Precise typing and dynamic treatment system for type 2 diabetes mellitus

Abstract

The invention relates to the technical field of medical information processing, in particular to a type 2 diabetes precise typing and dynamic treatment system which comprises a clinical feature extraction module, a metabolism dynamic modeling module, a precise typing module, a treatment parameter generation module and a dynamic treatment output module. Wherein the clinical feature extraction module is used for collecting a multi-dimensional clinical data set of a patient for continuous 14 days; the metabolism dynamic modeling module is used for extracting blood glucose fluctuation characteristics and pancreas islet function compensation characteristics; the precise typing module is used for dividing metabolism-driven subtypes to which the patients belong through a clustering algorithm; the treatment parameter generation module is used for mapping a subtype-treatment rule base and outputting a matched personalized treatment parameter combination; and the dynamic treatment output module outputs a dynamic treatment scheme. According to the invention, through a dynamic treatment mechanism based on typing, linkage control of individual difference identification and real-time adjustment of the medication scheme of the type 2 diabetes patients is realized, and the accuracy and adaptability of treatment are improved.

Description

Technical Field

[0001] This invention relates to the field of medical information processing technology, and in particular to a system for precise classification and dynamic treatment of type 2 diabetes. Background Technology

[0002] Currently, type 2 diabetes, a chronic metabolic disease characterized by insulin resistance and impaired β-cell function, is experiencing a continuous increase in the number of patients and a growing management burden. Traditional treatments often rely on fixed doses or empirical drug combinations, failing to fully consider the high heterogeneity among patients in terms of metabolic status, pancreatic function, and blood glucose fluctuations, leading to problems such as unstable blood glucose control, overdose, or slow response in some patients.

[0003] Although some studies have attempted to incorporate blood glucose monitoring and dynamic modeling technologies, most systems have not yet established a complete closed-loop mechanism from data collection, feature modeling, classification identification to personalized treatment output, making it difficult to support refined dynamic intervention. Therefore, there is an urgent need for a precise classification and dynamic treatment system for type 2 diabetes to address these issues. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides a system for precise classification and dynamic treatment of type 2 diabetes.

[0005] The Type 2 Diabetes Precision Classification and Dynamic Treatment System includes a clinical feature extraction module, a metabolic dynamic modeling module, a precision classification module, a treatment parameter generation module, and a dynamic treatment output module; among which: Clinical feature extraction module: used to collect patients' blood glucose time-series data, fasting C-peptide, and insulin resistance index for 14 consecutive days, and output multi-dimensional clinical datasets; Metabolic dynamic modeling module: used to receive multi-dimensional clinical datasets, extract blood glucose fluctuation features and pancreatic islet function compensation features through dynamic neural network model, and generate metabolic dynamic feature vectors; Precise subtyping module: used to receive metabolic dynamic feature vectors and classify patients into metabolic driving subtypes through clustering algorithms, specifically including insulin resistance-dominant type, β-cell exhaustion type and mixed compensatory type; Treatment parameter generation module: It receives metabolic driving subtypes, maps them to a preset subtype-treatment rule library, and outputs a matching personalized treatment parameter combination, including drug type, base dose and sensitivity coefficient; Dynamic treatment output module: Used to receive personalized treatment parameter combinations and generate dynamic treatment plans that include dose adjustment thresholds and follow-up cycles by combining real-time blood glucose data.

[0006] Optionally, the clinical feature extraction module includes a blood glucose acquisition unit, a C-peptide detection unit, an insulin resistance index calculation unit, and a feature integration unit; wherein: Blood glucose collection unit: Used to receive dynamic blood glucose data uploaded by continuously worn blood glucose monitoring devices, continuously collect blood glucose values ​​for 14 days at a frequency of once every 5 minutes, and generate sequence data containing timestamps and blood glucose concentration values; C-peptide detection unit: used to obtain fasting blood samples through a standardized blood collection process on days 1, 7 and 14, and to detect C-peptide concentration using chemiluminescent immunoassay. The average of the three samples is then processed to obtain a stable fasting C-peptide index. Insulin resistance index calculation unit: used to obtain the patient's fasting insulin concentration and fasting blood glucose concentration during the same period, and then calculate the insulin resistance index; Feature integration unit: Used to standardize blood glucose time series data, fasting C-peptide value and insulin resistance index, and integrate them into a structured multi-dimensional clinical dataset after aligning them according to a unified time benchmark.

