Skin noninvasive drug delivery control method and system based on big data analysis

Abstract

The invention discloses a non-invasive skin drug delivery control method and system based on big data analysis, and the method comprises the steps: collecting multi-dimensional data, such as skin thickness, metabolic capability, allergic constitution, electrical impedance value, skin temperature and humidity, through a physiological parameter sensor array, and carrying out the classification and recognition of individual physiological features through a clustering algorithm. When individual parameters exceed a preset penetration threshold value, a real-time monitoring module is automatically activated, the drug administration intensity adjustment direction is judged by analyzing electrical impedance value fluctuation characteristics, a penetration depth prediction model is constructed by fusing skin temperature data, and deviation feedback optimization of a prediction value and an actual monitoring value is achieved. Allergic physical characteristics are further extracted and combined with metabolic capacity data to construct a dynamic regulation and control sequence, personalized administration intensity parameters are generated to drive an actuator to adjust equipment output, continuous and accurate permeation rate control is achieved through loop iterative optimization, the contradiction between individual differential administration requirements and equipment standardized output is effectively solved, and the method is suitable for large-scale popularization and application. The safety and the effectiveness of noninvasive administration are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of non-invasive skin drug delivery technology, and in particular discloses a non-invasive skin drug delivery control method and system based on big data analysis. Background Technology

[0002] Non-invasive transdermal drug delivery technology, as an important tool in modern medical and aesthetic care, avoids the pain and infection risks associated with injections, and has irreplaceable value in the treatment of local diseases and drug delivery. This technology promotes drug penetration through the skin barrier using physical or chemical methods, allowing it to act directly on the target site and providing patients with a more comfortable and safer treatment experience.

[0003] Current non-invasive drug delivery protocols generally suffer from fixed dosing parameters. Most devices use preset standardized parameters for drug delivery, making it impossible to adjust them according to the physiological characteristics of individual patients. This standardized dosing mode ignores significant differences between individuals in skin thickness, metabolic capacity, and allergic constitution, leading to drastically different treatment responses in different patients despite using the same dosing regimen. Some patients may experience reduced efficacy due to insufficient drug penetration, while others may experience local irritation due to excessive drug penetration.

[0004] The root of this problem lies in the lack of real-time sensing capabilities for dynamic changes in the skin barrier. Skin physiological parameters such as impedance, temperature, and humidity continuously change during drug administration, directly affecting the drug's penetration rate and depth. However, current technology cannot capture these real-time changes, let alone adjust the dosage accordingly. This lack of sensing capability further leads to blind spots in the drug administration process. Operators cannot accurately determine whether the drug has penetrated to the expected skin depth, nor can they determine whether the current dosage is suitable for a specific patient. For example, during non-invasive administration of local anesthetics, if changes in skin condition cannot be monitored in real time, insufficient anesthetic depth may affect subsequent treatment, or an overdose may result in prolonged local numbness.

[0005] Therefore, establishing a precise identification mechanism for individual patient characteristics and achieving real-time monitoring and dynamic control of skin condition during drug administration has become a key issue in improving the accuracy and safety of non-invasive drug administration. Summary of the Invention

[0006] This invention provides a non-invasive skin drug delivery control method and system based on big data analysis, aiming to solve at least one of the defects existing in the prior art.

[0007] One aspect of the present invention relates to a non-invasive skin drug delivery control method based on big data analysis, comprising the following steps: S100. Collect individual physiological parameter data through a physiological parameter sensor array, and use a clustering algorithm to group the physiological parameter data to obtain individual physiological identification and classification results. The physiological parameter data includes skin thickness data, metabolic capacity data, allergy constitution data, electrical impedance data, skin temperature data, and skin humidity data. S200. The individual physiological identification and classification results are compared with the preset penetration rate threshold and penetration depth threshold. If the individual physiological identification and classification results show that the physiological parameters exceed the preset penetration rate threshold and penetration depth threshold, the real-time monitoring module is activated to continuously track the skin humidity change trend. S300: Extract the impedance value fluctuation features from the skin humidity change trend obtained from the real-time monitoring module, classify the impedance value fluctuation features using a classification algorithm, and determine the direction of drug administration intensity adjustment; S400: Based on the direction of drug administration intensity adjustment, a penetration depth prediction model is generated by integrating skin temperature data to obtain the predicted penetration depth value. The deviation between the predicted penetration depth value and the actual monitored penetration depth is calculated. If the deviation exceeds the preset threshold, the grouping parameters of the clustering algorithm are updated through the feedback loop to determine the optimized individual physiological identification and classification results. S500: Extract allergy-related features from the optimized individual physiological identification and classification results, and construct a dynamic regulation sequence by combining metabolic capacity data to obtain the drug administration intensity sequence value; S600: The actuator is driven by the drug delivery intensity sequence value to adjust the output parameters of the non-invasive drug delivery device. The above process is iterated by integrating real-time monitoring data to obtain a continuously optimized permeation rate control sequence.

[0008] Further, step S100 includes: S110. Acquire raw values ​​of six physiological parameters—skin thickness, metabolic capacity, allergy status, electrical impedance, skin temperature, and skin humidity—through a physiological parameter sensor array at a preset data acquisition frequency. Perform noise filtering on the raw values ​​of physiological parameters according to sensor accuracy standards to obtain a standardized physiological parameter dataset. S120. Use a data preprocessor to perform correlation verification on the skin thickness data and electrical impedance data in the standardized physiological parameter dataset. If the skin thickness value in the skin thickness data exceeds the preset threshold range, adjust the weight coefficient of the electrical impedance value. Normalize the skin temperature data and skin humidity data according to individual differences to obtain the corrected parameter matrix. S130. The corrected parameter matrix is ​​calculated in multiple dimensions by using a clustering algorithm. If the difference between the metabolic capacity index and the allergic constitution index is less than the classification threshold, the individuals are classified into the same cluster. The optimal number of clusters is determined according to the clustering accuracy standard to obtain the physiological characteristic clustering results. S140. Generate an individual physiological identification mapping table based on the cluster center point coordinates and boundary range in the physiological feature clustering results. If the newly input physiological parameter data falls within the boundary range of the cluster, it is determined that the individual belongs to the corresponding physiological type classification, and the individual physiological identification classification result is obtained.

[0009] Further, step S200 includes: S210. The data preprocessor compares the individual physiological identification and classification results with the preset penetration rate threshold and penetration depth threshold. If any parameter in the individual physiological identification and classification results exceeds the corresponding threshold, a trigger signal is generated to activate the real-time monitoring module. S220. The skin humidity data is continuously sampled through the real-time monitoring module, and the rate of change of skin humidity data is calculated by the sliding window algorithm to obtain the trend of skin humidity change.

[0010] Further, step S300 includes: S310. Based on the skin humidity change trend obtained by the real-time monitoring module, the impedance value fluctuation feature is extracted using a signal processor. If the frequency of the impedance value fluctuation feature is within a preset range, a first feature vector is generated. S320. The first feature vector is classified using a feature classifier. If the magnitude of the first feature vector shows an increasing trend, it is labeled as the first category; if it shows a decreasing trend, it is labeled as the second category, thus obtaining the classification result. S330. The decision-making unit matches the classification results with the preset drug administration intensity adjustment rules. If it is the first category, a reduction instruction is generated; if it is the second category, an increase instruction is generated to determine the direction of drug administration intensity adjustment.

[0011] Further, step S400 includes: S410. Based on the direction of drug delivery intensity adjustment, skin temperature data is integrated to generate a penetration depth prediction model; S420. The individual skin is processed using a penetration depth prediction model to generate a predicted penetration depth value. S430. The predicted penetration depth value is compared with the actual penetration depth value obtained by the real-time monitoring module using the deviation calculator. If the deviation between the predicted penetration depth value and the obtained actual penetration depth value exceeds a preset threshold, a deviation correction signal is generated. S440. Based on the deviation correction signal, the grouping parameters of the clustering algorithm are updated using a feedback loop to obtain optimized individual physiological identification and classification results.

[0012] Further, step S500 includes: S510. Receive the optimized individual physiological identification and classification results, and use a feature extractor to filter out physiological marker parameters related to allergic constitution from the individual physiological identification and classification results. If the individual allergic constitution score exceeds the preset threshold, obtain the corresponding allergic reaction intensity index. S520. Obtain individual drug metabolism rate data through a metabolic capacity monitoring device, and construct a dynamic regulation sequence based on the allergic reaction intensity index and the drug metabolism rate data. S530. The dosing interval for individuals is calculated using a dynamic regulation sequence, and the dosing intensity trend data is determined based on the drug metabolism rate data. S540: Receive drug administration intensity trend data through a sequence optimization processor, perform temporal arrangement and intensity calibration of the dynamic regulation sequence, and obtain drug administration intensity sequence values.

[0013] Further, step S600 includes: S610. Based on the drug delivery intensity sequence value, the output frequency and output intensity of the non-invasive drug delivery device are dynamically adjusted through the actuator control module to determine the real-time adjusted permeation rate parameter. S620. If the real-time adjusted permeation rate parameter is inconsistent with the preset physiological safety threshold, the real-time monitoring data will be analyzed for deviation through the feedback regulator to obtain the deviation correction sequence. S630. Using a deviation correction sequence, the dosing intensity sequence value is iteratively optimized by a time-series optimization processor to obtain a continuously optimized permeation rate control sequence.

