Microfluidic biochip data processing method and system based on drug release control
Patent Information
- Application Number
- CN202410722484.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-05
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-06-05
Smart Images

Figure CN118295236B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of drug release control, and in particular to a microfluidic biochip data processing method and system based on drug release control. Background Art
[0002] In recent years, microfluidic technology has been widely used due to its unique advantages in the biomedical field. As a technical platform that can accurately manipulate micro-scale fluids, microfluidic biochips can not only simulate the complex environment of drugs in the body, but also record and analyze key data of the drug release process in real time. By real-time monitoring of dynamic information such as drug concentration and release rate, microfluidic biochips provide strong technical support for drug research and development. However, although microfluidic biochips have shown great potential in drug release control, current data processing methods still have many shortcomings. On the one hand, traditional data analysis methods are often limited to surface data processing and processing of a few variables, lack in-depth analysis of factors affecting drug release and evaluation of drug effects, and cannot fully reveal the complex relationship between drug release and biological responses. On the other hand, the existing data processing system lacks intelligent control strategies, making it difficult to achieve accurate and adaptive adjustment of drug release, and cannot meet the needs of personalized treatment.
[0003] For example, a Chinese patent with publication number CN115078313A discloses a biomolecule analysis system based on a microfluidic chip. The system uses (M1) a temperature control module to control the predetermined temperature change of the microfluidic chip, so that each microdroplet in the microfluidic chip can independently perform a biochemical reaction; (M2) an imaging module to irradiate the microfluidic chip with excitation light so that the microdroplets produce fluorescence, and record the fluorescence signals generated by each microdroplet in the microfluidic chip; (M3) a data processing module to perform data processing on the fluorescence signals generated by each microdroplet in the microfluidic chip acquired by the imaging module, so as to obtain an analysis result; and (M4) an output module to output the analysis result.
[0004] Through the analysis of the above-mentioned prior art, it is found that the prior art is often limited to surface data processing and the processing of a few variables, lacks in-depth analysis of factors affecting drug release and evaluation of drug effects, and the existing data processing system lacks intelligent control strategies, making it difficult to achieve accurate and adaptive adjustment of drug release; in order to solve the above problems, the present invention provides a microfluidic biochip data processing method and system based on drug release control. Summary of the invention
[0005] In view of the shortcomings of the prior art, the present invention proposes a microfluidic biochip data processing method and system based on drug release control. The method first simulates the drug release process under different environments by constructing an experimental platform, and collects and preprocesses numerical and visual data; secondly, an adaptive control model constructed by a reinforcement learning strategy is used to calculate and obtain specific control parameters for drug release to control a PID controller to achieve precise drug release; at the same time, a visualization tool is used to simulate the drug diffusion dynamics and predict the release effect in real time; finally, the diffusion and effect errors are calculated by comparing the simulation and actual data, a dynamic function is constructed to obtain the dynamic optimization point, and the dynamic optimization point is fed back to the control model in real time to optimize the drug release control parameters; the present invention realizes adaptive control and optimization of the drug release process, and improves the accuracy and efficiency of drug release.
[0006] To achieve the above object, the present invention provides the following technical solutions:
[0007] The microfluidic biochip data processing method based on drug release control comprises the following specific steps:
[0008] S1: Build a microfluidic biochip experimental platform to simulate the in vivo drug release process under different environments, collect and preprocess the numerical data of the drug release process under different environments, the actual area data of drug release visualization fluorescence imaging and the actual effect data of drug release in real time through sensors;
[0009] S2: Based on the reinforcement learning strategy, a drug release adaptive control model is constructed, and the input state, joint execution action and reward function are constructed using the preprocessed numerical data in different environments. The constructed drug release adaptive control model is pre-trained to obtain the control parameters of drug release at different times of the microfluidic biochip, and the obtained control parameters are input into the PID controller to control the drug release process of the microfluidic biochip;
[0010] S3: Use visualization tools to simulate the dynamic visualization data of drug diffusion according to the specific control parameters of drug release in the microfluidic biochip, and calculate the simulated real-time diffusion area of the drug based on the simulated dynamic visualization data of drug diffusion. At the same time, use the machine learning prediction model to make real-time predictions on the drug release effect.