[0007] Optionally, the metabolic dynamic modeling module includes a feature preprocessing unit, a neural network modeling unit, and a feature encoding unit; wherein: Feature preprocessing unit: It receives multi-dimensional clinical datasets, performs noise reduction, normalization and missing value imputation on blood glucose time series data, and standardizes fasting C-peptide value and insulin resistance index, and outputs the processed clinical dataset. Neural network modeling unit: used to construct a dynamic neural network model with time-dependent features, inputting the processed clinical dataset to extract blood glucose fluctuation features and pancreatic function compensation features; Feature encoding unit: used to encode blood glucose fluctuation features and pancreatic islet function compensation features into a fixed-dimensional metabolic dynamic feature vector, including blood glucose fluctuation amplitude, rhythm changes and pancreatic islet compensation ability.

[0008] Optionally, the neural network modeling unit includes: Time series encoding subunit: Used to receive the clinical dataset output by the feature preprocessing unit, perform time-expanding processing on the blood glucose time series data for 14 consecutive days, and encode the daily data into a time series feature matrix; State memory modeling subunit: used to construct a gated recurrent neural network model, sequentially model the time series feature matrix, and extract the dynamic information state on which blood glucose fluctuation features and pancreatic function compensation features depend; Feature output sub-unit: used to perform pooling processing on the hidden state vectors at all times to extract the overall representative dynamic features, corresponding to blood glucose fluctuation features and pancreatic islet function compensation features, respectively.

[0009] Optionally, the feature encoding unit includes: The fluctuation amplitude calculation subunit is used to calculate the maximum fluctuation range of blood glucose within a continuous monitoring period. It obtains the blood glucose fluctuation amplitude by extracting the difference between the maximum and minimum values ​​in the blood glucose time series. ; The rhythm variation calculation subunit is used to analyze the variation pattern of blood glucose levels in the diurnal cycle, calculate its variability over 24 hours, and extract the amplitude of the dominant frequency component as the rhythm variation value after performing a Fourier transform on the blood glucose values ​​throughout the day. ; Pancreatic islet compensation capacity calculation subunit: used to assess the pancreatic islet function's response to changes in blood glucose, based on fasting C-peptide concentration. The ratio of insulin resistance index to insulin compensatory capacity is used to calculate pancreatic islet compensation capacity. .

[0010] Optionally, the precise typing module includes a feature normalization unit, a clustering analysis unit, and a subtype discrimination unit; wherein: Feature normalization unit: Used to receive the metabolic dynamic feature vector output by the metabolic dynamic modeling module, and perform zero-mean unit variance standardization on the blood glucose fluctuation amplitude, rhythm change value and pancreatic islet compensation capacity. Clustering Analysis Unit: Based on the normalized feature vectors, a feature sample space is constructed. The patient data is then unsupervised clustered using the K-means clustering algorithm with a preset number of clusters. The Euclidean distance between each sample and the cluster center is extracted. The output distance vector set; Subtype discrimination unit: It is used to divide the feature vector into the metabolic drive subtype represented by the nearest cluster center according to the principle of minimum distance, and output the classification result.

[0011] Optionally, the subtype discrimination unit includes: Minimum distance identification subunit: Used to receive the distance vector set output by the cluster analysis unit and identify the cluster center number corresponding to the minimum distance, denoted as: ,in, This represents the cluster center number that is closest to the patient's feature vector; Subtype label assignment subunit: used to assign metabolic dynamic feature vectors to cluster centers numbered as follows The subtype category is determined, and classification labels are output according to preset rules; Rule 1, when The condition is dominated by insulin resistance. Rule 2, At that time Cellular exhaustion type; Rule 3, It is a mixed compensation type.

[0012] Optionally, the treatment parameter generation module includes a subtype identification receiving unit, a rule base query unit, and a parameter combination output unit; wherein: Subtype identification receiving unit: used to receive the patient's metabolic drive subtype label output by the accurate typing module and use it as the basis for parameter generation query; Rule base query unit: used to access a preset subtype-treatment rule base, which uses three metabolic drive subtypes as index fields and stores the corresponding set of recommended treatment parameters; Parameter combination output unit: Used to output the treatment parameter entries obtained from the rule base in a structured manner to form a personalized treatment parameter combination.

