters contain data on the age, weight, height, and corresponding EGCG administration patterns of NAFLD patients; An association analysis of EGCG dosing patterns was performed on the first male cluster and the first female cluster to obtain male dosing association matrices and female dosing association matrices. The first dosing pattern correlation matrix is ​​obtained by jointly analyzing the male dosing correlation matrix and the female dosing correlation matrix. The analysis process of the second dosing modality correlation matrix includes: Acquire NAFLD symptom data, non-NAFLD symptom data, and EGCG dosing pattern data from the NAFLD analysis dataset; The NAFLD symptom data and the non-NAFLD symptom data are used as the input dataset, and the EGCG dosing pattern data is used as the output dataset. Furthermore, the input dataset and the output dataset are substituted into the relational analysis model to obtain the second drug administration pattern correlation matrix; Furthermore, the EGCG multimodal drug delivery planning model is trained to obtain the trained EGCG multimodal drug delivery planning model; The EGCG multimodal drug delivery planning model includes a modal drug delivery planning sub-model, and the training process includes: Furthermore, the processed personal information and disease information are used as input samples during training, NAFLD treatment records are used as real samples during training, and feedback forms are used as model optimization information during training. Furthermore, the input sample is received by the input layer, and the features of the input sample are extracted to obtain the drug administration pattern training features; Further, the drug administration pattern training features are input into the matching layer to obtain a drug administration model planning sub-model; wherein, the matching layer is used to match the drug administration pattern training features with clustering features to select a drug administration model planning sub-model that meets the requirements; the drug administration planning sub-model is a first drug administration pattern correlation matrix and a second drug administration pattern correlation matrix, and is a sub-model constructed by using the XGBoost regression model for different clusters; Further, the EGCG dosing pattern value is output by the dosing model planning sub-model; the dosing pattern planning error between the EGCG dosing pattern value and the real sample during training is calculated; Furthermore, the drug administration pattern planning error and the model optimization information are used to optimize the drug administration pattern planning sub-model; Furthermore, when the EGCG multimodal drug delivery planning model reaches the training stopping condition, the trained EGCG multimodal drug delivery planning model is obtained.

[0015] Furthermore, the data of the patients to be planned are collected, processed, and then input into the trained EGCG multimodal dosing planning model to obtain the initial planning scheme for EGCG multimodal dosing. Furthermore, the initial multimodal dosing plan for EGCG is optimized to obtain multiple final multimodal dosing plans for the patient to be planned; the specific process includes: The process involves obtaining an initial multimodal dosing plan for EGCG for the patient to be planned, and identifying key factors; assigning influencing parameters to each key factor; classifying the initial multimodal dosing plans for EGCG to obtain multiple classification sets; cross-combining the multiple classification sets to obtain multiple different subsets of initial dosing plans; calculating the fit between each subset of initial dosing plans and the patient to be planned; and selecting the subsets of initial dosing plans with a fit exceeding a fit threshold as the final multimodal dosing plan for EGCG.

[0016] Furthermore, multiple final EGCG dosing plans are fed back to the physician.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention proposes a multimodal dosing correlation analysis method to explore the correlation between NAFLD patients and EGCG dosing modalities. This method collects patient information from multiple NAFLD patients and obtains a first dosing modal correlation matrix and a second dosing modal correlation matrix from both non-correlation and correlation perspectives with NAFLD. The first dosing modal correlation matrix divides the collected data into two datasets based on gender. These datasets are then clustered according to multidimensional features to obtain multiple cluster sets. The cluster sets are analyzed to obtain the correlation between different clustering results and the dosing modalities. The second dosing modal correlation matrix analyzes NAFLD symptom data and non-NAFLD symptom data from NAFLD patients. Combining these two analytical approaches can provide scientific data support for subsequent multimodal EGCG dosing planning, which is beneficial for improving the efficiency of physicians in planning multimodal EGCG dosing for NAFLD patients.

[0018] 2. This invention proposes an EGCG multimodal dosing planning model for generating initial multimodal dosing regimens. The model consists of multiple modal dosing planning sub-models. By matching the information features of NAFLD patients with the corresponding sub-models, the dosing patterns among the most similar patients are sought. Furthermore, patient feedback records are used as additional optimization information during the model optimization process. This further improves the efficiency of doctors in planning EGCG multimodal dosing for NAFLD patients.