[0011] S4: using the obtained real-time drug simulation diffusion area and the actual area data of drug release visualization fluorescence imaging, the diffusion error is calculated, and the effect error is obtained according to the real-time predicted drug release effect data and the actual drug release effect data;
[0012] S5: constructing a dynamic function using the obtained diffusion error and effect error to obtain a dynamic optimization point, and feeding the dynamic optimization point back to the constructed drug release adaptive control model in real time to optimize the drug release control parameters in real time;
[0013] The numerical data collected in S1 during the drug release process under different environments include real-time drug concentration, fluid flow rate in the biochip microchannel, temperature difference between the internal and external environments of the biochip, drug release rate, pressure distribution at different positions in the microchannel, drug molecule diffusion coefficient, pH value of the internal and external environments of the biochip, ion concentration change value, drug diffusion area, drug effect value, and real-time change value of simulated biological hormones in the experimental environment;
[0014] The input status in S2 is as follows:
[0015] ,
[0016] in, represents the drug concentration at the current time t, represents the drug release flow rate in the biochip microchannel at the current time t, Indicates the temperature difference between the chip and the surrounding environment at the current time t. represents the drug release rate at the current time t, represents the pressure distribution at different positions in the microchannel at the current time t, represents the diffusion coefficient of drug molecules at the current time t, Indicates the pH value of the environment inside and outside the biochip at the current time t, Indicates the change in ion concentration in the experimental environment at the current time t, represents the drug diffusion error at the current time t, represents the drug effect error value at the current time t, Indicates the real-time change value of the simulated biological hormone at the current time t; the specific joint execution actions at the current time t are ,in, represents the execution action of adjusting the drug release rate, Indicates the execution action of adjusting the microchannel flow rate, represents the execution action of switching the drug release mode; the reward function is specifically:
[0017] ,
[0018] in, is the reward function The weighting coefficient of represents the target drug concentration, represents the optimal drug release rate;
[0019] The steps of the drug release process of the microfluidic biochip controlled by the PID controller in S2 include:
[0020] S201, initialize the PID controller proportional coefficient , integral coefficient , differential coefficient , and the error between the current drug release rate at time t and the optimal drug release rate The rate of change of drug concentration The product of is used as the input of the PID controller to calculate the output control signal at the current time t , the specific formula is:
[0021] ,
[0022] in, Indicates execution of an action The assessed value obtained, Indicates execution of an action The assessed value obtained, Indicates execution of an action The assessed value obtained, , and The proportionality coefficient , integral coefficient and the differential coefficient The adjustment coefficient is, T represents the duration of a controlled release of the drug, represents the differential operator;
[0023] The specific steps for obtaining the real-time diffusion area of drug simulation in S3 include:
[0024] S301, collecting specific parameters of the drug release environment, including but not limited to initial drug concentration, initial flow rate, initial release rate, initial environmental temperature, initial pH value and initial ion concentration change value;
[0025] S302, inputting specific parameters of the drug release environment into a visualization tool to initialize the simulation environment, and at the same time inputting control parameters of drug release of the microfluidic biochip at different times calculated by the drug release adaptive control model into the visualization tool to simulate the drug diffusion process, and output a diffusion visualization image;
[0026] S303, constructing an image recognition model, and inputting the output diffusion visualization image into the image recognition model to identify and segment the drug diffusion area, and outputting the area of the drug diffusion segmentation area;
[0027] The specific steps of drug release effect prediction in S3 are:
[0028] S311, collecting historical drug release concentration, microchannel flow rate, release rate, experimental environment temperature, experimental environment pH value and corresponding drug release effect value data, and preprocessing the collected data;
[0029] S312, constructing a prediction model based on the random forest algorithm, inputting the pre-treated historical drug release concentration, microchannel flow velocity, release rate, experimental environment temperature, experimental environment pH value and the corresponding drug release effect value into the prediction model to obtain the drug release prediction effect value, and obtaining the effect error according to the drug release prediction effect value and the actual historical drug release effect value in the experiment, and setting the error threshold to 0.1 to train the prediction model;
[0030] S313, inputting the drug release concentration, microchannel flow velocity, release rate, experimental environment temperature, and experimental environment pH value collected in real time during the experiment into the trained prediction model to calculate the real-time prediction effect value of drug release;
[0031] S5 specific steps include:
[0032] S501, based on the drug diffusion error at the current time t and the drug effect error at the current time t , build dynamic functions , the specific formula is:
[0033] ,
[0034] in, and The diffusion error and effect error Corresponding weight coefficient;
[0035] S502, according to the constructed dynamic function , calculate and obtain the dynamic optimization point , the calculated optimization point Input to input state The state is updated in the system, and the updated state is input into the drug release adaptive control model to obtain the updated control parameters.