[0013] Optionally, the dynamic treatment output module includes a blood glucose monitoring receiving unit, a dose adjustment calculation unit, and a follow-up examination cycle setting unit; wherein: Blood glucose monitoring receiving unit: used to receive blood glucose time-series data uploaded by the patient's continuous blood glucose monitoring device in real time; Dosage adjustment calculation unit: It receives the base dose and sensitivity coefficient in the personalized treatment parameter combination, combines the fluctuation range of real-time blood glucose data, target blood glucose range and blood glucose deviation trend, dynamically determines whether the current dose needs to be adjusted, and generates the corresponding dose adjustment threshold to guide the fine adjustment range of daily dosing dose. Follow-up cycle setting unit: used to individually set the follow-up cycle based on the stability and adjustment response of blood glucose control; When the blood glucose fluctuation exceeds 5 mmol / L within 3 consecutive days, or the difference between blood glucose before and after breakfast and dinner exceeds 4 mmol / L, and the blood glucose does not return to the target range of 4.4–10 mmol / L within 2 days after treatment adjustment, the re-examination cycle is set to 3 days. When the standard deviation of blood glucose is less than 1.5 mmol / L for 7 consecutive days and the average daily blood glucose is stable within the target range, the re-examination cycle is set to 10 days.

[0014] Optionally, the dose adjustment calculation unit includes: The blood glucose deviation assessment subunit is used to receive real-time blood glucose data and calculate the deviation between the patient's current daily average blood glucose value and the target blood glucose value. ; Adjustment range calculation subunit: used to calculate based on blood glucose deviation value Calculate the appropriate dose adjustment value based on the sensitivity coefficient in the personalized treatment parameters. ; Dose threshold generation subunit: used to adjust the dose based on the baseline dose in the personalized treatment parameters and the calculated dose adjustment value. The dynamically adjusted upper and lower limits of the dosing dose are generated and used as the dose adjustment threshold; the formula is: ; ,in, and These represent the lower and upper limits of the dose adjustment range, respectively. This indicates the baseline dose in the personalized treatment parameters.

[0015] The beneficial effects of this invention are: This invention, by constructing a system architecture consisting of five major modules—clinical feature extraction, metabolic dynamic modeling, precise subtyping, treatment parameter generation, and dynamic treatment output—achieves fully automated processing from raw clinical data to personalized treatment plans. It can dynamically generate targeted treatment strategies based on the patient's actual metabolic state, overcoming the shortcomings of traditional approaches that treat the same disease simultaneously and provide static responses.

[0016] This invention achieves automatic identification of metabolic-driven subtypes through clustering algorithms and calculates dosage adjustment thresholds and re-examination cycles by combining real-time blood glucose data, enabling the system to have adaptive adjustment capabilities. This significantly improves the accuracy and flexibility of treatment plans and provides efficient and stable technical support for the long-term refined management of type 2 diabetes. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the diabetes precision classification and dynamic treatment system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the metabolic dynamic modeling module in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0020] It should be noted that the use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0021] Generally, terms can be understood at least partly from their use in context. For example, depending at least partly on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood not necessarily to convey an exclusive set of factors, but rather, alternatively, depending at least partly on the context, to allow for the presence of other factors that are not necessarily explicitly described.

[0022] like Figures 1-2 As shown, the Type 2 Diabetes Precision Classification and Dynamic Treatment System includes a clinical feature extraction module, a metabolic dynamic modeling module, a precision classification module, a treatment parameter generation module, and a dynamic treatment output module; among which: Clinical feature extraction module: used to collect patients' blood glucose time-series data, fasting C-peptide, and insulin resistance index for 14 consecutive days, and output multi-dimensional clinical datasets; Metabolic dynamic modeling module: used to receive multi-dimensional clinical datasets, extract blood glucose fluctuation features and pancreatic islet function compensation features through dynamic neural network model, and generate metabolic dynamic feature vectors; Precise subtyping module: used to receive metabolic dynamic feature vectors and classify patients into metabolic driving subtypes through clustering algorithms, specifically including insulin resistance-dominant type, β-cell exhaustion type and mixed compensatory type; Treatment parameter generation module: It receives metabolic driving subtypes, maps them to a preset subtype-treatment rule library, and outputs a matching personalized treatment parameter combination, including drug type, base dose and sensitivity coefficient; Dynamic treatment output module: Used to receive personalized treatment parameter combinations and generate dynamic treatment plans that include dose adjustment thresholds and follow-up cycles by combining real-time blood glucose data.