[0019] 3. This invention proposes an EGCG multimodal dosing planning optimization method for generating personalized EGCG multimodal dosing plans. This method categorizes initial plans and generates multiple subsets of plans through cross-combination. To achieve personalized EGCG multimodal dosing, key factors are set based on the patient's data, and the final EGCG multimodal dosing plan is derived by calculating the suitability of the subsets of plans with the key factors. Through precise drug planning and dynamic adjustment, this not only provides patients with more personalized plans but also improves the efficiency of physicians in planning NAFLD using EGCG multimodal dosing. Attached Figure Description

[0020] Figure 1 This is a structural diagram of the NAFLD treatment planning system for multimodal drug delivery of EGCG provided in an embodiment of the present invention; Figure 2 A flowchart of a multimodal dosing method for NAFLD treatment planning using EGCG provided in an embodiment of the present invention; Figure 3 This is a clustering diagram of the NAFLD analysis dataset provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] Non-alcoholic fatty liver disease, also known as pseudo-alcoholic liver disease, is a clinicopathological syndrome characterized by excessive fat deposition in hepatocytes, excluding alcohol and other clearly defined liver-damaging factors. It can lead to decompensated cirrhosis, hepatocellular carcinoma, and recurrence after liver transplantation. By taking appropriate amounts of EGCG, NAFLD patients can benefit from improved liver health, reduced fat accumulation, and decreased inflammation. EGCG, or epigallocatechin gallate, is a natural compound derived from green tea that has antioxidant, anti-inflammatory, and metabolic-improving effects.

[0023] Current EGCG multimodal administration for NAFLD treatment primarily involves physicians developing treatment plans based on their professional knowledge, clinical experience, and the patient's medical history. However, due to limited physician resources and the increasing number of patients and complex clinical situations, physicians often cannot develop treatment plans in a timely manner.

[0024] Therefore, this invention proposes a multimodal EGCG administration system and method for NAFLD treatment planning to generate reference plans, thereby improving physician efficiency; this will be specifically described from the following two embodiments. Example

[0025] In this application, the system and method proposed in this invention are used for NAFLD treatment planning with multimodal dosing of EGCG; see also [link to invention]. Figure 1 and Figure 2 Content, Figure 1 The content describes the system structure of this invention, which includes: a data storage unit, a data acquisition unit, a data processing unit, a data analysis unit, an EGCG multimodal dosing planning unit, a dynamic monitoring unit, a dosing multimodal optimization unit, and a feedback unit. Figure 2 The method flow steps of this invention include: S10. Obtaining information on multiple patients with NAFLD; S20. Processing the obtained patient information and constructing a NAFLD analysis dataset; S30. Performing dosing modality correlation analysis; S40. Generating an initial multimodal dosing plan for EGCG for the patients to be dosing; S50. Optimizing the initial multimodal dosing plan for EGCG; S60. Feeding back multiple final multimodal dosing plans for EGCG to the physician. Figure 1 There is a correspondence between the units and the method steps. The following process combines... Figure 1 and Figure 2 content; Patient information for multiple treated NAFLD patients is retrieved from the system's data storage unit, corresponding to... Figure 1Step S10; wherein the patient information is stored in a database management system, and the database management system sets user roles to ensure that only authorized personnel can access or modify sensitive patient information; and the stored patient information is encrypted to protect the privacy of NAFLD patients.

[0026] Furthermore, the data processing unit of the system processes the acquired NAFLD patient information and constructs a NAFLD analysis dataset, corresponding to step S20 above; see Table 1, which provides a data description of acquiring NAFLD patient information in the embodiments of this application;

[0027] In this embodiment, patient information includes multiple aspects such as personal information, disease information, NAFLD treatment records, and feedback forms. Personal information includes name, age, gender, weight, height, and medical record number. Disease information includes NAFLD disease information and non-NAFLD disease information. NAFLD disease information involves the patient's consultation date, disease severity, and liver function indicators, such as ALT (alanine aminotransferase), AST (aspartate aminotransferase), and liver ultrasound results. Non-NAFLD disease information includes related diseases and allergy symptoms, such as diabetes and hypertension; drug allergies, food allergies, and drug administration mode allergies. The NAFLD treatment records include treatment plan, EGCG administration mode, and treatment cycle. The feedback form contains feedback and suggestions obtained from follow-up visits after the completion of treatment for NAFLD patients. In this embodiment of the application, information from multiple NAFLD patients in the database is collected as analytical data for the multimodal dosing of EGCG for NAFLD treatment planning. This data includes personal information, disease information, NAFLD treatment records, and feedback forms. By collecting and analyzing information from a large number of NAFLD patients, data-based analysis can help compare the effects of different treatment regimens, discover best practices, and optimize the EGCG dosing strategy.