[0036] The microfluidic biochip data processing system based on drug release control includes: a data collection and preprocessing module, a model pretraining and control module, a visual simulation and prediction module, and an error calculation and feedback module; the data collection and preprocessing module is used to build an experimental platform, and collect and preprocess drug release control data, drug diffusion visualization data, and drug diffusion effect data; the model training and control module is used to build, pretrain, and control the drug release adaptive control model; the visual simulation and prediction module is used to visualize the drug release diffusion process and predict the drug release effect in real time; the error calculation and feedback module is used to calculate the drug diffusion error and effect error, and use the drug diffusion error and effect error to provide real-time feedback and update optimization of the drug release adaptive control model;
[0037] Specifically, the data collection and preprocessing module includes an experimental platform unit and a data collection and preprocessing unit; the model building and pretraining module includes a reinforcement learning unit, a control parameter output unit and a PID control unit; the visual simulation and prediction module includes a diffusion simulation unit and an effect prediction unit; the error calculation and feedback module includes an error calculation unit and an intelligent feedback unit;
[0038] The experimental platform unit is used to build the experimental platform and generate experimental data of drug release in different environments; the data collection and preprocessing unit is used for real-time collection and weight reduction and denoising preprocessing of data generated by the experimental platform; the reinforcement learning unit is used to build, train and update the drug release adaptive control model in real time; the control parameter output unit is used to calculate and output the drug release control parameters in real time according to the trained drug release adaptive control model; the PID control unit is used to control the drug release process of the microfluidic biochip using the PID controller according to the specific control parameters obtained; the diffusion simulation unit is used to simulate the dynamic visualization imaging data of drug diffusion according to the specific control parameters of the microfluidic biochip drug release obtained by using visualization tools; the effect prediction unit is used to make real-time predictions on the drug release effect using the machine learning prediction model; the error calculation unit is used to calculate the diffusion error and effect error in real time, and use the diffusion error and effect error to construct a dynamic function to obtain the dynamic optimization point; the intelligent feedback unit is used to feed back the calculated dynamic optimization point to the drug release adaptive control model in real time to update and optimize the model in real time.
[0039] An electronic device comprises a memory and a processor, wherein the memory stores a computer program and the processor implements a microfluidic biochip data processing method based on drug release control when executing the computer program.
[0040] Compared with the prior art, the present invention has the following beneficial effects:
[0041] 1. In view of the shortcomings of the prior art, the present invention realizes accurate simulation and adaptive control of the drug release process by constructing a microfluidic biochip experimental platform and combining reinforcement learning strategy with PID control. In addition, this method not only improves the accuracy and real-time performance of drug release simulation, but also intuitively displays the dynamic process of drug diffusion through visualization tools, and uses machine learning prediction models to predict drug release effects in real time, further improving the efficiency and accuracy of drug research and development. At the same time, by calculating diffusion errors and effect errors and constructing dynamic functions for real-time optimization, the continuous optimization of drug release control parameters is ensured. The accuracy and stability of drug release are improved, which helps to achieve personalized medicine and precision treatment.
[0042] 2. The present invention aims at the problem that the prior art is limited to surface data processing and a few variable processing. By collecting the real-time drug concentration, fluid flow rate in the microchannel of the biochip, the temperature difference between the internal and external environments of the biochip, the drug release rate, the pressure distribution at different positions in the microchannel, the diffusion coefficient of drug molecules, the pH value of the internal and external environments of the biochip, the change in ion concentration, the drug diffusion area, the drug effect value and the real-time change value of the simulated biological hormone in the experimental environment, and using the collected variables to construct a drug controlled release process, a more comprehensive and in-depth analysis and control of the drug release process is achieved, and the precision control of drug release is further improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 This is a flow chart of a microfluidic biochip data processing method based on drug release control according to Example 1 of the present invention;
[0044] Figure 2 This is a diagram showing the architecture of a microfluidic biochip data processing method based on drug release control according to Example 1 of the present invention;
[0045] Figure 3 This is a module structure diagram of a microfluidic biochip data processing system based on drug release control according to Example 2 of the present invention. DETAILED DESCRIPTION
[0046] Example 1
[0047] See also Figure 1-Figure 2 , an embodiment provided by the present invention: a microfluidic biochip data processing method based on drug release control, comprising the following steps:
[0048] S1: Construct a microfluidic biochip experimental platform to simulate the in vivo drug release process under different environments, collect the numerical data of the drug release process under different environments, the actual area data of drug release visualization fluorescence imaging and the actual effect data of drug release in real time through sensors and preprocess them; wherein, the numerical data collected during the drug release process under different environments include the real-time drug concentration, the fluid flow rate in the microchannel of the biochip, the temperature difference between the inside and outside of the biochip, the drug release rate, the pressure distribution at different positions in the microchannel, the diffusion coefficient of drug molecules, the pH value of the environment inside and outside the biochip, the change value of ion concentration, the drug diffusion area, the drug effect value and the real-time change value of the simulated biological hormone in the experimental environment. This implementation In this example, adrenaline is selected as a simulated biological hormone; an experimental platform of a microfluidic chip is constructed, including but not limited to a fluid control subsystem, a temperature control subsystem, and an optical detection subsystem to support drug release simulation under multiple environmental conditions; wherein the fluid control subsystem is used for the specific control process of drug release; the temperature control subsystem is used for temperature control of the entire experimental environment, and the optical detection subsystem is used for detection and visual processing of the drug diffusion process; different environments in this embodiment use the collected pH value as the main measurement variable, including a weakly acidic environment: a pH value between 5.0 and 6.5, a weakly alkaline environment: a pH value between 7.5 and 8.5, and a neutral environment: a pH value between 6.5 and 7.5;
[0049] S2: Based on the reinforcement learning strategy, a drug release adaptive control model is constructed, and the input state, joint execution action and reward function are constructed using the preprocessed numerical data in different environments. The constructed drug release adaptive control model is pre-trained to obtain the control parameters of drug release at different times of the microfluidic biochip, and the obtained control parameters are input into the PID controller to control the drug release process of the microfluidic biochip; wherein the input state is specifically:
[0050] ,
[0051] in, represents the drug concentration at the current time t, represents the drug release flow rate in the biochip microchannel at the current time t, Indicates the temperature difference between the chip and the surrounding environment at the current time t. represents the drug release rate at the current time t, represents the pressure distribution at different positions in the microchannel at the current time t, represents the diffusion coefficient of drug molecules at the current time t, Indicates the pH value of the environment inside and outside the biochip at the current time t, Indicates the change in ion concentration in the experimental environment at the current time t, represents the drug diffusion error at the current time t, represents the drug effect error value at the current time t, Indicates the real-time change value of the simulated biological hormone at the current time t;
[0052] The specific joint execution action at the current time t is ,in, represents the execution action of adjusting the drug release rate, Indicates the execution action of adjusting the microchannel flow rate, Indicates the execution action of switching the drug release mode, where the drug release mode includes two specific methods: pulse release and continuous release. When the real-time change value of the simulated biological hormone in the experimental environment changes sharply in a short period of time, the control chip is switched to the pulse release mode. When the real-time change value of the simulated biological hormone in the experimental environment remains stable for a long time, the control chip is switched to the continuous release mode. The real-time change detection of the simulated biological hormone is realized by a high-precision hormone detector.
[0053] The specific reward function is:
[0054] ,
[0055] in, is the reward function The weighting coefficient of represents the target drug concentration, represents the optimal drug release rate, where the reward function of this example aims to minimize the drug diffusion error and the drug effect error while being as close to the target drug concentration and the optimal drug release rate as possible.
[0056] S3: Use visualization tools to simulate the dynamic visualization data of drug diffusion according to the specific control parameters of drug release in the microfluidic biochip, and calculate the simulated real-time diffusion area of the drug based on the simulated dynamic visualization data of drug diffusion. At the same time, use the machine learning prediction model to make real-time predictions on the drug release effect.
[0057] S4: using the obtained real-time drug simulation diffusion area and the actual area data of drug release visualization fluorescence imaging, the diffusion error is calculated, and the effect error is obtained according to the real-time predicted drug release effect data and the actual drug release effect data;
[0058] S5: A dynamic function is constructed using the obtained diffusion error and effect error to obtain a dynamic optimization point, and the dynamic optimization point is fed back in real time to the constructed drug release adaptive control model to optimize the drug release control parameters in real time.
[0059] The drug release adaptive control model is constructed based on the SCA algorithm, which specifically includes a strategy network, a target strategy network, a first evaluation network, a first target evaluation network, a second evaluation network, a second target evaluation network and an improved experience replay pool B. The strategy network and the target strategy network are composed of a first neural network with the same structure. The first neural network includes a first convolutional layer with a convolution kernel size of 32, a ReLU layer, a Gaussian distribution layer, and a Tanh layer. The first evaluation network, the first target evaluation network, the second evaluation network, and the second target evaluation network are composed of a second neural network with the same structure. The second neural network includes a first convolutional layer, a ReLU layer, a Gaussian distribution layer, and a Tanh layer. U layer, a second convolution layer with a convolution kernel size of 16, a ReLU layer and a second convolution layer; the first evaluation network corresponds to the first target evaluation network, the second evaluation network corresponds to the second target evaluation network, and the improved experience replay pool B includes a historical experience layer replay pool and a real-time experience layer replay pool, wherein the historical experience layer replay pool is used to store the experience values generated by the collected historical drug release control data and environmental data, and the real-time experience layer replay pool is used to store the experience data generated by controlling the drug release of the chip at the current moment; and each time a drug release control process is completed, the data stored in the real-time experience layer is transferred to the historical experience layer for storage.