[0023] The clinical feature extraction module includes a blood glucose acquisition unit, a C-peptide detection unit, an insulin resistance index calculation unit, and a feature integration unit; among which: Blood glucose collection unit: Used to receive dynamic blood glucose data uploaded by continuously worn blood glucose monitoring devices, continuously collect blood glucose values ​​for 14 days at a frequency of once every 5 minutes, and generate sequence data containing timestamps and blood glucose concentration values; C-peptide detection unit: used to obtain fasting blood samples through a standardized blood collection process on days 1, 7 and 14, and to detect C-peptide concentration using chemiluminescent immunoassay. The average of the three samples is then processed to obtain a stable fasting C-peptide index. Insulin Resistance Index Calculation Unit: Used to obtain the patient's fasting insulin concentration and fasting blood glucose concentration during the same period, and then calculate the insulin resistance index. The formula is as follows: ,in, Indicator of insulin resistance, Indicates fasting insulin concentration. Indicates fasting blood glucose concentration; The feature integration unit is used to standardize blood glucose time-series data, fasting C-peptide values, and insulin resistance index, align them according to a unified time benchmark, integrate them into a structured, multi-dimensional clinical dataset, and output it to the metabolic dynamic modeling module for feature modeling. The above sub-units achieve standardized extraction of key clinical features of type 2 diabetes patients through explicit sensor data acquisition, biochemical index measurement, and index calculation, ensuring the stability and integrity of the input data for subsequent modeling processes, thereby providing a high-quality data foundation for subsequent metabolic dynamic modeling.

[0024] The metabolic dynamic modeling module includes a feature preprocessing unit, a neural network modeling unit, and a feature encoding unit; among which: Feature preprocessing unit: It receives multi-dimensional clinical datasets, performs noise reduction, normalization and missing value imputation on blood glucose time series data, and standardizes fasting C-peptide value and insulin resistance index, and outputs the processed clinical dataset. Neural network modeling unit: used to construct a dynamic neural network model with time-dependent features, inputting the processed clinical dataset to extract blood glucose fluctuation features and pancreatic function compensation features; Feature encoding unit: used to encode blood glucose fluctuation features and pancreatic islet function compensation features into a fixed-dimensional metabolic dynamic feature vector, including blood glucose fluctuation amplitude, rhythm changes and pancreatic islet compensation ability, and used as input for subsequent accurate typing module.

[0025] The neural network modeling unit includes: The time-series encoding subunit receives the clinical dataset output by the feature preprocessing unit, performs time-expanding processing on 14 consecutive days of blood glucose time-series data, and encodes daily data into a time-series feature matrix for subsequent modeling; the corresponding expression is: ,in, Represents the feature matrix of the time series; Indicates the first The input feature vector of the day; This indicates the number of time steps, which is 14. State memory modeling subunit: used to construct a gated recurrent neural network (GRU) model, sequentially modeling the time series feature matrix, and extracting the dynamic information state upon which blood glucose fluctuation features and pancreatic function compensation features depend; the corresponding formula is: ,in, This represents the hidden state vector at the current moment; This represents the input feature vector at the current moment; This represents the hidden state vector from the previous time step. The computation function representing the gated loop unit; Feature output subunit: used to perform pooling processing on the hidden state vectors at all times to extract representative dynamic features, corresponding to blood glucose fluctuation features and pancreatic islet function compensation features respectively; through the above subunit design, the neural network modeling unit can extract the metabolic response process of individual patients from the temporal structure, enhance the model's ability to model the dynamic changes in blood glucose and the changes in pancreatic islet function, and improve the accuracy of feature extraction.

[0026] Feature coding units include: The fluctuation amplitude calculation subunit is used to calculate the maximum fluctuation range of blood glucose within a continuous monitoring period. It obtains the blood glucose fluctuation amplitude by extracting the difference between the maximum and minimum values ​​in the blood glucose time series. The calculation formula is: ,in, Indicates the range of blood glucose fluctuations; Indicates the first Blood glucose level at any given time; and These represent the maximum and minimum blood glucose levels over a 14-day period, respectively. The rhythm variation calculation subunit is used to analyze the variation pattern of blood glucose levels in the diurnal cycle, calculate its variability over 24 hours, and extract the amplitude of the dominant frequency component as the rhythm variation value after performing a Fourier transform on the blood glucose values ​​throughout the day. The formula is: ,in, Indicates the value of rhythm variation. Fourier transform representing the blood glucose time series signal; Indicates extraction of main frequency The amplitude at that point; Pancreatic islet compensation capacity calculation subunit: used to assess the pancreatic islet function's response to changes in blood glucose, based on fasting C-peptide concentration. The ratio of insulin resistance index to insulin compensatory capacity is used to calculate pancreatic islet compensation capacity. The formula is: ,in, Indicates the compensatory capacity of the pancreas; Indicates fasting C-peptide concentration; This represents the insulin resistance index. The indicators extracted from the above three sub-units encode the fluctuation amplitude A, rhythm change value R, and pancreatic islet compensation capacity D into a fixed-dimensional metabolic dynamic feature vector, which helps to quantify the patient's metabolic dynamic state and provides a highly expressive input basis for subsequent precise subtyping and individualized treatment.