[0014] Another aspect of the present invention relates to a skin-based non-invasive drug delivery control system based on big data analysis, for executing the above-described skin-based non-invasive drug delivery control method based on big data analysis, comprising: The individual physiological identification and classification result acquisition module is used to collect individual physiological parameter data through a physiological parameter sensor array, and to group the physiological parameter data using a clustering algorithm to obtain individual physiological identification and classification results. The physiological parameter data includes skin thickness data, metabolic capacity data, allergy constitution data, electrical impedance data, skin temperature data, and skin humidity data. The skin humidity change trend acquisition module is used to compare the individual physiological identification and classification results with the preset penetration rate threshold and penetration depth threshold. If the individual physiological identification and classification results show that the physiological parameters exceed the preset penetration rate threshold and penetration depth threshold, the real-time monitoring module is activated to continuously track the skin humidity change trend. The drug administration intensity adjustment direction determination module is used to extract the impedance value fluctuation characteristics from the skin humidity change trend obtained by the real-time monitoring module, and use a classification algorithm to classify the impedance value fluctuation characteristics to determine the drug administration intensity adjustment direction. The individual physiological identification and classification result optimization module is used to adjust the direction of drug administration intensity, integrate skin temperature data to generate a penetration depth prediction model, obtain the predicted penetration depth value, calculate the deviation between the predicted penetration depth value and the actual monitored penetration depth, and if the deviation exceeds the preset threshold, update the grouping parameters of the clustering algorithm through the feedback loop to determine the optimized individual physiological identification and classification result. The drug administration intensity sequence value acquisition module is used to extract allergy-related features from the optimized individual physiological identification and classification results, and construct a dynamic regulation sequence by combining metabolic capacity data to obtain the drug administration intensity sequence value. The permeation rate control acquisition module is used to drive the actuator to adjust the output parameters of the non-invasive drug delivery device by using the drug delivery intensity sequence value. By integrating real-time monitoring data and iteratively repeating the above process, a continuously optimized permeation rate control sequence is obtained.

[0015] The beneficial effects achieved by this invention are as follows: This invention discloses a non-invasive skin drug delivery control method and system based on big data analysis. Addressing the problem that traditional drug delivery methods cannot achieve precise drug delivery intensity control based on individual differences, this invention collects multi-dimensional data such as skin thickness, metabolic capacity, allergic constitution, electrical impedance value, skin temperature, and humidity through a physiological parameter sensor array. Clustering algorithms are used to classify and identify individual physiological characteristics. When individual parameters exceed a preset penetration threshold, a real-time monitoring module is automatically activated. By analyzing the fluctuation characteristics of electrical impedance value, the direction of drug delivery intensity adjustment is determined. A penetration depth prediction model is constructed by integrating skin temperature data, achieving feedback optimization of the deviation between predicted and actual monitoring values. This invention further extracts allergic constitution characteristics and combines them with metabolic capacity data to construct a dynamic control sequence, generating personalized drug delivery intensity parameters to drive the actuator to adjust the device output. Through iterative optimization, continuous and precise penetration rate control is achieved, effectively resolving the contradiction between individualized drug delivery needs and standardized device output, significantly improving the safety and effectiveness of non-invasive drug delivery. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an embodiment of a non-invasive skin drug delivery control method based on big data analysis according to the present invention. Figure 2 This is a functional block diagram of an embodiment of a non-invasive skin drug delivery control system based on big data analysis according to the present invention.

[0017] Explanation of icon numbers: 10. Individual physiological identification and classification result acquisition module; 20. Skin humidity change trend acquisition module; 30. Drug administration intensity adjustment direction judgment module; 40. Individual physiological identification and classification result optimization module; 50. Drug administration intensity sequence value acquisition module; 60. Permeation rate control acquisition module. Detailed Implementation

[0018] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0019] like Figure 1 As shown, the first embodiment of the present invention proposes a non-invasive skin drug delivery control method based on big data analysis, comprising the following steps: Step S100: Collect individual physiological parameter data through a physiological parameter sensor array, and use a clustering algorithm to group the physiological parameter data to obtain individual physiological identification and classification results. The physiological parameter data includes skin thickness data, metabolic capacity data, allergy constitution data, electrical impedance data, skin temperature data, and skin humidity data.

[0020] A flexible integrated sensor array (including multiple types of sensors such as ultrasound, electrochemical, and bioimpedance) is used to simultaneously collect six key physiological parameters on the skin in the drug delivery area: skin thickness data reflecting skin structure, metabolic capacity data reflecting drug metabolism, allergic constitution data reflecting allergy risk, electrical impedance data reflecting skin barrier function, skin temperature data reflecting local microcirculation, and skin humidity data reflecting the hydration state of the stratum corneum. The collected parameters are standardized and preprocessed (to eliminate dimensional differences) to construct a 12-dimensional physiological feature vector. An improved K-means clustering algorithm (based on the elbow rule to determine the optimal number of categories) is used to classify individuals into five significantly different physiological types (such as "strong barrier-slow metabolizer" and "weak barrier-fast metabolizer"). The final output includes the individual's category and the deviation of key parameters, realizing refined management by "grouping by physiological characteristics".

[0021] Step S200: Compare the individual physiological identification and classification results with the preset penetration rate threshold and penetration depth threshold. If the individual physiological identification and classification results show that the physiological parameters exceed the preset penetration rate threshold and penetration depth threshold, then activate the real-time monitoring module to continuously track the skin humidity change trend.

[0022] Based on the individual physiological identification and classification results generated in step S100 (such as "strong barrier - slow metabolizer" and its parameter characteristics), a preset penetration threshold matrix is ​​invoked (the matrix rows correspond to the physiological type, and the columns contain the safe range of penetration rate and the safe range of penetration depth). The deviation between the key physiological parameters (impedance, skin humidity, etc.) in the classification results and the corresponding type threshold is calculated using Euclidean distance. If the deviation exceeds the critical value (such as rate deviation > 30% or depth deviation > 0.2mm), it is judged as "permeability abnormal risk". The real-time monitoring module is immediately activated. This real-time monitoring module increases the skin humidity sampling frequency to 10Hz, generates a continuous humidity change curve (including rising / falling slope, fluctuation amplitude, etc.), and simultaneously records the dynamic correlation data of impedance and temperature, continuously tracking the real-time evolution of the skin barrier state, providing a high temporal resolution monitoring basis for subsequent drug administration intensity adjustment.

[0023] Step S300: Extract the impedance value fluctuation features from the skin humidity change trend obtained from the real-time monitoring module, classify the impedance value fluctuation features using a classification algorithm, and determine the direction of drug administration intensity adjustment.

[0024] From the high-frequency (10Hz) "humidity-impedance" time series collected by the real-time monitoring module, impedance data for periods with a humidity change slope > 2%RH / min (during which the barrier state undergoes significant dynamic changes) are selected. Three core features of impedance fluctuations are extracted: amplitude, frequency, and trend (e.g., peak-to-trough difference, fluctuation period, and overall rise / fall slope). The feature vectors are then input into a trained random forest classifier, which is trained based on 1000+ sets of labeled data for "fluctuation features - optimal adjustment direction." This classifier can output three adjustment directions: when impedance fluctuations show skin barrier enhancement (e.g., increased impedance and narrowing fluctuation amplitude), it is determined to "enhance drug administration"; when the barrier weakens (e.g., decreased impedance and violent fluctuations), it is determined to "weaken drug administration"; when the barrier state is stable (fluctuations within a safe range), it is determined to "maintain the current strength." The adjustment direction needs to be cross-validated with the humidity change trend (e.g., when increased humidity is accompanied by decreased impedance, the confidence of "weaken drug administration" is strengthened) to ensure the accuracy of the decision.

[0025] Step S400: Based on the direction of drug administration intensity adjustment, a penetration depth prediction model is generated by integrating skin temperature data to obtain the predicted penetration depth value. The deviation between the predicted penetration depth value and the actual monitored penetration depth is calculated. If the deviation exceeds a preset threshold, the grouping parameters of the clustering algorithm are updated through a feedback loop to determine the optimized individual physiological identification classification result.

[0026] For the drug administration intensity adjustment direction (enhancement / weakening / maintenance) determined in step S300, the adjustment direction parameters (such as enhancement coefficient, weakening ratio) and real-time skin temperature data (32-37℃) are used as input. An LSTM neural network is used to construct a penetration depth prediction model. The penetration depth prediction model learns from historical "adjustment direction-temperature-depth" correlation data (1000+ samples) and outputs the predicted penetration depth value (unit: mm, error ≤0.1mm) for the next 5 minutes. Simultaneously, the actual penetration depth is collected using a confocal laser scanning microscope (accuracy 5μm), and the prediction deviation is calculated. If the deviation exceeds a preset threshold (e.g., 0.2 mm), a feedback loop is activated: the grouping parameters of the K-means clustering algorithm are corrected based on the deviation value (e.g., adjusting the cluster center and weight coefficients), the individual physiological parameters are reclassified, and an optimized individual physiological identification classification result is generated (e.g., corrected from "medium barrier type" to "medium barrier-temperature sensitive subtype"), ensuring that the matching degree between the classification result and the actual penetration characteristics is improved by ≥15%.