[0028] The patient data described above will be processed. Since the patient data contains various types of attribute fields, different data processing methods are required. The specific process is as follows: The patient information in the data storage unit is screened; the principle of screening is to remove information that is not related to EGCG multimodal dosing analysis, such as the name, contact information and medical record number of NAFLD patients, or information that is not related to EGCG multimodal dosing in non-NAFLD disease information. Furthermore, the selected key analytical information is cleaned to obtain the first key analytical data; the data cleaning process includes the following aspects: Deduplication: Check and delete duplicate records to ensure that each NAFLD patient's information appears only once in the dataset; Handling missing values: For important fields (such as NAFLD diagnosis information, EGCG treatment records, etc.), if the proportion of missing values ​​is small, you can consider filling them with the mean, median or mode, or directly removing the record. Standardized data: Ensure all data uses a consistent format. For example, the gender field should use a consistent identifier (e.g., 0 / 1; 0 represents male, 1 represents female); the age field should be in a consistent numeric format; and treatment records should use standardized codes or descriptions. Furthermore, the first key analysis data is transformed to obtain the second key analysis data; for example, age should be an integer, weight should be a floating-point number, performance should be a character type, and treatment records should be categorical data, etc. Furthermore, the second key analysis data is normalized to obtain the third key analysis data; wherein, the third key analysis data is obtained by normalizing the numerical information contained in the second key analysis data; the normalization operation can be performed using the following methods: (1) Min-Max Normalization scales the data to the range [0,1], as shown in the formula: ;in, Represented as normalized key analysis data; This is represented as the original key analysis data in the second key analysis data; Represented as the smallest second key analysis data value in the second key analysis data; Represented as the smallest second key analysis data value in the second key analysis data; (2) Z-score standardization transforms the data into a distribution with a mean of 0 and a standard deviation of 1. The formula is: in, Represented as key analysis data after Z-score standardization; This is represented as the mean of the data in the second key analysis. This is represented as the standard deviation of the data in the second key analysis. Furthermore, feature engineering is used to process the third key analysis data to obtain the fourth key analysis data; wherein, in the embodiments of this application, the processing of the third key analysis data using feature engineering mainly includes the following process: (1) Feature construction: Construct new features using interactive data from the third key analysis data; for example, age, height, weight, etc. are closely related to NAFLD; therefore, a new feature can be created by combining age and weight. For example, the interaction between weight and height, known as Body Mass Index (BMI), is calculated using the following formula: ; This represents the weight of a NAFLD patient; The height of NAFLD patients is represented by Age_Weight_Interaction, which is calculated using the following formula: ; This represents the age of the NAFLD patient; (2) Feature selection: The variance selection method is used to select the most relevant feature data from the existing data; (3) Feature encoding: convert type features such as gender, treatment records and age group into unique heat encoding format.

[0029] Furthermore, the fourth key analysis data were integrated based on the medical record number in the original patient information to obtain the NAFLD analysis dataset; In this embodiment of the application, the collected patient information is processed to establish a NAFLD analysis dataset. The data processing involves data screening, data cleaning, normalization, feature engineering techniques, and data integration. The systematic data processing facilitates subsequent multimodal EGCG dosing correlation analysis for NAFLD patients.

[0030] Further, the data analysis unit of the system performs EGCG dosing pattern analysis on the NAFLD analysis dataset, corresponding to step S30 above; the data analysis unit includes a first analysis module and a second analysis module; The analysis process of the first analysis module includes: Obtain the NAFLD analysis dataset; Furthermore, based on the gender attribute field, the NAFLD analysis dataset is divided into two datasets: the first dataset and the second dataset; wherein, the first dataset is the dataset for males; and the second dataset is the dataset for females. Furthermore, the first and second datasets are clustered based on age, weight, and height; the clustering steps are as follows: Step 1: Construct three clustering criteria based on age, weight, and height, as follows: (1) Age clustering criteria: Based on the age segmentation of the NAFLD analysis dataset, NAFLD patients were assigned to four age groups; First age group: Second age group: Third age group: Fourth age group .

[0031] (2) Weight clustering criteria, based on the body mass index (BMI) derived from the weight and height data of NAFLD patients in the dataset, are set, including: First weight segment: Second weight group: Third weight group: Fourth weight group: .

[0032] (3) Height clustering criteria: Obtain the height data of NAFLD patients in the first and second datasets; perform statistics on the height data in the two datasets respectively; set the segments according to the percentile of the statistical height; first height segment: below the 25th percentile; second height segment: from the 25th percentile to the 75th percentile; third height segment: above the 75th percentile.