[0027] The precise typing module includes a feature normalization unit, a clustering analysis unit, and a subtype discrimination unit; among which: Feature normalization unit: It receives the metabolic dynamic feature vector output by the metabolic dynamic modeling module, and performs zero-mean unit variance normalization on the blood glucose fluctuation amplitude, rhythm change value and pancreatic islet compensation capacity to ensure that the features of each dimension have a consistent dimension and weight basis in the clustering process. Clustering Analysis Unit: Based on the normalized feature vectors, a feature sample space is constructed. The patient data is then unsupervised clustered using the K-means clustering algorithm with a preset number of clusters. The Euclidean distance between each sample and the cluster center is extracted. The output distance vector set is shown in the following formula: ,in, Representing the eigenvector To the Cluster centers The Euclidean distance; This represents the patient's metabolic dynamic feature vector; Indicates the first Individual subtype cluster centers; Represents the L2 norm; Subtype discrimination unit: It is used to divide the feature vector into the metabolic driving subtype represented by the nearest cluster center according to the principle of minimum distance, and output the classification result. Among them, the three cluster centers correspond to the insulin resistance-dominated type, β-cell exhaustion type and mixed compensation type, respectively. Through the above structure, patients can be stably classified into pathological subtypes dominated by specific metabolic mechanisms based on their metabolic dynamic characteristics, providing a clear classification basis for subsequent formulation of personalized treatment parameters.

[0028] Subtype discrimination unit includes: Minimum distance identification subunit: Used to receive the distance vector set output by the cluster analysis unit and identify the cluster center number corresponding to the minimum distance, denoted as: ,in, This represents the cluster center number that is closest to the patient's feature vector; Subtype label assignment subunit: used to assign metabolic dynamic feature vectors to cluster centers numbered as follows The subtype category is determined, and classification labels are output according to preset rules; Rule 1, when The condition is dominated by insulin resistance. Rule 2, At that time Cellular exhaustion type; Rule 3, The subtype is a hybrid compensatory type. Through direct identification and label mapping of the minimum distance, the subtype discrimination unit achieves accurate classification of metabolic-driven subtypes, ensuring that the subtyping module has the characteristics of clear judgment basis and high execution efficiency in structure, which is conducive to improving the accuracy of subsequent treatment plan matching.

[0029] The treatment parameter generation module includes a subtype identification receiving unit, a rule base query unit, and a parameter combination output unit; wherein: Subtype identification receiving unit: used to receive the patient's metabolic drive subtype label output by the accurate typing module and use it as the basis for parameter generation query; Rule base query unit: used to access the preset subtype treatment rule base, which uses the patient's metabolic driving subtype and weight stratification (overweight / obese, normal, underweight) as index fields and stores the corresponding recommended drug types and basic dosage standards; For patients with insulin resistance as the primary risk factor, when they are overweight or obese, medications that combine weight loss and improved insulin sensitivity are preferred, including GLP-1 receptor agonists (liraglutide, smegglutide) or SGLT2 inhibitors (dapagliflozin). The recommended dosage range is calculated based on the patient's weight as follows: ,in, The range is 0.02-0.05 mg / kg·d. For the patient's weight; When body weight is normal or low, medications that improve insulin sensitivity, including metformin or pioglitazone, should be given priority. The baseline dose is calculated based on body weight as follows: The range is 15–25 mg / kg·d; for For patients with cellular exhaustion, insulin pump therapy is recommended. The pump rate and dosage are automatically adjusted based on continuous glucose monitoring data. The baseline rate is set as follows: ,in, The range is 0.3–0.6 U / kg·d, and the pump speed is dynamically adjusted according to the blood glucose deviation value through the feedback control module; For patients with mixed compensatory type: Metformin combined with low-dose insulin therapy is recommended.