[0027] Step S500: Extract allergy-related features from the optimized individual physiological identification and classification results, and construct a dynamic regulation sequence by combining metabolic capacity data to obtain the drug administration intensity sequence value.

[0028] From the optimized individual physiological identification and classification results (e.g., "Type 4a - High Allergy Risk + Fast Metabolizer Subtype"), four-dimensional core features of allergic constitution (mean histamine concentration, allergy history score, erythema reaction threshold, and drug contact dermatitis risk level) were extracted. Z-score standardization was used to quantify these features into a risk index in the 0-1 range (0 for extremely low risk, 1 for extremely high risk). Simultaneously, three key indicators from metabolic capacity data—drug clearance rate, half-life, and hepatic and renal metabolic load—were retrieved to construct a two-dimensional correlation matrix of "allergy risk - metabolic rate." Based on this matrix, a modular approach was adopted. The predictive control (MPC) algorithm generates a time-gradient dosing intensity sequence value with the constraint that "the allergy risk index is ≤0.3 and the effective drug concentration is maintained within the therapeutic window". The dosing intensity sequence value is updated every 3 minutes and includes the current permeation power (e.g., ultrasound intensity 0.5-2.0 W / cm²), duration of action (10-60 seconds), and interval period (1-5 minutes). The change rate of adjacent intensities in the sequence is ≤20% (to avoid stimulation). The final output sequence value needs to be verified by allergy risk simulation (predicted allergy probability <5%).

[0029] Step S600: The actuator is driven by the drug delivery intensity sequence value to adjust the output parameters of the non-invasive drug delivery device. The above process is iterated by integrating real-time monitoring data to obtain a continuously optimized permeation rate control sequence.

[0030] The drug delivery intensity sequence value generated in step S500 (e.g., ultrasonic power 1.2 W / cm², duration 30 seconds) is converted into executable parameters (e.g., voltage, frequency, pulse width) for the non-invasive drug delivery device, driving the actuator (piezoelectric transducer, iontophoresis electrode, etc.) to output energy sequentially; real-time monitoring data (skin humidity, electrical impedance, temperature, actual penetration rate, etc.) are collected simultaneously during the drug delivery process, and the deviation between the current penetration rate and the target rate is calculated. Based on the deviation value, the dosing intensity sequence is corrected through a proportional-integral-derivative (PID) control algorithm, the actuator is re-driven and the monitoring-correction process is repeated, and one cycle iteration is completed every 30 minutes. After 3-5 iterations, when the permeation rate fluctuation is ≤±8% and stable for two consecutive cycles, a continuously optimized permeation rate control sequence is output. This permeation rate control sequence includes four elements: timestamp, target rate value, device output parameters and safety constraint threshold. It can directly guide the automated dosing of subsequent treatments and ensure the long-term stability and individual adaptability of drug permeation rate.

[0031] Furthermore, the skin-based non-invasive drug delivery control method based on big data analysis provided in this embodiment includes step S100 as follows: Step S110: Acquire raw values ​​of six physiological parameters—skin thickness, metabolic capacity, allergy status, electrical impedance, skin temperature, and skin humidity—through a physiological parameter sensor array at a preset data acquisition frequency. Perform noise filtering on the raw values ​​of the physiological parameters according to sensor accuracy standards to obtain a standardized physiological parameter dataset.

[0032] The standardized physiological parameter dataset is obtained using the following formula: (1) In formula (1), Indicates the first Each sensor at time Standardized physiological parameter data, Indicates the first Each sensor at time The collected raw physiological parameter values, Indicates the first The average benchmark of each sensor, Indicates the first The standard deviation of each sensor, Indicates based on sampling frequency and sensor accuracy Weighting factors.

[0033] The raw values ​​of six physiological parameters constitute the raw physiological parameter dataset, which is derived using the following formula: (2) In formula (2), This represents the raw physiological parameter dataset. arrive These correspond to six different types of physiological parameters. This indicates skin thickness data. Data representing metabolic capacity, This indicates data on allergic constitution. This represents the electrical impedance value data. This represents skin temperature data. This indicates skin moisture data.

[0034] The physiological parameter sensor array achieves comprehensive data acquisition through multi-point deployment. The skin thickness sensor uses ultrasonic measurement technology to continuously acquire data from areas such as the forearm and back of the hand at a frequency of 50Hz, obtaining thickness values ​​within the range of 0.8-2.5 mm. Metabolic capacity data is obtained through bioelectrical impedance analysis, reflecting cell membrane permeability and metabolic activity. Allergy predisposition data is detected through changes in skin conductivity, with values ​​typically ranging from 2-15 microSiemens.

[0035] Noise filtering employs an adaptive filtering algorithm. When abnormal fluctuations in skin temperature data exceeding ±0.5 degrees Celsius are detected, the system automatically removes outliers and performs interpolation compensation. This approach effectively eliminates data deviations caused by environmental interference and equipment vibration, ensuring the accuracy of subsequent analysis.

[0036] Step S120: Use a data preprocessor to perform correlation verification on the skin thickness data and electrical impedance data in the standardized physiological parameter dataset. If the skin thickness value in the skin thickness data exceeds the preset threshold range, adjust the weight coefficient of the electrical impedance value. Normalize the skin temperature data and skin humidity data according to individual differences to obtain the corrected parameter matrix.

[0037] The adjusted resistance value weighting coefficient is obtained by the following formula: (3) In formula (3), This represents the weighting coefficient for the adjusted electrical reactance value. This represents the weighting coefficient of the original electrical reactance value. This indicates the numerical value of skin thickness. and These represent the minimum and maximum values ​​within the preset threshold range, respectively. and This indicates the weight adjustment factor when the skin thickness exceeds the preset range.

[0038] Normalized skin temperature data are obtained using the following formula: (4) In formula (4), This represents the normalized skin temperature data. This represents the raw skin temperature data. Indicates based on individual differences The average skin temperature, Indicates based on individual differences The standard deviation of skin temperature.

[0039] Normalized skin moisture data is obtained using the following formula: (5) In formula (5), This represents the normalized skin moisture data. This represents the raw skin moisture data. Indicates based on individual differences The average skin moisture level, Indicates based on individual differences The standard deviation of skin moisture.

[0040] The correlation verification mechanism plays a crucial role. When an individual's skin thickness exceeds a preset threshold of 1.8 mm, the system adjusts the impedance value weighting coefficient from the standard 1.0 to 0.7, because thicker skin significantly affects the accuracy of impedance measurement. Simultaneously, the normalization process is dynamically adjusted based on individual differences such as age and gender; for example, the baseline skin moisture value for elderly individuals will be reduced by 15-20%.

[0041] Step S130: Perform multi-dimensional grouping calculations on the corrected parameter matrix using a clustering algorithm. If the numerical difference between the metabolic capacity index and the allergic constitution index is less than the classification threshold, then the individuals are grouped into the same cluster. Determine the optimal number of clusters based on the clustering accuracy standard to obtain the physiological characteristic clustering results.

[0042] The clustering rules are defined by the following formula: when the distance between an individual and the cluster center is less than the classification threshold, the individual is assigned to that cluster: (6) In formula (6), Indicates the first A cluster, Indicates the first The feature vector of each individual Indicates the first The center point of each cluster, Represents the distance function. This represents the classification threshold.

[0043] The following formula is used to evaluate the accuracy of clustering results, determining the optimal number of clusters by calculating the average similarity between individuals within all clusters and their cluster centers: (7) In formula (7), This represents the evaluation index for clustering accuracy. This represents the total number of clusters. Indicates the first A cluster, Indicates the first The number of individuals in each cluster Represents an individual within a cluster. Indicates the first The center of each cluster, This represents the similarity function.

[0044] The clustering algorithm employs a modified K-means method for multi-dimensional grouping. When the metabolic capacity index is 7.2 and the allergy constitution index is 7.8, the difference between the two (0.6) is less than the preset classification threshold of 1.0, and therefore they are grouped into the same cluster. The system determines the optimal number of clusters through silhouette coefficient evaluation, typically between 3 and 8 clusters, ensuring a clustering accuracy of over 85%.

[0045] Step S140: Generate an individual physiological identification mapping table based on the cluster center point coordinates and boundary range in the physiological feature clustering results. If the newly input physiological parameter data falls within the boundary range of the cluster, it is determined that the individual belongs to the corresponding physiological type classification, and the individual physiological identification classification result is obtained.

[0046] The following formula is used to determine the boundary range of each physiological type cluster, and to determine whether a new individual belongs to that physiological type: (8) In formula (8), Indicates the first The range of the boundary radius of each cluster. Indicates belonging to the first The set of all samples in each cluster Indicates the first in the cluster The feature vector of each sample Indicates the first The center point of each cluster.