[0033] Step 2: Combine the age clustering criterion, weight clustering criterion, and height clustering criterion to obtain the joint clustering criterion; Step 3: Perform clustering on the first and second datasets using the OPTICS algorithm. During the OPTICS clustering process, a joint clustering criterion is used as the criterion for evaluating the distance between data samples. The OPTICS algorithm is a density-based clustering algorithm that sorts the data in a dataset and outputs an ordered list of objects containing all the information needed to extract clusters. This algorithm does not require pre-specifying the number of clusters, can discover clusters of arbitrary shapes, and can identify noise points and outliers.

[0034] Further, see Figure 3 After the clustering process from step one to step three, the data in the first dataset is clustered into 5 male first clusters and 4 female first clusters; wherein, the male first clusters and the female first clusters contain data on the age, weight, height and corresponding EGCG administration patterns of NAFLD patients. Furthermore, an association analysis of EGCG dosing patterns was performed on the first cluster of males and the first cluster of females to obtain the male dosing association matrix and the female dosing association matrix; see Tables 2 and 3;

[0035] Tables 2 and 3 provide the relevant information. Figure 3 The dosing association between male first cluster 1 and female first cluster 1; Furthermore, the dosing association matrices calculated from the five male first clusters and the four female first clusters were jointly analyzed to obtain the first dosing pattern association matrix; see Table 4;

[0036] In Table 4, the values ​​range from 0 to 1, reflecting the strength of the association between different individual characteristics and the dosing regimen. For example, the association between age and dosing regimen is 0.75, indicating a strong correlation between these two variables; the association between dosage and weight is 0.70, indicating that weight affects the required dosage to some extent.

[0037] The analysis process of the second analysis module includes: Acquire NAFLD symptom data, non-NAFLD symptom data, and EGCG dosing pattern data from the NAFLD analysis dataset; Furthermore, a relational analysis model was used to analyze NAFLD symptom data, non-NAFLD symptom data, and EGCG dosing pattern data. NAFLD symptom data and non-NAFLD symptom data were used as input data, and EGCG dosing pattern data were used as output data. This relational analysis model was based on an LSTM model and analyzed NAFLD (non-alcoholic fatty liver disease) symptom data, non-NAFLD symptom data, and EGCG (epigallocatechin gallate) dosing pattern data according to different stages of NAFLD patients. In this embodiment, a data analysis unit is used to analyze the information of NAFLD patients. This unit is divided into a first analysis module and a second analysis module. The first analysis module analyzes the correlation between the age, height, and weight of NAFLD patients and EGCG multimodal dosing. The second analysis module uses an LSTM model to analyze the patient's symptom information and non-NAFLD symptom information, such as complications or allergies present in NAFLD patients, and the EGCG dosing mode. Combining the above two analytical bases can provide scientific data support for subsequent EGCG multimodal dosing planning, which is conducive to improving the efficiency of doctors' EGCG multimodal dosing planning for NAFLD.

[0038] Further, the EGCG multimodal drug delivery planning unit of the system performs initial protocol planning for patient A, corresponding to step S40 above; the basic information of patient A involved in this embodiment is: male, 50 years old, obese, and according to medical record information, his NAFLD is at a moderate level; before the EGCG multimodal drug delivery planning unit performs planning, an EGCG multimodal drug delivery planning model needs to be trained; the training process of this model includes: Obtain a training set related to NAFLD; wherein, the NAFLD analysis dataset in this embodiment is used as the training set; relevant personal information in the set, such as gender, age, height, and weight, along with disease information, are used as input samples for training; the real samples in the training set are the EGCG dosing patterns saved by NAFLD patients in the database, including: dosing method, dosage, and dosing frequency; and the feedback forms of NAFLD patients are obtained as additional optimization information for the model during training; the specific training process is as follows: The input layer receives input samples and extracts features from these samples to obtain training features for the drug administration pattern. Further, the dosing pattern training features are input into the matching layer to obtain the dosing model planning sub-model; wherein, the matching layer is used to match the dosing pattern training features with the clustering features in the data analysis unit to select the dosing model planning sub-model that meets the requirements; the dosing pattern planning sub-model is a first dosing pattern correlation matrix and a second dosing pattern correlation matrix, and the sub-model is constructed by using the XGBoost regression model for different clusters; Further, the EGCG dosing pattern value is output by the dosing model planning sub-model; the dosing pattern planning error between the EGCG dosing pattern value and the real sample during training is calculated; Furthermore, the drug administration pattern planning error and the characteristic feedback information are used to optimize the pattern drug administration planning sub-model; Furthermore, when the EGCG multimodal drug delivery planning model reaches the training stopping condition, the trained EGCG multimodal drug delivery planning model is obtained. A tolerance (such as a threshold) can be set for the model's training stopping condition; if the loss on the validation set does not decrease beyond this threshold within a certain number of epochs, training stops.