[0030] Table 1 Subtype-Treatment Rule Base In Table 1 above, the metabolic drive subtype is identified by the classification label output by the precise subtyping module, which distinguishes different pathogenesis mechanisms; the drug type refers to the recommended drug regimen for different subtypes, which can be a single drug or a combination therapy; the basal dose refers to the recommended initial dose, expressed in mg (oral) or units (injection), which is the reference starting point for personalized treatment adjustments; the sensitivity coefficient characterizes the sensitivity of the patient's blood glucose response to changes in drug dosage and is used to calculate the dose adjustment threshold in subsequent dynamic treatment plans.

[0031] The parameter combination output unit is used to output the treatment parameter entries obtained from the rule base in a structured manner, forming a personalized treatment parameter combination. Through the above module design, the treatment parameter generation module can quickly map and match the personalized treatment strategy according to the patient's specific metabolic drive subtype, ensuring that the generated results are highly consistent with the pathological mechanism, and providing clear and structured parameter support for subsequent dynamic treatment.

[0032] The dynamic treatment output module includes a blood glucose monitoring receiving unit, a dose adjustment calculation unit, and a follow-up examination cycle setting unit; among which: Blood glucose monitoring receiving unit: used to receive blood glucose time-series data uploaded by the patient's continuous blood glucose monitoring device in real time, and update and cache it to ensure that subsequent parameter adjustments are based on the latest blood glucose trend; Dosage adjustment calculation unit: It receives the base dose and sensitivity coefficient in the personalized treatment parameter combination, combines the fluctuation range of real-time blood glucose data, target blood glucose range and blood glucose deviation trend, dynamically determines whether the current dose needs to be adjusted, and generates the corresponding dose adjustment threshold to guide the fine adjustment range of daily dosing dose. Follow-up cycle setting unit: used to individually set the follow-up cycle based on the stability and adjustment response of blood glucose control; When the blood glucose fluctuation exceeds 5 mmol / L within 3 consecutive days, or the difference between blood glucose before and after breakfast and dinner exceeds 4 mmol / L, and the blood glucose does not return to the target range of 4.4–10 mmol / L within 2 days after treatment adjustment, the re-examination cycle is set to 3 days. When the standard deviation of blood glucose is less than 1.5 mmol / L for 7 consecutive days and the average daily blood glucose is stable within the target range, the re-examination cycle is set to 10 days. Through the above module structure, the dynamic treatment output module can generate a highly responsive and adaptable dynamic treatment plan based on the received personalized basic parameters and blood glucose monitoring results, so as to realize the continuous individualized adjustment of medication dosage and re-examination frequency, which helps to improve the dynamic control ability in the treatment of type 2 diabetes.

[0033] The dose adjustment calculation unit includes: The blood glucose deviation assessment subunit is used to receive real-time blood glucose data and calculate the deviation between the patient's current daily average blood glucose value and the target blood glucose value. The formula is: ,in, Indicates the blood glucose deviation value; This indicates the current average daily blood glucose concentration; This indicates the preset target blood glucose level; Adjustment range calculation subunit: used to calculate based on blood glucose deviation value Calculate the appropriate dose adjustment value based on the sensitivity coefficient in the personalized treatment parameters. The formula is: ,in, This indicates the recommended dose adjustment value; Indicates the sensitivity coefficient; Dose threshold generation subunit: used to adjust the dose based on the baseline dose in the personalized treatment parameters and the calculated dose adjustment value. The dynamically adjusted upper and lower limits of the dosing dose are generated and used as the dose adjustment threshold; the formula is: ; ,in, and These represent the lower and upper limits of the dose adjustment range, respectively. This represents the baseline dose in personalized treatment parameters. Through the above three sub-units, the dose adjustment calculation sub-unit can flexibly adjust the dosage according to the deviation of blood glucose from the target, output a reasonable dose variation range, and enhance the adaptability and individualized control of the treatment plan.