[0047] The generated individual physiological identification mapping table includes the six-dimensional coordinate center point and boundary range of each cluster. The coordinates of the center point of the sensitive skin type cluster are: skin thickness 1.2 mm, metabolic capacity 6.5, allergy index 8.2, electrical impedance 45 ohms, temperature 32.1 degrees Celsius, and humidity 42%. The boundary range is set as a fluctuation range of ±15% for each dimension.

[0048] When a new individual's physiological parameters fall within the boundary of a specific cluster, the system calculates its Euclidean distance to the cluster center; the closer the distance, the higher the confidence level of the classification. This identification method enables rapid and accurate classification of individual physiological types, providing a scientific basis for the development of personalized skincare plans and significantly improving the suitability and effectiveness of skincare products.

[0049] Furthermore, in the skin non-invasive drug delivery control method based on big data analysis provided in this embodiment, step S200 includes: Step S210: The data preprocessor compares the individual physiological identification classification results with the preset penetration rate threshold and penetration depth threshold. If any parameter in the individual physiological identification classification results exceeds the corresponding threshold, a trigger signal is generated to activate the real-time monitoring module.

[0050] Individual physiological identification and classification results include measured permeation rate parameters and measured permeation depth parameters. The generation conditions for trigger signals are defined by the following formula: (9) In formula (9), Indicates a trigger signal. This represents the measured permeation rate parameter. This indicates the preset permeation rate threshold. This represents the measured penetration depth parameter. This indicates the preset penetration depth threshold.

[0051] The activation status of the real-time monitoring module is determined by the following formula: (10) In formula (10), Indicates the activation status of the real-time monitoring module. Represents the step function. Indicates a trigger signal. This indicates the startup parameters of the monitoring module. The monitoring module is activated when the trigger signal is positive, otherwise it remains in the off state.

[0052] The data preprocessor establishes a multi-level threshold comparison mechanism. The penetration rate threshold is differentiated based on different skin types: 0.8 mm / min for sensitive skin and 1.2 mm / min for normal skin. The penetration depth threshold is also stratified: 2.5 mm for shallow penetration and 4.8 mm for deep penetration. When an individual's physiological classification shows a skin thickness of 1.1 mm and an allergy index of 8.5, the system automatically categorizes the individual as highly sensitive and adjusts the monitoring standards for penetration parameters accordingly.

[0053] The trigger signal generation employs a multi-parameter joint judgment mechanism. When an individual's metabolic capacity index is detected to be 5.2, lower than the preset threshold of 6.0, and their electrical impedance value reaches 52 ohms, exceeding the upper limit of 50 ohms, the system immediately generates a high-priority trigger signal. At this time, the data preprocessor increases the monitoring frequency from the standard once every 30 seconds to once every 10 seconds to ensure timely capture of abnormal states. The trigger signal also includes a risk level indicator, divided into three levels: low risk, medium risk, and high risk, with different monitoring strategies and response mechanisms corresponding to different levels.

[0054] Step S220: Continuously sample skin humidity data through the real-time monitoring module, and use the sliding window algorithm to calculate the rate of change of skin humidity data to obtain the trend of skin humidity change.

[0055] The instantaneous rate of change of skin moisture data is calculated using the following formula: (11) In formula (11), Indicates time The rate of change in skin humidity, Indicates the current time Skin moisture level This indicates the skin's moisture level at the previous moment. Indicates the sampling time interval.

[0056] The overall trend of skin moisture change is obtained by averaging the rate of change over a certain time window using the following formula: (12) In formula (12), Indicates the first Skin moisture change trend values ​​over a period of time This indicates the length of the time window used for trend calculation. Indicates the first The rate of humidity change at each time point.

[0057] The real-time monitoring module employs a high-precision humidity sensor array for continuous sampling. Sensors are positioned on key areas of the face, including the T-zone, cheeks, and jawline, with a sampling frequency set every 5 seconds, achieving a humidity change detection accuracy of 0.1%. The module also features an environmental compensation algorithm that automatically corrects the baseline skin humidity value when ambient humidity changes, preventing interference from external factors. A sliding window algorithm calculates the rate of humidity change using a 15-minute time window. The environmental compensation algorithm continuously collects all humidity data points within the window and calculates the slope of the trend through linear regression analysis. When skin humidity drops from an initial 45% to 38% within 15 minutes, the system calculates a rate of change of 0.47% per minute, which is marked as a rapid dehydration state. The sliding window updates every 3 minutes to ensure the real-time nature and accuracy of trend analysis.

[0058] Preferably, the skin-based non-invasive drug delivery control method based on big data analysis provided in this embodiment includes step S300 as follows: Step S310: Based on the skin humidity change trend obtained by the real-time monitoring module, the impedance value fluctuation feature is extracted using a signal processor. If the frequency of the impedance value fluctuation feature is within a preset range, a first feature vector is generated.

[0059] The following formula is used to extract the characteristics of electrical impedance fluctuation: (13) In formula (13), It represents the power of signal fluctuations within a specific time window, serving as a quantitative indicator of fluctuation characteristics; This represents the real-time impedance value as it changes over time. It is the average value of the impedance within that time window. Represents the starting time of the integration interval, used to specify the starting point of the time range for analysis; This represents the length of the time window used for analysis.

[0060] The signal processor employs a multi-channel parallel acquisition architecture to extract impedance value fluctuation characteristics. The system is configured with eight independent impedance detection channels, each corresponding to changes in the skin's electrical properties in different facial regions. When skin humidity changes, the alteration in intercellular fluid content directly affects the tissue's conductivity, thus causing corresponding fluctuations in impedance values.

[0061] The signal processor continuously monitors impedance changes at a frequency of 200 times per second using high-frequency sampling technology and extracts effective fluctuation signals within the frequency range of 0.1Hz to 5Hz using a bandpass filtering algorithm. Frequency analysis of the fluctuation characteristics is performed using spectral decomposition technology for precise identification. When a sustained fluctuation in the impedance value is detected at 2.3Hz, and the fluctuation amplitude reaches 15% of the reference value, the system determines that the frequency characteristic is within a preset effective range. At this point, the signal processor automatically extracts key parameters such as the amplitude, phase, and duration of the fluctuation signal, constructing a first feature vector containing 12 dimensions of feature data. These dimensions cover multiple quantitative indicators such as peak intensity, rate of change, and frequency stability of the fluctuation.

[0062] Step S320: Classify the first feature vector using a feature classifier. If the magnitude of the first feature vector shows an increasing trend, it is labeled as the first category; if it shows a decreasing trend, it is labeled as the second category, thus obtaining the classification result.

[0063] The core discrimination rule of the classifier is embodied in the following formula, which is based on the trend metric. The sign of the value determines whether a data sample belongs to the first or second category, which directly corresponds to the core logic of classification based on increasing or decreasing trends: (14) In formula (14), This indicates the final classification result. Indicates the first category, Indicates the second category, This represents a trend measurement value.

[0064] The following formula is a key mathematical model for determining whether a trend is increasing or decreasing. If the trend measurement value... A positive value indicates an increasing trend, while a negative value indicates a decreasing trend. (15) In formula (15), This represents a trend measure calculated using a linear regression method. Indicates the length of the feature vector sequence. Indicates the first The time sequence number corresponding to each vector. Indicates the first The magnitude of each vector. and These represent the average values ​​of time and amplitude, respectively.

[0065] The feature classifier uses a pattern recognition algorithm to perform in-depth analysis of the first feature vector. The classifier first calculates the temporal trend of the amplitude parameter in the feature vector, determining whether it is an increasing or decreasing pattern by continuously monitoring the data change trajectory over 30 minutes. When the amplitude rises from an initial 0.8 units to 1.2 units within the monitoring period, the system identifies a clear increasing trend and labels it as the first category. Conversely, if the amplitude decreases from 1.1 units to 0.7 units, it is classified as the second category.

[0066] Step S330: Use a decision-making device to match the classification results with the preset drug administration intensity adjustment rules. If it is the first category, generate a reduction instruction; if it is the second category, generate an increase instruction to determine the direction of drug administration intensity adjustment.

[0067] The core logic of the decision-making system is described by the following formula: based on the classification result of the input... Belongs to the first category Or the second category Which preset category will be used to match and generate a unique, specific adjustment instruction? (16) In formula (16), This indicates the adjustment instruction generated after the decision-making process. This represents the input classification result. This represents the set of all results that fall into the first category. This represents the set of all results that fall into the second category. This indicates a reduction instruction. This indicates an elevation command.

[0068] The decision-making system establishes an intelligent dosing intensity adjustment rule base. This base contains optimal dosing strategies for different skin conditions. For the first category of increasing trends, the system generates a reduction instruction, adjusting the dosing intensity from the current 75% to 60% to avoid skin burden caused by overdosing. For the second category of decreasing trends, the system generates an increase instruction, raising the dosing intensity from 55% to 70% to ensure full drug penetration.

[0069] Furthermore, in the skin non-invasive drug delivery control method based on big data analysis provided in this embodiment, step S400 includes: Step S410: Adjust the direction of drug administration intensity and generate a penetration depth prediction model by integrating skin temperature data.