[0039] Furthermore, the data acquisition unit of the system acquires the data of the patient A to be planned, and after processing, it is input into the trained EGCG multimodal drug delivery planning model to obtain the initial planning scheme for EGCG multimodal drug delivery. The initial dosing plan for EGCG using multiple transdermal dosing modalities includes: dosing methods such as oral administration and transdermal administration; dosage and frequency of administration; the EGCG multimodal dosing planning model can output a variety of different initial dosing regimens; see Table 5, which shows some of the initial dosing regimens generated by the EGCG multimodal dosing planning model for patient A to be planned.

[0040] In this embodiment, an initial dosing pattern planning scheme is generated for a patient to be planned by training an EGCG multimodal dosing planning model. The model establishes multiple drug planning sub-models based on the XGBoost model through cluster analysis results. This not only provides patients with more accurate and personalized treatment plans, but also uses patient feedback records as additional optimization information during the model optimization process, which further improves the efficiency of doctors in planning NAFLD using EGCG multimodal dosing.

[0041] Furthermore, the system's multi-modal dosing optimization unit is used to optimize the EGCG multi-modal dosing planning model for generating a treatment plan for patient A, corresponding to step S50 above; the specific optimization process is as follows: Obtain the initial multimodal dosing plan for EGCG in patient A; wherein the initial multimodal dosing plan is expressed as: ; This represents the set of initial planning schemes for multimodal drug delivery of EGCG; The number of initial dosing plans generated for patient A in the EGCG multimodal dosing planning model is represented by this number. Represented as the first The administration method for an initial drug administration plan; Represented as the first Dosage for an initial dosing plan; Represented as the first Dosing frequency for an initial dosing plan; Furthermore, key factors are identified; these key factors refer to information about the patient's data that influences the treatment planning process. Examples include: basic patient information such as age, gender, weight, height, and complications; physiological indicators such as liver function indicators (ALT, AST, GGT, etc.) and the severity of fatty liver (assessed through imaging or biochemical indicators); past medical history such as allergies to specific medications and previous treatment responses; and lifestyle habits such as diet, exercise habits, and other lifestyle factors. All of the factors listed above will affect the treatment of NAFLD patients, and this list does not include all factors; professionals may add more based on their expertise. Furthermore, an influence parameter is assigned to each of the key factors; wherein the assignment of the influence parameter can be achieved through sensitivity analysis combined with expert knowledge, and the specific steps include: Step 1: Set the initial range of influence parameters for each key factor; Step 2: Design a structured questionnaire and ask experts to rate the degree of influence of each key factor; Step 3: Adjust the initial influencing parameters based on expert feedback; Step 4: Establish a sensitivity analysis model based on the global sensitivity analysis method; Step 5: Input the adjusted preliminary impact parameters into the sensitivity analysis model; Step 6: Gradually change the influence parameters of each key factor and record the changes in the model output; Step 7: Evaluate the results of the sensitivity analysis and determine the influencing parameters; Furthermore, EGCG was classified using a multimodal dosing initial planning scheme, resulting in multiple classification sets. The classification was carried out according to the following methods: by dosing method, by dosing dose, and by dosing frequency; see Tables 6, 7, and 8 for the classifications given.

[0042] Furthermore, by cross-combining multiple classification sets, multiple different subsets of initial drug delivery planning schemes are obtained; Further, the fitness of each subset of initial dosing plans with respect to the patient A to be planned is calculated; wherein the fitness is calculated using the following formula:

[0043] in, Represented as a subset of the initial drug administration planning scheme Suitability; Represented as a subset of the initial drug administration planning scheme ; This indicates information about patient A, who is to be planned. This is expressed as a drug compatibility score, based on drug interactions and the allergy history of patient A to be planned; This is expressed as a dosage rationality score, based on factors such as the patient A's weight, age, and liver and kidney function. The appropriateness score for the route of administration is used to assess the suitability of the administration method for patient A under treatment. The feasibility score for the frequency of drug administration is represented by a score that takes into account the lifestyle and accessibility of medical resources for patient A, who is to be planned for. This is represented as the characteristic score of patient A to be planned, based on patient A's specific medical history and health status; Furthermore, the subset of the initial dosing plans whose fitness exceeds the fitness threshold is used as the final multimodal dosing plan for EGCG; wherein, in this embodiment of the application, the fitness threshold is set to 0.65 using big data analysis based on the dosing regimens of NAFLD patients in the database management system; Furthermore, the system's dynamic monitoring unit is used to monitor the physical health indicators of the patient A during the treatment process; Furthermore, the system's feedback unit feeds back multiple final EGCG multimodal dosing plans to the physician, corresponding to step S60 above.