[0034] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0035] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

A precise classification and dynamic treatment system for type 1.2 diabetes is characterized by, It includes a clinical feature extraction module, a metabolic dynamic modeling module, a precise subtyping module, a treatment parameter generation module, and a dynamic treatment output module; among which: Clinical feature extraction module: used to collect patients' blood glucose time-series data, fasting C-peptide, and insulin resistance index for 14 consecutive days, and output multi-dimensional clinical datasets; Metabolic dynamic modeling module: used to receive multi-dimensional clinical datasets, extract blood glucose fluctuation features and pancreatic islet function compensation features through dynamic neural network model, and generate metabolic dynamic feature vectors; Precise subtyping module: used to receive metabolic dynamic feature vectors and classify patients into metabolic driving subtypes through clustering algorithms, specifically including insulin resistance-dominant type, β-cell exhaustion type and mixed compensatory type; Treatment parameter generation module: It receives metabolic driving subtypes, maps them to a preset subtype-treatment rule library, and outputs a matching personalized treatment parameter combination, including drug type, base dose and sensitivity coefficient; Dynamic treatment output module: Used to receive personalized treatment parameter combinations and generate dynamic treatment plans that include dose adjustment thresholds and follow-up cycles by combining real-time blood glucose data.

2. The type 2 diabetes precision classification and dynamic treatment system according to claim 1, characterized in that, The clinical feature extraction module includes a blood glucose acquisition unit, a C-peptide detection unit, an insulin resistance index calculation unit, and a feature integration unit; wherein: Blood glucose collection unit: Used to receive dynamic blood glucose data uploaded by continuously worn blood glucose monitoring devices, continuously collect blood glucose values ​​for 14 days at a frequency of once every 5 minutes, and generate sequence data containing timestamps and blood glucose concentration values; C-peptide detection unit: used to obtain fasting blood samples through a standardized blood collection process on days 1, 7 and 14, and to detect C-peptide concentration using chemiluminescent immunoassay. The average of the three samples is then processed to obtain a stable fasting C-peptide index. Insulin resistance index calculation unit: used to obtain the patient's fasting insulin concentration and fasting blood glucose concentration during the same period, and then calculate the insulin resistance index; Feature integration unit: Used to standardize blood glucose time series data, fasting C-peptide value and insulin resistance index, and integrate them into a structured multi-dimensional clinical dataset after aligning them according to a unified time benchmark.

3. The type 2 diabetes precision classification and dynamic treatment system according to claim 1, characterized in that, The metabolic dynamic modeling module includes a feature preprocessing unit, a neural network modeling unit, and a feature encoding unit; wherein: Feature preprocessing unit: It receives multi-dimensional clinical datasets, performs noise reduction, normalization and missing value imputation on blood glucose time series data, and standardizes fasting C-peptide value and insulin resistance index, and outputs the processed clinical dataset. Neural network modeling unit: used to construct a dynamic neural network model with time-dependent features, inputting the processed clinical dataset to extract blood glucose fluctuation features and pancreatic islet function compensation features; Feature encoding unit: used to encode blood glucose fluctuation features and pancreatic islet function compensation features into a fixed-dimensional metabolic dynamic feature vector, including blood glucose fluctuation amplitude, rhythm changes and pancreatic islet compensation capacity.

4. The type 2 diabetes precision classification and dynamic treatment system according to claim 3, characterized in that, The neural network modeling unit includes: Time series encoding subunit: Used to receive the clinical dataset output by the feature preprocessing unit, perform time-expanding processing on the blood glucose time series data for 14 consecutive days, and encode the daily data into a time series feature matrix; State memory modeling subunit: used to construct a gated recurrent neural network model, sequentially model the time series feature matrix, and extract the dynamic information state on which blood glucose fluctuation features and pancreatic function compensation features depend; Feature output sub-unit: used to perform pooling processing on the hidden state vectors at all times to extract the overall representative dynamic features, corresponding to blood glucose fluctuation features and pancreatic function compensation features, respectively.

5. The type 2 diabetes precision classification and dynamic treatment system according to claim 3, characterized in that, The feature encoding unit includes: The fluctuation amplitude calculation subunit is used to calculate the maximum fluctuation range of blood glucose within a continuous monitoring period. It obtains the blood glucose fluctuation amplitude by extracting the difference between the maximum and minimum values ​​in the blood glucose time series. ; The rhythm variation calculation subunit is used to analyze the variation pattern of blood glucose levels in the diurnal cycle, calculate its variability over 24 hours, and extract the amplitude of the dominant frequency component as the rhythm variation value after performing a Fourier transform on the blood glucose values ​​throughout the day. ; Pancreatic islet compensation capacity calculation subunit: used to assess the pancreatic islet function's response to changes in blood glucose, based on fasting C-peptide concentration. The ratio of insulin resistance index to insulin compensatory capacity is used to calculate pancreatic islet compensation capacity. .