[0070] For a given direction of drug administration intensity adjustment (enhancement / weakening / maintenance), quantitative parameters of the adjustment direction (such as enhancement coefficient and weakening ratio) and real-time collected skin temperature data (including instantaneous temperature, temperature change rate, and temperature-baseline deviation) are used as input variables. A Long Short-Term Memory (LSTM) network is used to construct a penetration depth prediction model. The penetration depth prediction model learns the three-dimensional correlation between "drug administration adjustment - temperature change - depth evolution" in a historical database (covering 5000+ samples from 1000+ individuals) and outputs the predicted penetration depth value (in mm, with an accuracy of ±0.05 mm) every 30 seconds for the next 5 minutes. The penetration depth prediction model has a built-in temperature sensitivity correction module (based on the Arrhenius equation, the diffusion coefficient increases by about 5%-8% for every 1°C increase in temperature) and sets differentiated weight parameters for different skin types (such as strong barrier / weak barrier) to ensure that the predicted value can dynamically adapt to the individual skin barrier characteristics and temperature response differences.

[0071] The penetration depth prediction model constructs a prediction algorithm by fusing multi-dimensional skin temperature data. The system is equipped with an infrared temperature sensor array to collect real-time temperature distribution information on the skin surface and at different depths. When the skin temperature rises from a baseline of 32.5 degrees Celsius to 34.2 degrees Celsius, the penetration depth prediction model predicts the diffusion path of drug molecules in the skin tissue based on the temperature gradient change. The model uses thermal conduction theory combined with pharmacokinetic principles to calculate the impact of temperature changes on the skin barrier function, thereby estimating the maximum potential penetration depth of the drug under the current temperature conditions.

[0072] Step S420: Process individual skin using a penetration depth prediction model to generate predicted penetration depth values.

[0073] The predicted penetration depth value is generated using the following formula: (17) In formula (17), This represents the predicted drug penetration depth value. This represents a baseline function for penetration depth determined solely by the intensity of drug administration. This represents the penetration depth influence function determined solely by skin temperature. The fusion effect function representing the interaction between drug administration intensity and skin temperature. The fusion weighting coefficient is used to adjust the contribution of the fusion term to the final prediction result.

[0074] The penetration depth prediction model establishes a correlation between temperature and penetration depth. The built-in algorithm analyzes the thermal resistance characteristics of different skin layers. When the epidermal temperature is detected at 33.8 degrees Celsius and the dermal temperature at 33.1 degrees Celsius, the system predicts a drug penetration depth of approximately 1.2 millimeters. Simultaneously, the model considers individual differences, including physiological parameters such as skin thickness and pore density, to personalize the prediction results.

[0075] Step S430: Compare the predicted penetration depth value with the actual penetration depth value obtained by the real-time monitoring module using the deviation calculator. If the deviation between the predicted penetration depth value and the obtained actual penetration depth value exceeds a preset threshold, a deviation correction signal is generated.

[0076] The following formula is used to quantify the overall degree of deviation between forecasts and actual values: (18) In formula (18), This represents the weighted average deviation value calculated within a specific time window. This indicates the number of sample points used in the calculation. Indicates the first The weighting coefficients for each sample point Indicates the first Predicted penetration depth values ​​for each sample point Indicates the first The actual penetration depth value obtained from each sample point.

[0077] The deviation calculator uses a real-time comparison mechanism to verify prediction accuracy. The system uses ultrasonic detection technology to obtain the actual distribution depth of the drug in the skin. When the measured penetration depth is 0.9 mm and the predicted value is 1.2 mm, the deviation calculator will identify a difference of 0.3 mm. If this deviation exceeds the preset threshold range of 0.2 mm, the system will immediately generate a deviation correction signal, triggering an adaptive adjustment mechanism for the model parameters.

[0078] Step S440: Based on the deviation correction signal, update the grouping parameters of the clustering algorithm using a feedback loop to obtain optimized individual physiological identification classification results.

[0079] The deviation correction signal is defined using the following formula: (19) In formula (19), Indicates the first The bias correction signal vector of each cluster is used to quantify the average degree of deviation of data points within that cluster; Indicates the current division to the th The set of all physiological data points in each cluster, Represents a set The number of data points in the middle, This represents a vector of physiological data of an individual in the set. This represents the current center position of the k-th cluster, serving as a key grouping parameter.

[0080] Upon receiving the bias correction signal, the feedback loop automatically updates the key grouping parameters in the clustering algorithm. The system reassesses the skin characteristics of the current individual, reclassifying skin originally categorized as standard permeability type into low permeability type. The clustering algorithm adjusts the temperature sensitivity coefficient from 1.5 to 1.2, while simultaneously correcting the permeability resistance parameter from 0.8 to 1.1, ensuring that subsequent predictions are closer to reality.

[0081] The optimized individual physiological classification results lead to a more precise drug delivery strategy. Based on the revised classification information, the system will re-formulate a personalized drug delivery intensity adjustment plan. For newly identified low-permeability skin types, the system will increase the drug delivery intensity to 85% and extend the delivery time to 45 minutes, while adjusting the temperature control range to 34-36 degrees Celsius to promote full drug penetration. This closed-loop feedback mechanism continuously optimizes the accuracy of the predictive model, achieving precise control of the drug delivery process.

[0082] Furthermore, in the skin non-invasive drug delivery control method based on big data analysis provided in this embodiment, step S500 includes: Step S510: Receive the optimized individual physiological identification and classification results, and use a feature extractor to filter out physiological marker parameters related to allergic constitution from the individual physiological identification and classification results. If the individual allergic constitution score exceeds the preset threshold, obtain the corresponding allergic reaction intensity index.

[0083] The conditions for initiating the acquisition of indicators of allergic reaction intensity are defined by the following formula: (20) In formula (20), Indicators representing the intensity of allergic reactions This indicates a set of relevant physiological data. The function to obtain the intensity index in the middle. It is an indicator function that has a value of 1 when the condition inside the parentheses is true, and 0 otherwise. This represents an individual's allergy predisposition score. This indicates the preset scoring threshold. The system will only activate and acquire intensity metrics when the score exceeds the threshold.

[0084] The feature extractor identifies individual allergic constitution characteristics through multi-dimensional physiological parameter analysis. The system is equipped with a biomarker detection module to monitor serum immunoglobulin E levels, histamine release, and skin sensitivity indicators in real time. When an individual's serum IgE concentration reaches 150 IU / ml, and the skin sensitivity test shows an erythema reaction area exceeding 8 square millimeters, the feature extractor calculates a comprehensive allergic constitution score of 7.2. If this score exceeds a preset threshold of 6.0, the system automatically obtains the corresponding allergic reaction intensity indicators, including the rate of inflammatory factor release and the degree of change in vascular permeability.

[0085] Step S520: Obtain individual drug metabolism rate data through a metabolic capacity monitoring device, and construct a dynamic regulation sequence based on the allergic reaction intensity index combined with the drug metabolism rate data.

[0086] The following formula is used to construct the entire dynamic regulation sequence: (twenty one) In formula (21), This represents the optimal dynamic control sequence obtained through optimization calculation. It is any candidate regulatory sequence, consisting of a series of regulatory instructions. composition; It is the total cycle of the regulation process. It is a function that represents the control command. and metabolic rate The intensity of the allergic reaction produced by the combined effects; It is a function representing the therapeutic effect of the drug. The desired therapeutic effect is... and These are weighting factors, used to balance the goals of minimizing allergic reactions and maximizing treatment efficacy.

[0087] The metabolic monitoring device uses enzyme activity detection technology to acquire individual drug metabolism characteristic data. The system detects the activity of the liver cytochrome P450 enzyme system; when the CYP3A4 enzyme activity is 0.7 times the standard value, it indicates that the individual's drug metabolism rate is relatively slow. Combined with the histamine release rate of 2.3 ng / ml / min, an indicator of allergic reaction intensity, the system constructs a dynamic regulatory sequence for that individual, setting the drug concentration increment to 5% at 90-minute intervals.

[0088] Step S530: The dosing interval for individuals is calculated using a dynamic regulation sequence, and the dosing intensity trend data is determined based on the drug metabolism rate data.

[0089] The following formula adaptively adjusts the dosing interval using a nonlinear function based on the cumulative deviation between the dynamic regulation sequence and the target value: (twenty two) In formula (22), Indicates the time interval for the next dose. This represents a reference dosing interval baseline value. This represents a sensitivity adjustment coefficient. Indicates the weight of historical measurement data, Indicates the first in the dynamic regulation sequence One measurement value, Indicates the target value of the regulatory sequence. This indicates the total number of measured values.

[0090] The construction of the dynamic regulatory sequence fully considers the balance between allergy risk and metabolic capacity. When an individual's allergy intensity index shows moderate sensitivity and the drug metabolism rate is 0.8 ml / min, the system generates a regulatory sequence containing 6 time points. The regulatory sequence sets the initial dosing intensity to 40%, increasing by 8% every 75 minutes to ensure that the drug concentration gradually increases within a safe range, avoiding allergic reactions while ensuring therapeutic efficacy.