[0060] The steps of the drug release process of the microfluidic biochip controlled by the PID controller in S2 include:
[0061] S201, initialize the PID controller proportional coefficient , integral coefficient , differential coefficient , and the error between the current drug release rate at time t and the optimal drug release rate The rate of change of drug concentration The product of is used as the input of the PID controller to calculate the output control signal at the current time t , the specific formula is:
[0062] ,
[0063] in, Indicates execution of an action The assessed value obtained, Indicates execution of an action The assessed value obtained, Indicates execution of an action The assessed value obtained, , and The proportionality coefficient , integral coefficient and the differential coefficient The adjustment coefficient is, T represents the duration of a controlled release of the drug, represents a differential operator; the setting of the optimal drug release rate is determined by multiple artificial experiments according to the type of drug released in the experiment;
[0064] The specific steps for obtaining the real-time diffusion area of drug simulation in S3 include:
[0065] S301, collecting specific parameters of the drug release environment, including but not limited to initial drug concentration, initial flow rate, initial release rate, initial environmental temperature, initial pH value and initial ion concentration change value;
[0066] S302, inputting specific parameters of the drug release environment into a visualization tool to initialize the simulation environment, and at the same time inputting control parameters of drug release of the microfluidic biochip at different times calculated by the drug release adaptive control model into the visualization tool to simulate the drug diffusion process, and output a diffusion visualization image;
[0067] S303, constructing an image recognition model, and inputting the output diffusion visualization image into the image recognition model to identify and segment the drug diffusion area, and outputting the area of the drug diffusion segmentation area; specifically, the steps for calculating the area of the drug diffusion segmentation area are as follows:
[0068] 1) First, the identified and segmented image is binarized, and the drug diffusion area and background are converted to black and white respectively using Python. Then, the number of black pixels (i.e., drug diffusion area) in the binarized image is counted.
[0069] 2) According to the resolution of the image (i.e. the actual area represented by each pixel), the number of black pixels is converted into the actual area. The calculation formula is as follows:
[0070] ,
[0071] in represents the area of drug diffusion segmentation region, q represents the number of black pixels (i.e. drug diffusion area), It indicates the area represented by a single black pixel. The actual area of a single pixel is determined according to the image resolution processed in the experiment.
[0072] This process collects specific parameters of the drug release environment, uses visualization tools to simulate the drug diffusion process, and combines image recognition models to identify and segment the drug diffusion area, thereby achieving accurate acquisition of the real-time diffusion area of drug simulation. This process not only improves the efficiency and accuracy of drug research and development, but also provides strong support for the evaluation of drug diffusion effects, helping to promote innovation and development in the pharmaceutical field.
[0073] The specific steps of drug release effect prediction in S3 are:
[0074] S311, collecting historical drug release concentration, microchannel flow rate, release rate, experimental environment temperature, experimental environment pH value and corresponding drug release effect value data, and preprocessing the collected data;
[0075] S312, constructing a prediction model based on the random forest algorithm, inputting the pre-treated historical drug release concentration, microchannel flow velocity, release rate, experimental environment temperature, experimental environment pH value and the corresponding drug release effect value into the prediction model to obtain the drug release prediction effect value, and obtaining the effect error according to the drug release prediction effect value and the actual historical drug release effect value in the experiment, and setting the error threshold to 0.1 to train the prediction model;
[0076] S313, inputting the drug release concentration, microchannel flow velocity, release rate, experimental environment temperature, and experimental environment pH value collected in real time during the experiment into the trained prediction model, and calculating the real-time prediction effect value of drug release; further, the real-time code of this process is as follows:
[0077] Import the Python code library:
[0078] import pandas as pd
[0079] from sklearn.model_selection import train_test_split
[0080] from sklearn.ensemble import RandomForestRegressor
[0081] from sklearn.metrics import mean_squared_error;
[0082] S311: Collect and preprocess data:
[0083] data = pd.read_csv('drug_release_data.csv'),
[0084] Data preprocessing: fill missing values with mean:
[0085] imputer = SimpleImputer(strategy='mean')
[0086] data_filled = imputer.fit_transform(data)
[0087] data_df = pd.DataFrame(data_filled, columns=data.columns) ,
[0088] Data preprocessing: Standardization features:
[0089] scaler = StandardScaler(),
[0090] Split features and target variables:
[0091] X = data[['drug release concentration', 'microchannel flow rate', 'release rate', 'experimental environment temperature', 'experimental environment pH value']],
[0092] y = data['Effect value of drug release'],
[0093] S312: Build and train a prediction model based on the random forest algorithm:
[0094] Divide the training set and the test set into a ratio of 8:2.