6. The type 2 diabetes precision classification and dynamic treatment system according to claim 1, characterized in that, The precise typing module includes a feature normalization unit, a clustering analysis unit, and a subtype discrimination unit; wherein: Feature normalization unit: Used to receive the metabolic dynamic feature vector output by the metabolic dynamic modeling module, and perform zero-mean unit variance standardization on the blood glucose fluctuation amplitude, rhythm change value and pancreatic islet compensation capacity. Clustering Analysis Unit: Based on the normalized feature vectors, a feature sample space is constructed. The patient data is then unsupervised clustered using the K-means clustering algorithm with a preset number of clusters. The Euclidean distance between each sample and the cluster center is extracted. The output distance vector set; Subtype discrimination unit: It is used to divide the feature vector into the metabolic driving subtype represented by the nearest cluster center according to the principle of minimum distance, and output the classification result.

7. The type 2 diabetes precision classification and dynamic treatment system according to claim 6, characterized in that, The subtype discrimination unit includes: Minimum distance identification subunit: Used to receive the distance vector set output by the cluster analysis unit and identify the cluster center number corresponding to the minimum distance, denoted as: ,in, This represents the cluster center number that is closest to the patient's feature vector; Subtype label assignment subunit: used to assign metabolic dynamic feature vectors to cluster centers numbered as follows The subtype category is determined, and classification labels are output according to preset rules; Rule 1, when The condition is dominated by insulin resistance. Rule 2, At that time Cellular exhaustion type; Rule 3, It is a mixed compensation type.

8. The type 2 diabetes precision classification and dynamic treatment system according to claim 1, characterized in that, The treatment parameter generation module includes a subtype identification receiving unit, a rule base query unit, and a parameter combination output unit; wherein: Subtype identification receiving unit: used to receive the patient's metabolic drive subtype label output by the accurate typing module and use it as the basis for parameter generation query; Rule base query unit: used to access a preset subtype-treatment rule base, which uses three metabolic drive subtypes as index fields and stores the corresponding set of recommended treatment parameters; Parameter combination output unit: Used to output the treatment parameter entries obtained from the rule base in a structured manner, forming a personalized treatment parameter combination.

9. The type 2 diabetes precision classification and dynamic treatment system according to claim 1, characterized in that, The dynamic treatment output module includes a blood glucose monitoring receiving unit, a dose adjustment calculation unit, and a follow-up examination cycle setting unit; wherein: Blood glucose monitoring receiving unit: used to receive blood glucose time-series data uploaded by the patient's continuous blood glucose monitoring device in real time; Dosage adjustment calculation unit: It receives the base dose and sensitivity coefficient in the personalized treatment parameter combination, combines the fluctuation range of real-time blood glucose data, target blood glucose range and blood glucose deviation trend, dynamically determines whether the current dose needs to be adjusted, and generates the corresponding dose adjustment threshold to guide the fine adjustment range of daily dosing dose. Follow-up cycle setting unit: used to individually set the follow-up cycle based on the stability and adjustment response of blood glucose control; When the blood glucose fluctuation exceeds 5 mmol / L within 3 consecutive days, or the difference between blood glucose before and after breakfast and dinner exceeds 4 mmol / L, and the blood glucose does not return to the target range of 4.4–10 mmol / L within 2 days after treatment adjustment, the re-examination cycle is set to 3 days. When the standard deviation of blood glucose is less than 1.5 mmol / L for 7 consecutive days and the average daily blood glucose is stable within the target range, the re-examination cycle is set to 10 days.

10. The type 2 diabetes precision classification and dynamic treatment system according to claim 9, characterized in that, The dose adjustment calculation unit includes: The blood glucose deviation assessment subunit is used to receive real-time blood glucose data and calculate the deviation between the patient's current daily average blood glucose value and the target blood glucose value. ; Adjustment range calculation subunit: used to calculate based on blood glucose deviation value Calculate the appropriate dose adjustment value based on the sensitivity coefficient in the personalized treatment parameters. ; Dose threshold generation subunit: used to adjust the dose based on the baseline dose in the personalized treatment parameters and the calculated dose adjustment value. The dynamically adjusted upper and lower limits of the dosing dose are generated and used as the dose adjustment threshold; the formula is: ; ,in, and These represent the lower and upper limits of the dose adjustment range, respectively. This indicates the baseline dose in personalized treatment parameters.

Images