[0091] The determination of dosing intensity trend data is based on real-time changes in drug metabolism rate. The system continuously monitors fluctuations in individual metabolic enzyme activity. When the metabolic rate is detected to increase from the initial 0.8 ml / min to 1.1 ml / min, the dosing intensity trend data will be adjusted accordingly to an incremental mode, suggesting that the subsequent dosing intensity be increased from the current 56% to 65% to compensate for the decrease in drug concentration caused by accelerated metabolism.

[0092] Step S540: Receive drug administration intensity trend data through a sequence optimization processor, perform temporal arrangement and intensity calibration on the dynamic regulation sequence, and obtain the drug administration intensity sequence value.

[0093] The core idea of ​​arranging dynamic regulation sequences in time is represented by the following formula: (twenty three) In formula (23), This represents the optimal dosing sequence arrangement. Represents the set of all possible temporal arrangements. This represents a utility evaluation function used to evaluate the benefits of using a specific intensity under specific conditions; Indicates according to the permutation scheme At the point of time The selected dosing intensity trend data, This represents the physiological state of a group of patients at time point t. Indicates the total duration.

[0094] After receiving dosing intensity trend data, the sequence optimization processor refines the temporal arrangement of the dynamic regulation sequence. Analyzing individual physiological changes across different time periods, the processor readjusts the originally uniformly distributed dosing times to a non-equidistant pattern. During periods of high metabolic activity, the dosing interval is shortened to 60 minutes, while during periods of low metabolic activity, the interval is extended to 100 minutes. Simultaneously, intensity calibration is performed, limiting peak dosing intensity to below 72%. This ultimately generates dosing intensity sequence values ​​containing intensity values ​​of 42%, 51%, 58%, 65%, 70%, and 72%, enabling personalized and precise dosing control.

[0095] Furthermore, in the skin non-invasive drug delivery control method based on big data analysis provided in this embodiment, step S600 includes: Step S610: Based on the drug delivery intensity sequence value, the output frequency and output intensity of the non-invasive drug delivery device are dynamically adjusted by the actuator control module to determine the real-time adjusted permeation rate parameter.

[0096] The following formula describes the relationship between the actuator control module dynamically adjusting the output frequency according to the drug administration intensity sequence: (twenty four) In formula (24), Indicates the actuator at time... The output frequency, Indicates the fundamental frequency parameter. Indicates the frequency adjustment coefficient. Indicates time The corresponding dosing intensity sequence value.

[0097] The following formula is used to calculate the actual output intensity under actuator control: (25) In formula (25), Indicates the output intensity of the non-invasive drug delivery device. This indicates the maximum output power of the device. Indicates the control signal strength value. Indicates the maximum control signal value. This represents the equipment efficiency coefficient.

[0098] The final permeation rate parameters after dynamic adjustment of frequency and intensity were determined using the following formula: (26) In formula (26), This represents the permeation rate parameter after real-time adjustment. Indicates the baseline permeation rate, This indicates the adjusted frequency value. Indicates the reference frequency value. This indicates the adjusted strength value. Indicates the reference strength value. and These represent the indices representing the influence of frequency and intensity on the permeation rate, respectively.

[0099] The actuator control module controls drug penetration by precisely adjusting the electromagnetic pulse frequency of the non-invasive drug delivery device. When the system receives a drug delivery intensity sequence value of 42%, the actuator control module sets the device output frequency to 15Hz, corresponding to an electric field strength of 3.2V / cm. As the drug delivery intensity increases to 51%, the system automatically increases the output frequency to 18Hz and simultaneously strengthens the electric field strength to 3.8V / cm, ensuring that the drug molecule penetration rate in the skin barrier remains consistent with the preset intensity. Real-time adjustment of the penetration rate parameter is based on the physical characteristics of iontophoresis technology. The system assesses drug penetration by monitoring changes in skin impedance. When the skin impedance decreases from the initial 2.1kΩ to 1.8kΩ, it indicates increased skin permeability. At this point, the actuator control module correspondingly reduces the output intensity to 2.9V / cm to prevent the drug penetration rate from exceeding the safe range. This dynamic adjustment mechanism ensures that the drug penetrates the skin barrier at a stable rate of 0.15mg / cm² / h.

[0100] Step S620: If the real-time adjusted permeation rate parameter is inconsistent with the preset physiological safety threshold, then the real-time monitoring data is analyzed by the feedback regulator to obtain the deviation correction sequence.

[0101] The bias correction sequence is derived using the following formula: (27) In formula (27), Indicates the first in the deviation correction sequence One correction value, Indicates the correction amplitude coefficient. Indicates the attenuation factor. Indicates the correction frequency. Indicates phase shift, Represents a sequence index. Represents the natural constant.

[0102] The physiological safety threshold is set by comprehensively considering individual skin tolerance and drug concentration limitations. The system's preset upper limit for penetration rate is 0.20 mg / cm² / h. When real-time monitoring shows that the penetration rate reaches 0.23 mg / cm² / h, the feedback regulator immediately initiates the deviation analysis program. If the analysis results show that the penetration rate exceeds the safety threshold by 15%, the system will generate a deviation correction sequence containing a frequency reduction instruction, adjusting the output frequency from the current 20 Hz to 16 Hz.

[0103] Step S630: Using a deviation correction sequence, the drug administration intensity sequence value is iteratively optimized through a time-series optimization processor to obtain a continuously optimized permeation rate control sequence.

[0104] The iteratively obtained dosing intensity sequence value is obtained using the following formula: (28) In formula (28), Indicates the first The dosing intensity sequence value after the next iteration Indicates the first The dosing intensity sequence value of the next iteration This represents the learning rate parameter of the timing optimization processor. The gradient of the objective function representing the drug administration intensity is given. This represents the bias correction weighting coefficient. Indicates the first The increment of the deviation correction sequence in the next iteration.

[0105] The continuously optimized permeation rate control sequence is derived using the following formula: (29) In formula (29), Indicates at time The permeation rate control sequence value, This indicates the total number of control sequence segments. Indicates the first Segment weighting factor Indicates the first Each control time point Indicates the first The time-diffusion parameter of the segment, Indicates the first Concentration control parameters for the segment, Indicates the current time.

[0106] The bias correction sequence was constructed using a multi-parameter fusion analysis method. When an individual's skin temperature was detected to rise from 36.5℃ to 37.8℃, while skin humidity increased by 12%, the feedback regulator calculated the influence coefficients of temperature and humidity on the penetration rate. The system-generated correction sequence included a temperature compensation factor of 0.85 and a humidity adjustment factor of 1.12, which were used to correct subsequent dosing parameter settings.

[0107] After receiving the deviation correction sequence, the timing optimization processor iteratively adjusts the original dosing intensity sequence. Analyzing the periodic changes in skin physiological states, the processor identifies peaks in skin permeability at 45 and 90 minutes post-drug administration. Based on this finding, the system adjusts the original constant-intensity increasing pattern to a fluctuating increasing pattern, reducing the dosing intensity to 48% during peak permeability periods and increasing it to 68% during trough permeability periods, ultimately achieving stable drug concentration maintenance and continuously optimized permeation rate control.

[0108] Please see Figure 2This embodiment provides a skin-based non-invasive drug delivery control system based on big data analysis, used to execute the aforementioned skin-based non-invasive drug delivery control method based on big data analysis. It includes an individual physiological identification and classification result acquisition module 10, a skin humidity change trend acquisition module 20, a drug delivery intensity adjustment direction judgment module 30, an individual physiological identification and classification result optimization module 40, a drug delivery intensity sequence value acquisition module 50, and a penetration rate control acquisition module 60. The individual physiological identification and classification result acquisition module 10 collects individual physiological parameter data through a physiological parameter sensor array, and uses a clustering algorithm to group the physiological parameter data to obtain individual physiological identification and classification results. The physiological parameter data includes skin thickness data, metabolic capacity data, allergy constitution data, electrical impedance data, skin temperature data, and skin humidity data. The skin humidity change trend acquisition module 20 compares the individual physiological identification and classification results with preset penetration rate thresholds and penetration depth thresholds. If the individual physiological identification and classification results show that the physiological parameters exceed the preset penetration rate thresholds and penetration depth thresholds, real-time monitoring is activated. The module continuously tracks the trend of skin humidity changes; the drug administration intensity adjustment direction judgment module 30 is used to extract the impedance value fluctuation characteristics from the skin humidity change trend obtained by the real-time monitoring module, classify the impedance value fluctuation characteristics using a classification algorithm, and determine the direction of drug administration intensity adjustment; the individual physiological identification classification result optimization module 40 is used to generate a penetration depth prediction model by integrating skin temperature data for the drug administration intensity adjustment direction, obtain the predicted penetration depth value, calculate the deviation between the predicted penetration depth value and the actual monitored penetration depth, and if the deviation exceeds a preset threshold, update the grouping parameters of the clustering algorithm through a feedback loop to determine the optimized individual physiological identification classification result; the drug administration intensity sequence value acquisition module 50 is used to extract allergy-related features from the optimized individual physiological identification classification result, construct a dynamic regulation sequence by combining metabolic capacity data, and obtain the drug administration intensity sequence value; the penetration rate control acquisition module 60 is used to drive the actuator to adjust the output parameters of the non-invasive drug delivery device through the drug administration intensity sequence value, integrate real-time monitoring data and iterate the above process to obtain a continuously optimized penetration rate control sequence.