[0095] X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42),
[0096] Create a random forest regression model and set the number of decision trees to 100 and the random parameter to 42:
[0097] rf = RandomForestRegressor(n_estimators=100, random_state=42) ,
[0098] Fit the training model:
[0099] rf.fit(X_train, y_train),
[0100] Evaluate model performance:
[0101] y_pred = rf.predict(X_test),
[0102] mse = mean_squared_error(y_test, y_pred),
[0103] print(f'Mean Squared Error on test set: {mse}'),
[0104] Set the error threshold to 0.1:
[0105] error_threshold = 0.1,
[0106] historical_effects = y_test,
[0107] predicted_effects = rf.predict(X_test),
[0108] Calculate the effect error:
[0109] errors = abs(predicted_effects - historical_effects),
[0110] Check if any errors exceed the threshold:
[0111] if (errors > error_threshold).any():
[0112] print("Error threshold exceeded for some samples. Model may need further adjustment."),
[0113] else:
[0114] print("All errors are within the acceptable threshold."),
[0115] S313: Use the trained model for real-time prediction:
[0116] Assume that the real-time collected data is a DataFrame named new_drug_release_data.csv.
[0117] real_time_data = pd.DataFrame('new_drug_release_data.csv'),
[0118] Use the model to make real-time predictions:
[0119] real_time_predictions = rf.predict(real_time_data),
[0120] print("Real-time predicted drug release effect value:"),
[0121] print(real_time_predictions).
[0122] S5 specific steps include:
[0123] S501, based on the drug diffusion error at the current time t and the drug effect error at the current time t , build dynamic functions , the specific formula is:
[0124] ,
[0125] in, and The diffusion error and effect error Corresponding weight coefficient;
[0126] S502, according to the constructed dynamic function , calculate and obtain the dynamic optimization point , the calculated optimization point Input to input state The state is updated in the system, and the updated state is input into the drug release adaptive control model to obtain the updated control parameters.
[0127] This process not only improves the accuracy of drug release control, but also enhances the system's adaptability, which helps to optimize drug release effects and improve drug research and development efficiency.
[0128] Example 2
[0129] See also Figure 3 , another embodiment provided by the present invention: a microfluidic biochip data processing system based on drug release control, comprising: a data collection and preprocessing module, a model pretraining and control module, a visual simulation and prediction module, and an error calculation and feedback module; the data collection and preprocessing module is used to build an experimental platform, and collect and preprocess drug release control data, drug diffusion visualization data, and drug diffusion effect data; the model training and control module is used to build, pretrain, and control the drug release adaptive control model; the visual simulation and prediction module is used for visual simulation of the drug release diffusion process and real-time prediction of the drug release effect; the error calculation and feedback module is used to calculate the drug diffusion error and the effect error, and use the drug diffusion error and the effect error to perform real-time feedback and update optimization on the drug release adaptive control model;
[0130] The data collection and preprocessing module includes an experimental platform unit and a data collection and preprocessing unit; the model building and pretraining module includes a reinforcement learning unit, a control parameter output unit and a PID control unit; the visual simulation and prediction module includes a diffusion simulation unit and an effect prediction unit; the error calculation and feedback module includes an error calculation unit and an intelligent feedback unit;
[0131] The experimental platform unit is used to build the experimental platform and generate experimental data of drug release in different environments; the data collection and preprocessing unit is used for real-time collection and weight reduction and denoising preprocessing of data generated by the experimental platform; the reinforcement learning unit is used to build, train and update the drug release adaptive control model in real time; the control parameter output unit is used to calculate and output the drug release control parameters in real time according to the trained drug release adaptive control model; the PID control unit is used to control the drug release process of the microfluidic biochip using the PID controller according to the specific control parameters obtained; the diffusion simulation unit is used to simulate the dynamic visualization imaging data of drug diffusion according to the specific control parameters of the microfluidic biochip drug release obtained by using visualization tools; the effect prediction unit is used to make real-time predictions on the drug release effect using the machine learning prediction model; the error calculation unit is used to calculate the diffusion error and effect error in real time, and use the diffusion error and effect error to construct a dynamic function to obtain the dynamic optimization point; the intelligent feedback unit is used to feed back the calculated dynamic optimization point to the drug release adaptive control model in real time to update and optimize the model in real time.