[0044] In this embodiment, the present invention realizes the multimodal dosing of EGCG for NAFLD treatment planning for patient A. This process mainly includes the following: First, patient information of multiple NAFLD patients is obtained from a database and analyzed. The analysis process is divided into two modules: a first analysis module and a second analysis module. The first analysis module classifies NAFLD patients by gender and clusters the classified data using external features, namely age, height, and weight, to obtain multiple cluster sets. The correlation between dosing patterns in different cluster structures is analyzed. The second analysis model analyzes the NAFLD and non-NAFLD symptoms of NAFLD patients against the dosing patterns. This achieves the function of multimodal dosing analysis of EGCG from multiple dimensions, providing sufficient analysis for subsequent dosing pattern planning. Secondly, an E... The GCG multimodal dosing planning model is used to generate initial dosing pattern plans. Based on the XGBoost model, this model establishes multiple drug planning sub-models to find suitable planning models for similar patients, and incorporates patient feedback records as additional optimization information during model optimization. Finally, an optimization method for GCG multimodal dosing planning is proposed. This method classifies the initial planning schemes and generates multiple planning scheme subsets through cross-combination. To achieve personalized GCG multimodal dosing, key factors are set based on the data of the patients to be planned, and the final GCG multimodal dosing plan is obtained by calculating the suitability of the planning scheme subsets with the key factors. Through precise drug planning and dynamic adjustment, this not only provides patients with personalized reference plans but also improves the efficiency of doctors in planning NAFLD using GCG multimodal dosing.

[0045] Example 2: In Example 1, the system and method proposed in this invention were used to implement the NAFLD treatment plan for patient A with EGCG multimodal dosing. To further illustrate the effectiveness of this invention, in Example 2, the NAFLD treatment plan for patient B with EGCG multimodal dosing was implemented. See Table 9 for the basic information of patient B.

[0046] The following describes the EGCG multimodal dosing NAFLD treatment planning process for patient B: Information on multiple patients with NAFLD was obtained from the database management system mentioned in Example 1; Furthermore, the patient information is processed to obtain a NAFLD analysis dataset; Furthermore, a dosing pattern correlation analysis is performed on the data in the NAFLD analysis dataset to obtain a first dosing pattern correlation matrix and a second dosing pattern correlation matrix; the first dosing pattern correlation matrix and the second dosing pattern correlation matrix in this embodiment are the same matrices as the correlation matrix in Embodiment 1; Furthermore, the data of patient B to be planned is processed and input into the trained EGCG multimodal dosing planning model to obtain the initial planning scheme for EGCG multimodal dosing. Furthermore, the initial multimodal dosing plan for EGCG in patient B was obtained, and key factors were identified. Furthermore, an influence parameter is assigned to each of the key factors; Furthermore, the EGCG is classified using a multimodal drug delivery initial planning scheme to obtain multiple classification sets; Furthermore, by cross-combining multiple classification sets, multiple different subsets of initial drug delivery planning schemes are obtained; Further, the suitability of each subset of the initial dosing plans for the patient B to be planned is calculated; Furthermore, the subset of the initial dosing plans whose fitness exceeds the fitness threshold is taken as the final multimodal dosing plan for EGCG; Furthermore, multiple final EGCG dosing plans are fed back to the physician.

[0047] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multimodal drug delivery system for EGCG to treat NAFLD, characterized in that, The system includes: a data storage unit for storing information on multiple patients with NAFLD; a data acquisition unit for acquiring information on patients to be planned; a data processing unit for processing the patient information of NAFLD patients to obtain an NAFLD analysis dataset; a data analysis unit for analyzing the patient information in the data storage unit; an EGCG multimodal dosing planning unit for initial multimodal dosing planning for the patients to be planned; a dynamic monitoring unit for real-time monitoring of the clinical indicators of the patients to be planned; and a multimodal dosing optimization unit for optimizing the initial dosing regimen of the patients to be planned. The optimization process of the multimodal dosing optimization unit includes: obtaining the initial multimodal dosing plan for EGCG for the patient to be planned and identifying key factors; assigning influence parameters to each key factor; classifying the initial multimodal dosing plan for EGCG to obtain multiple classification sets; cross-combining the multiple classification sets to obtain multiple different subsets of initial dosing plans; calculating the suitability of each subset of initial dosing plans with the patient to be planned; and using the subsets of initial dosing plans with suitability exceeding a suitability threshold as the final multimodal dosing plan for EGCG. The feedback unit is used to provide feedback to the physician on multiple final EGCG dosing plans.