[0109] This embodiment discloses a non-invasive skin drug delivery control method and system based on big data analysis. Compared with existing technologies, it collects multi-dimensional data such as skin thickness, metabolic capacity, allergic constitution, electrical impedance value, skin temperature, and humidity through a physiological parameter sensor array, and uses a clustering algorithm to classify and identify individual physiological characteristics. When individual parameters exceed a preset penetration threshold, a real-time monitoring module is automatically activated. By analyzing the fluctuation characteristics of electrical impedance value, the direction of drug delivery intensity adjustment is determined. A penetration depth prediction model is constructed by integrating skin temperature data to achieve feedback optimization of the deviation between predicted and actual monitoring values. This embodiment further extracts allergic constitution characteristics and combines them with metabolic capacity data to construct a dynamic regulation sequence, generating personalized drug delivery intensity parameters to drive the actuator to adjust the device output. Through iterative optimization, continuous and precise penetration rate control is achieved, effectively solving the contradiction between individual differentiated drug delivery needs and standardized device output, and significantly improving the safety and effectiveness of non-invasive drug delivery.

[0110] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A non-invasive skin drug delivery control method based on big data analysis, characterized in that, Includes the following steps: S100. Collect individual physiological parameter data through a physiological parameter sensor array, and use a clustering algorithm to group the physiological parameter data to obtain individual physiological identification and classification results. The physiological parameter data includes skin thickness data, metabolic capacity data, allergy constitution data, electrical impedance data, skin temperature data, and skin humidity data. S200. The individual physiological identification and classification results are compared with preset penetration rate thresholds and penetration depth thresholds. If the individual physiological identification and classification results show that the physiological parameters exceed the preset penetration rate thresholds and penetration depth thresholds, the real-time monitoring module is activated to continuously track the skin humidity change trend. S300: Extract the impedance value fluctuation features from the skin humidity change trend obtained by the real-time monitoring module, classify the impedance value fluctuation features using a classification algorithm, and determine the direction of drug administration intensity adjustment; S400. Based on the direction of drug administration intensity adjustment, a penetration depth prediction model is generated by integrating skin temperature data to obtain the predicted penetration depth value. The deviation between the predicted penetration depth value and the actual monitored penetration depth is calculated. If the deviation exceeds a preset threshold, the grouping parameters of the clustering algorithm are updated through a feedback loop to determine the optimized individual physiological identification classification result. S500: Extract allergy-related features from the optimized individual physiological identification and classification results, and construct a dynamic regulation sequence by combining metabolic capacity data to obtain the drug administration intensity sequence value; S600: The actuator is driven by the drug delivery intensity sequence value to adjust the output parameters of the non-invasive drug delivery device. The above process is iterated by integrating real-time monitoring data to obtain a continuously optimized permeation rate control sequence.

2. The skin-based non-invasive drug delivery control method based on big data analysis according to claim 1, characterized in that, Step S100 includes: S110. Acquire the raw values ​​of six physiological parameters—skin thickness, metabolic capacity, allergy constitution, electrical impedance, skin temperature, and skin humidity—through a physiological parameter sensor array at a preset data acquisition frequency. Perform noise filtering on the raw values ​​of the physiological parameters according to the sensor accuracy standard to obtain a standardized physiological parameter dataset. The standardized physiological parameter dataset is obtained using the following formula: in, Indicates the first Each sensor at time Standardized physiological parameter data, Indicates the first Each sensor at time The collected raw physiological parameter values, Indicates the first The average benchmark of each sensor, Indicates the first The standard deviation of each sensor, Indicates based on sampling frequency and sensor accuracy Weighting factors; The raw values ​​of six physiological parameters constitute the raw physiological parameter dataset, which is derived using the following formula: in, This represents the raw physiological parameter dataset. arrive These correspond to six different types of physiological parameters. This indicates skin thickness data. Data representing metabolic capacity, This indicates data on allergic constitution. This represents the electrical impedance value data. This represents skin temperature data. This indicates skin moisture data; S120. A data preprocessor is used to perform correlation verification on the skin thickness data and electrical impedance data in the standardized physiological parameter dataset. If the skin thickness value in the skin thickness data exceeds the preset threshold range, the electrical impedance weight coefficient is adjusted. The skin temperature data and skin humidity data are normalized according to individual difference characteristics to obtain the corrected parameter matrix. The adjusted resistance value weighting coefficient is obtained by the following formula: in, This represents the weighting coefficient for the adjusted electrical reactance value. This represents the weighting coefficient of the original electrical reactance value. This indicates the numerical value of skin thickness. and These represent the minimum and maximum values ​​within the preset threshold range, respectively. and This indicates the weight adjustment factor when the skin thickness exceeds the preset range; Normalized skin temperature data are obtained using the following formula: in, This represents the normalized skin temperature data. This represents the raw skin temperature data. Indicates based on individual differences The average skin temperature, Indicates based on individual differences The standard deviation of skin temperature; Normalized skin moisture data is obtained using the following formula: in, This represents the normalized skin moisture data. This represents the raw skin moisture data. Indicates based on individual differences The average skin moisture level, Indicates based on individual differences The standard deviation of skin moisture; S130. The corrected parameter matrix is ​​calculated in multiple dimensions by using a clustering algorithm. If the difference between the metabolic capacity index and the allergic constitution index is less than the classification threshold, the individuals are classified into the same cluster. The optimal number of clusters is determined according to the clustering accuracy standard to obtain the physiological characteristic clustering results. The clustering rules are defined by the following formula: when the distance between an individual and the cluster center is less than the classification threshold, the individual is assigned to that cluster: in, Indicates the first A cluster, Indicates the first The feature vector of each individual Indicates the first The center point of each cluster, Represents the distance function. Indicates the classification threshold; The following formula is used to evaluate the accuracy of clustering results, determining the optimal number of clusters by calculating the average similarity between individuals within all clusters and their cluster centers: in, This represents the clustering accuracy evaluation metric. This represents the total number of clusters. Indicates the first A cluster, Indicates the first The number of individuals in each cluster Represents an individual within a cluster. Indicates the first The center of each cluster, Represents the similarity function; S140. Generate an individual physiological identification mapping table based on the cluster center point coordinates and boundary range in the physiological feature clustering results. If the newly input physiological parameter data falls within the boundary range of the cluster, it is determined that the individual belongs to the corresponding physiological type classification, and the individual physiological identification classification result is obtained. The following formula is used to determine the boundary range of each physiological type cluster, and to determine whether a new individual belongs to that physiological type: in, Indicates the first The range of the boundary radius of each cluster. Indicates belonging to the first The set of all samples in each cluster Indicates the first in the cluster The feature vector of each sample Indicates the first The center point of each cluster.

3. The non-invasive skin drug delivery control method based on big data analysis according to claim 1, characterized in that, Step S200 includes: S210. The data preprocessor compares the individual physiological identification and classification results with the preset penetration rate threshold and penetration depth threshold. If any parameter in the individual physiological identification and classification results exceeds the corresponding threshold, a trigger signal is generated to activate the real-time monitoring module. Individual physiological identification and classification results include measured permeation rate parameters and measured permeation depth parameters. The generation conditions for trigger signals are defined by the following formula: in, Indicates a trigger signal. This represents the measured permeation rate parameter. This indicates the preset permeation rate threshold. This represents the measured penetration depth parameter. Indicates the preset penetration depth threshold; The activation status of the real-time monitoring module is determined by the following formula: in, Indicates the activation status of the real-time monitoring module. Represents the step function. Indicates a trigger signal. This indicates the startup parameters of the monitoring module. The monitoring module is activated when the trigger signal is positive, and otherwise remains in the off state. S220. The real-time monitoring module continuously samples the skin humidity data, and the sliding window algorithm is used to calculate the rate of change of the skin humidity data to obtain the skin humidity change trend. The instantaneous rate of change of skin moisture data is calculated using the following formula: in, Indicates time The rate of change in skin humidity, Indicates the current time Skin moisture level This indicates the skin's moisture level at the previous moment. Indicates the sampling time interval; The overall trend of skin moisture change is obtained by averaging the rate of change over a certain time window using the following formula: in, Indicates the first Skin moisture change trend values ​​over a period of time This indicates the length of the time window used for trend calculation. Indicates the first The rate of humidity change at each time point.