[0132] The system first builds an experimental environment through the experimental platform unit to simulate the drug release process under different conditions. Secondly, the data collection and preprocessing unit is used to collect experimental data in real time, and preprocessing is performed to reduce weight and remove noise to ensure the accuracy and reliability of the data. Thirdly, the reinforcement learning unit uses the collected data to build and train the drug release adaptive control model, and the control parameter output unit is used to calculate and output the drug release control parameters in real time according to the model. Fourthly, after the PID control unit receives the control parameters, the PID controller is used to accurately control the drug release process in the microfluidic biochip. Fifthly, the diffusion simulation unit uses visualization tools to simulate the dynamic process of drug diffusion according to the acquired control parameters, and the effect prediction unit uses the machine learning model to predict the drug release effect in real time. Finally, the error calculation unit calculates the diffusion error and effect error according to the acquired diffusion data and effect prediction, and uses the diffusion error and effect error to calculate the dynamic optimization point. The intelligent feedback unit uses the calculated dynamic optimization point to feed back the calculated dynamic optimization point to the drug release adaptive control model in real time to optimize the control model in real time.
[0133] The system simulates drug release under different environmental conditions through the experimental platform, providing a real and reliable experimental basis for data collection; the introduction of data collection and preprocessing units ensures the purity and accuracy of the data, providing solid support for subsequent model training; the collaborative work of the reinforcement learning unit and the PID control unit enables the drug release control model to be adaptively optimized and achieve precise control; at the same time, the use of visual simulation and prediction modules allows researchers to intuitively observe the drug diffusion process and predict the release effect, greatly enhancing the scientificity and accuracy of decision-making; most importantly, the real-time effect of the error calculation and feedback module enables the system to dynamically adjust the control model according to the real-time error, continuously optimize the drug release process, and ultimately achieve precise and efficient control of drug release; this process not only improves the efficiency and success rate of drug research and development, but also provides patients with safer and more effective treatment options, promoting progress and development in the medical field.
[0134] Example 3
[0135] A computer-readable storage medium, characterized in that computer instructions are stored thereon, and when the computer instructions are executed, a microfluidic biochip data processing method based on drug release control is executed.
[0136] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the scope of protection of the purpose of the present invention and the claims, and all of these are within the protection of the present invention.
Claims
1. A microfluidic biochip data processing method based on drug release control, characterized in that: The specific steps include: S1: Build a microfluidic biochip experimental platform to simulate the in vivo drug release process under different environments, collect and preprocess the numerical data of the drug release process under different environments, the actual area data of drug release visualization fluorescence imaging and the actual effect data of drug release in real time through sensors; S2: Based on the reinforcement learning strategy, a drug release adaptive control model is constructed, and the input state, joint execution action and reward function are constructed using the preprocessed numerical data in different environments. The constructed drug release adaptive control model is pre-trained to obtain the control parameters of drug release at different times of the microfluidic biochip, and the obtained control parameters are input into the PID controller to control the drug release process of the microfluidic biochip; S3: Use visualization tools to simulate the dynamic visualization data of drug diffusion according to the specific control parameters of drug release in the microfluidic biochip, and calculate the simulated real-time diffusion area of the drug based on the simulated dynamic visualization data of drug diffusion. At the same time, use the machine learning prediction model to make real-time predictions on the drug release effect. S4: using the obtained real-time drug simulation diffusion area and the actual area data of drug release visualization fluorescence imaging, the diffusion error is calculated, and the effect error is obtained according to the real-time predicted drug release effect data and the actual drug release effect data; S5: constructing a dynamic function using the obtained diffusion error and effect error to obtain a dynamic optimization point, and feeding the dynamic optimization point back to the constructed drug release adaptive control model in real time to update and optimize drug release control parameters in real time; The numerical data collected in S1 during the drug release process under different environments include real-time drug concentration, fluid flow rate in the biochip microchannel, temperature difference between the internal and external environments of the biochip, drug release rate, pressure distribution at different positions in the microchannel, drug molecular diffusion coefficient, pH value of the internal and external environments of the biochip, ion concentration change value, drug diffusion area, drug effect value and real-time change value of simulated biological hormones in the experimental environment; The input state in S2 is specifically: , in, represents the drug concentration at the current time t, represents the drug release flow rate in the biochip microchannel at the current time t, Indicates the temperature difference between the chip and the surrounding environment at the current time t. represents the drug release rate at the current time t, represents the pressure distribution at different positions in the microchannel at the current time t, represents the diffusion coefficient of drug molecules at the current time t, Indicates the pH value of the environment inside and outside the biochip at the current time t, Indicates the change in ion concentration in the experimental environment at the current time t, represents the drug diffusion error at the current time t, represents the drug effect error value at the current time t, represents the real-time change value of the simulated biological hormone at the current time t; the joint execution