2. The EGCG multimodal drug delivery NAFLD treatment planning system according to claim 1, characterized in that, The data storage unit consists of a database management system, which is responsible for storing and managing the patient information of the NAFLD patients, including: personal information, symptom information, NAFLD treatment records, and feedback forms; wherein, the personal information includes: name, age, gender, weight, height, and medical record number; the symptom information includes: NAFLD symptom information and non-NAFLD symptom information; the NAFLD treatment records include: treatment plan, EGCG administration mode, and treatment cycle; the feedback form contains the NAFLD patients' feedback and suggestions after treatment.

3. The EGCG multimodal drug delivery NAFLD treatment planning system according to claim 1, characterized in that, The data processing unit includes: acquiring patient information for all NAFLD patients in the data storage unit; filtering the patient information to obtain key analysis information; cleaning the key analysis information to obtain first key analysis data; transforming the first key analysis data to obtain second key analysis data; normalizing the second key analysis data to obtain third key analysis data; processing the third key analysis data using feature engineering to obtain fourth key analysis data; and integrating multiple sets of fourth key analysis data according to medical record numbers to obtain the NAFLD analysis data set.

4. The EGCG multimodal drug delivery NAFLD treatment planning system according to claim 1, characterized in that, The data analysis unit includes: a first analysis module and a second analysis module; The first analysis module includes: acquiring the NAFLD analysis dataset; classifying the NAFLD analysis dataset into a first dataset and a second dataset based on the gender field attribute; clustering the data in the first dataset and the second dataset according to the age, weight, and height field attributes, respectively, to obtain P male first clusters and Q female first clusters; wherein, both the male and female first clusters contain the age, weight, height, and corresponding EGCG dosing pattern data of the NAFLD patients; performing EGCG dosing pattern correlation analysis on the male and female first clusters to obtain a male dosing correlation matrix and a female dosing correlation matrix; and jointly analyzing the male and female dosing correlation matrices to obtain a first dosing pattern correlation matrix. The second analysis module includes: acquiring NAFLD symptom data, non-NAFLD symptom data, and EGCG administration pattern data from the NAFLD analysis dataset; using the NAFLD symptom data and the non-NAFLD symptom data as input datasets and the EGCG administration pattern data as output datasets; and substituting the input datasets and the output datasets into a relational analysis model to obtain a second administration pattern correlation matrix.

5. The EGCG multimodal drug delivery NAFLD treatment planning system according to claim 1, characterized in that, The EGCG multimodal dosing planning unit uses the EGCG multimodal dosing planning model to perform initial dosing pattern planning for the patient to be planned; wherein, the EGCG multimodal dosing planning model includes multiple modal dosing planning sub-models, and the specific process includes: The NAFLD analysis dataset is used as the training set; wherein, the personal information and disease information processed by the data processing unit are used as input samples during training; the NAFLD treatment records processed by the data processing unit are used as real samples during training; and the feedback table processed by the data processing unit is used as model optimization information during training. The input samples are received by the input layer, and the features of the input samples are extracted to obtain the drug administration pattern training features; The drug administration pattern training features are input into the matching layer to obtain the drug administration model planning sub-model; wherein, the matching layer is used to match the drug administration pattern training features with the clustering features in the data analysis unit to select the drug administration model planning sub-model that meets the requirements; the drug administration planning sub-model is a first drug administration pattern correlation matrix and a second drug administration pattern correlation matrix, and the sub-model is constructed by using the XGBoost regression model for different clusters; The EGCG dosing pattern value is output from the dosing model planning sub-model; the dosing pattern planning error between the EGCG dosing pattern value and the real sample during training is calculated. The dosing pattern planning sub-model is optimized based on the dosing pattern planning error and the model optimization information; When the EGCG multimodal drug delivery planning model reaches the training stopping condition, the trained EGCG multimodal drug delivery planning model is obtained. The data acquisition unit acquires the input data of the patient to be planned, and after processing by the data processing unit, it is input into the trained EGCG multimodal dosing planning model to obtain the initial planning scheme for EGCG multimodal dosing.