4. The non-invasive skin drug delivery control method based on big data analysis according to claim 1, characterized in that, Step S300 includes: S310. Based on the skin humidity change trend obtained by the real-time monitoring module, the impedance value fluctuation feature is extracted using a signal processor. If the frequency of the impedance value fluctuation feature is within a preset range, a first feature vector is generated. The following formula is used to extract the characteristics of electrical impedance fluctuation: in, It represents the power of signal fluctuations within a specific time window, serving as a quantitative indicator of fluctuation characteristics; This represents the real-time impedance value as it changes over time. It is the average value of the impedance within that time window. Represents the starting time of the integration interval, used to specify the starting point of the time range for analysis; This represents the length of the time window used for analysis; S320. The first feature vector is classified using a feature classifier. If the magnitude of the first feature vector shows an increasing trend, it is labeled as the first category; if it shows a decreasing trend, it is labeled as the second category, thus obtaining the classification result. The core discrimination rule of the classifier is embodied in the following formula, which is based on the trend metric. The sign of the value determines whether a data sample belongs to the first or second category, which directly corresponds to the core logic of classification based on increasing or decreasing trends: in, This indicates the final classification result. Indicates the first category, Indicates the second category, Indicates a trend measurement value; The following formula is a key mathematical model for determining whether a trend is increasing or decreasing. If the trend measurement value... A positive value indicates an increasing trend, while a negative value indicates a decreasing trend. in, This represents a trend measure calculated using a linear regression method. Indicates the length of the feature vector sequence. Indicates the first The time sequence number corresponding to each vector. Indicates the first The magnitude of each vector. and These represent the average values ​​of time and amplitude, respectively. S330. The decision-making unit matches the classification result with the preset drug administration intensity adjustment rules. If it is the first category, a reduction instruction is generated; if it is the second category, an increase instruction is generated to determine the direction of drug administration intensity adjustment. The core logic of the decision-making system is described by the following formula: based on the classification result of the input... Belongs to the first category Or the second category Which preset category will be used to match and generate a unique, specific adjustment instruction? in, This indicates the adjustment instruction generated after the decision-making process. This represents the input classification result. This represents the set of all results that fall into the first category. This represents the set of all results that fall into the second category. This indicates a reduction instruction. This indicates an elevation command.

5. The non-invasive skin drug delivery control method based on big data analysis according to claim 1, characterized in that, Step S400 includes: S410. Based on the direction of drug delivery intensity adjustment, skin temperature data is integrated to generate a penetration depth prediction model; S420. The penetration depth prediction model is used to process individual skin to generate a predicted penetration depth value. The predicted penetration depth value is generated using the following formula: in, This represents the predicted drug penetration depth value. This represents a baseline function for penetration depth determined solely by the intensity of drug administration. This represents the penetration depth influence function determined solely by skin temperature. The fusion effect function representing the interaction between drug administration intensity and skin temperature. The fusion weighting coefficient is used to adjust the contribution of the fusion term to the final prediction result; S430. The predicted penetration depth value is compared with the actual penetration depth value obtained by the real-time monitoring module using the deviation calculator. If the deviation between the predicted penetration depth value and the obtained actual penetration depth value exceeds a preset threshold, a deviation correction signal is generated. The following formula is used to quantify the overall degree of deviation between forecasts and actual values: in, This represents the weighted average deviation value calculated within a specific time window. This indicates the number of sample points used in the calculation. Indicates the first The weighting coefficients of each sample point Indicates the first Predicted penetration depth values ​​for each sample point Indicates the first The actual penetration depth value obtained from each sample point; S440. Based on the deviation correction signal, the grouping parameters of the clustering algorithm are updated using a feedback loop to obtain optimized individual physiological identification and classification results; The deviation correction signal is defined using the following formula: in, Indicates the first The bias correction signal vector of each cluster is used to quantify the average degree of deviation of data points within that cluster; Indicates the current division to the th The set of all physiological data points in each cluster, Represents a set The number of data points in the middle, This represents a vector of physiological data of an individual in the set. This represents the current center position of the k-th cluster, serving as a key grouping parameter.

6. The non-invasive skin drug delivery control method based on big data analysis according to claim 1, characterized in that, Step S500 includes: S510. Receive the optimized individual physiological identification and classification results, and use a feature extractor to filter out physiological marker parameters related to allergic constitution from the individual physiological identification and classification results. If the individual allergic constitution score exceeds a preset threshold, obtain the corresponding allergic reaction intensity index. The conditions for initiating the acquisition of indicators of allergic reaction intensity are defined by the following formula: in, Indicators representing the intensity of allergic reactions This indicates a set of relevant physiological data. The function to obtain the intensity index in the middle. It is an indicator function that has a value of 1 when the condition inside the parentheses is true, and 0 otherwise. This represents an individual's allergy predisposition score. This indicates a preset scoring threshold; the system will only activate and acquire intensity metrics when the score exceeds the threshold. S520. Obtain individual drug metabolism rate data through a metabolic capacity monitoring device, and construct a dynamic regulation sequence based on the allergic reaction intensity index and the drug metabolism rate data. The following formula is used to construct the entire dynamic regulation sequence: in, This represents the optimal dynamic control sequence obtained through optimization calculation. It is any candidate regulatory sequence, consisting of a series of regulatory instructions. composition; It is the total cycle of the regulation process. It is a function that represents the control command. and metabolic rate The intensity of the allergic reaction produced by the combined effects; It is a function representing the therapeutic effect of the drug. The desired therapeutic effect is... and These are weighting factors, used to balance the two objectives of minimizing allergic reactions and maximizing treatment efficacy; S530. The dynamic regulation sequence is used to calculate the dosing time interval for an individual, and the dosing intensity trend data is determined based on the drug metabolism rate data. S540. The drug administration intensity trend data is received by the sequence optimization processor, and the dynamic regulation sequence is time-series arranged and intensity-calibrated to obtain the drug administration intensity sequence value.

7. The non-invasive skin drug delivery control method based on big data analysis according to claim 6, characterized in that, In step S530, the following formula adaptively adjusts the dosing interval using a nonlinear function based on the cumulative deviation between the dynamic regulation sequence and the target value: in, Indicates the time interval for the next dose. This represents a reference dosing interval baseline value. This represents a sensitivity adjustment coefficient. Indicates the weight of historical measurement data, Indicates the first in the dynamic regulation sequence One measurement value, Indicates the target value of the regulatory sequence. This indicates the total number of measured values.

8. The non-invasive skin drug delivery control method based on big data analysis according to claim 7, characterized in that, In step S540, the core idea of ​​arranging the dynamic regulation sequence in time is represented by the following formula: in, This represents the optimal dosing sequence arrangement. Represents the set of all possible temporal arrangements. This represents a utility evaluation function used to evaluate the benefits of using a specific intensity under specific conditions; Indicates according to the permutation scheme At the point of time The selected dosing intensity trend data, This represents the physiological state of a group of patients at time point t. Indicates the total duration.

9. The non-invasive skin drug delivery control method based on big data analysis according to claim 1, characterized in that, Step S600 includes: S610. Based on the drug delivery intensity sequence value, the output frequency and output intensity of the non-invasive drug delivery device are dynamically adjusted by the actuator control module to determine the real-time adjusted penetration rate parameter. S620. If the real-time adjusted permeation rate parameter is inconsistent with the preset physiological safety threshold, the real-time monitoring data will be analyzed for deviation through the feedback regulator to obtain the deviation correction sequence. S630. Using the aforementioned deviation correction sequence, the drug administration intensity sequence value is iteratively optimized through a timing optimization processor to obtain a continuously optimized penetration rate control sequence.

10. A skin-based non-invasive drug delivery control system based on big data analysis, used to execute the skin-based non-invasive drug delivery control method based on big data analysis as described in any one of claims 1 to 9, characterized in that, include: The individual physiological identification and classification result acquisition module (10) is used to collect individual physiological parameter data through a physiological parameter sensor array, and to group the physiological parameter data using a clustering algorithm to obtain individual physiological identification and classification results. The physiological parameter data includes skin thickness data, metabolic capacity data, allergic constitution data, electrical impedance data, skin temperature data, and skin humidity data. The skin humidity change trend acquisition module (20) is used to compare the individual physiological identification classification result with the preset penetration rate threshold and penetration depth threshold. If the individual physiological identification classification result shows that the physiological parameters exceed the preset penetration rate threshold and penetration depth threshold, the real-time monitoring module is activated to continuously track the skin humidity change trend. The drug administration intensity adjustment direction judgment module (30) is used to extract the impedance value fluctuation features from the skin humidity change trend obtained by the real-time monitoring module, classify the impedance value fluctuation features using a classification algorithm, and judge the drug administration intensity adjustment direction. The individual physiological identification classification result optimization module (40) is used to generate a penetration depth prediction model by integrating skin temperature data for the direction of drug administration intensity adjustment, obtain the predicted penetration depth value, calculate the deviation between the predicted penetration depth value and the actual monitored penetration depth, and if the deviation exceeds the preset threshold, update the grouping parameters of the clustering algorithm through the feedback loop to determine the optimized individual physiological identification classification result. The drug administration intensity sequence value acquisition module (50) is used to extract allergy-related features from the optimized individual physiological identification and classification results, and construct a dynamic regulation sequence by combining metabolic capacity data to obtain the drug administration intensity sequence value; The permeation rate control acquisition module (60) is used to drive the actuator to adjust the output parameters of the non-invasive drug delivery device through the drug delivery intensity sequence value, and to iterate the above process by integrating real-time monitoring data to obtain a continuously optimized permeation rate control sequence.

Images