6. A treatment planning method for NAFLD using EGCG with multimodal dosing, characterized in that, include: Retrieve patient information for multiple NAFLD patients from the database management system; The patient information is processed to obtain a NAFLD analysis dataset; A dosing pattern correlation analysis was performed on the data in the NAFLD analysis dataset to obtain a first dosing pattern correlation matrix and a second dosing pattern correlation matrix; The analysis process of the first dosing pattern correlation matrix includes: classifying the NAFLD analysis dataset into a first dataset and a second dataset based on the gender field attribute; clustering the data in the first dataset and the second dataset according to the age, weight, and height field attributes, respectively, to obtain P male first clusters and Q female first clusters; wherein, both the male and female first clusters contain the age, weight, height, and corresponding EGCG dosing pattern data of NAFLD patients; performing EGCG dosing pattern correlation analysis on the male and female first clusters to obtain male dosing correlation matrices and female dosing correlation matrices; and jointly analyzing the male and female dosing correlation matrices to obtain the first dosing pattern correlation matrix. The analysis process of the second dosing mode correlation matrix includes: acquiring NAFLD symptom data, non-NAFLD symptom data, and EGCG dosing mode data from the NAFLD analysis dataset; using the NAFLD symptom data and the non-NAFLD symptom data as input datasets and the EGCG dosing mode data as output datasets; and substituting the input datasets and the output datasets into the relationship analysis model to obtain the second dosing mode correlation matrix. Train the EGCG multimodal drug delivery planning model to obtain the trained EGCG multimodal drug delivery planning model; After the data of the patients to be planned are collected and processed, they are input into the trained EGCG multimodal dosing planning model to obtain the initial planning scheme for EGCG multimodal dosing. The initial multimodal dosing plan for EGCG was optimized to obtain multiple final multimodal dosing plans for EGCG for the patient to be planned. The multiple final EGCG dosing plans will be fed back to the physician.

7. The method for planning NAFLD treatment with multimodal EGCG administration according to claim 6, characterized in that, The patient information includes: personal information, disease information, NAFLD treatment records, and a feedback form; wherein, the personal information includes: name, age, gender, weight, height, and medical record number; the disease information includes: NAFLD disease information and non-NAFLD disease information; the NAFLD treatment records include: treatment plan, EGCG administration mode, and treatment cycle; the feedback form is the NAFLD patient's feedback and suggestions after treatment.

8. The method for planning NAFLD treatment with multimodal EGCG administration according to claim 6, characterized in that, The patient information processing procedure includes: filtering the patient information to obtain key analysis information; cleaning the key analysis information to obtain first key analysis data; transforming the first key analysis data to obtain second key analysis data; normalizing the second key analysis data to obtain third key analysis data; processing the third key analysis data using feature engineering to obtain fourth key analysis data; and integrating multiple sets of fourth key analysis data according to the medical record number to obtain the NAFLD analysis data set.

9. The method for planning NAFLD treatment with multimodal EGCG administration according to claim 6, characterized in that, The EGCG multimodal drug delivery planning model includes a modal drug delivery planning sub-model, and the training process includes: The processed personal and disease information are used as input samples during training, NAFLD treatment records are used as real samples during training, and feedback forms are used as model optimization information during training. The input samples are received by the input layer, and the features of the input samples are extracted to obtain the drug administration pattern training features; The drug administration pattern training features are input into the matching layer to obtain the drug administration model planning sub-model; wherein, the matching layer is used to match the drug administration pattern training features with clustering features to select the drug administration model planning sub-model that meets the requirements; the drug administration planning sub-model is a first drug administration pattern correlation matrix and a second drug administration pattern correlation matrix, and the sub-model is constructed by using the XGBoost regression model for different clusters; The EGCG dosing pattern value is output from the dosing model planning sub-model; the dosing pattern planning error between the EGCG dosing pattern value and the real sample during training is calculated. The drug administration pattern planning error and the model optimization information are used to optimize the drug administration pattern planning sub-model; When the EGCG multimodal drug delivery planning model reaches the training stopping condition, the trained EGCG multimodal drug delivery planning model is obtained.

10. The method for planning NAFLD treatment with multimodal EGCG administration according to claim 6, characterized in that, The optimization process of the initial EGCG multimodal dosing plan includes: obtaining the initial EGCG multimodal dosing plan for the patient to be planned and identifying key factors; assigning influence parameters to each key factor; classifying the initial EGCG multimodal dosing plan to obtain multiple classification sets; cross-combining the multiple classification sets to obtain multiple different subsets of the initial dosing plan; calculating the fit of each subset of the initial dosing plan with the patient to be planned; and using the subsets of the initial dosing plan with a fit exceeding a fit threshold as the final EGCG multimodal dosing plan.

Images