action at the current time t is specifically: ,in, represents the execution action of adjusting the drug release rate, Indicates the execution action of adjusting the microchannel flow rate, represents the execution action of switching the drug release mode; the reward function is specifically: , in, is the reward function The weighting coefficient of represents the target drug concentration, represents the optimal drug release rate; The steps of controlling the drug release process of the microfluidic biochip by the PID controller in S2 include: S201, initialize the PID controller proportional coefficient , integral coefficient , differential coefficient , and the error between the current drug release rate at time t and the optimal drug release rate The rate of change of drug concentration The product of is used as the input of the PID controller to calculate the output control signal at the current time t , the specific formula is: , in, Indicates execution of an action The assessed value obtained, Indicates execution of an action The assessed value obtained, Indicates execution of an action The assessed value obtained, , and The proportionality coefficient , integral coefficient and the differential coefficient The adjustment coefficient is, T represents the duration of a controlled release of the drug, represents the differential operator; The specific steps for obtaining the real-time diffusion area of drug simulation in S3 include: S301, collecting specific parameters of the drug release environment, including but not limited to initial drug concentration, initial flow rate, initial release rate, initial environmental temperature, initial pH value and initial ion concentration change value; S302, inputting specific parameters of the drug release environment into a visualization tool to initialize the simulation environment, and at the same time inputting control parameters of drug release of the microfluidic biochip at different times calculated by the drug release adaptive control model into the visualization tool to simulate the drug diffusion process, and output a diffusion visualization image; S303, constructing an image recognition model, and inputting the output diffusion visualization image into the image recognition model to identify and segment the drug diffusion area, and outputting the area of the drug diffusion segmentation area; The specific steps of drug release effect prediction in S3 are: S311, collecting historical drug release concentration, microchannel flow rate, release rate, experimental environment temperature, experimental environment pH value and corresponding drug release effect value data, and preprocessing the collected data; S312, constructing a prediction model based on the random forest algorithm, inputting the pre-treated historical drug release concentration, microchannel flow velocity, release rate, experimental environment temperature, experimental environment pH value and the corresponding drug release effect value into the prediction model to obtain the drug release prediction effect value, and obtaining the effect error according to the drug release prediction effect value and the actual historical drug release effect value in the experiment, and setting the error threshold to 0.1 to train the prediction model; S313, inputting the drug release concentration, microchannel flow velocity, release rate, experimental environment temperature, and experimental environment pH value collected in real time during the experiment into the trained prediction model to calculate the real-time prediction effect value of drug release; The specific steps of S5 include: S501, based on the drug diffusion error at the current time t and the drug effect error at the current time t , build dynamic functions , the specific formula is: , in, and The diffusion error and effect error Corresponding weight coefficient; S502, according to the constructed dynamic function , calculate and obtain the dynamic optimization point , the calculated optimization point Input to input state The state is updated in the system, and the updated state is input into the drug release adaptive control model to obtain the updated control parameters.
2. A microfluidic biochip data processing system based on drug release control, which is implemented based on the microfluidic biochip data processing method based on drug release control described in claim 1, characterized in that: include: Data collection and preprocessing module, model pretraining and control module, visual simulation and prediction module and error calculation and feedback module; the data collection and preprocessing module is used to build an experimental platform and collect and preprocess drug release control data, drug diffusion visualization data and drug diffusion effect data; the model training and control module is used to build, pretrain and control the drug release adaptive control model; the visual simulation and prediction module is used for visual simulation of the drug release diffusion process and real-time prediction of the drug release effect; the error calculation and feedback module is used to calculate the drug diffusion error and effect error, and use the drug diffusion error and effect error to provide real-time feedback and update optimization of the drug release adaptive control model.
3. The microfluidic biochip data processing system based on drug release control according to claim 2, characterized in that: The data collection and preprocessing module includes an experimental platform unit and a data collection and preprocessing unit; the model building and pretraining module includes a reinforcement learning unit, a control parameter output unit and a PID control unit; the visual simulation and prediction module includes a diffusion simulation unit and an effect prediction unit; the error calculation and feedback module includes an error calculation unit and an intelligent feedback unit; The experimental platform unit is used for building the experimental platform and generating experimental data of drug release in different environments; the data collection and preprocessing unit is used for real-time collection and weight reduction and denoising preprocessing of data generated by the experimental platform; the reinforcement learning unit is used for building, training and real-time updating of the drug release adaptive control model; the control parameter output unit is used for real-time calculation and output of drug release control parameters according to the trained drug release adaptive control model; the PID control unit is used for controlling the drug release process of the microfluidic biochip using a PID controller according to the acquired specific control parameters; the diffusion simulation unit is used for simulating the dynamic visualization imaging data of drug diffusion according to the acquired specific control parameters of the microfluidic biochip drug release using a visualization tool; the effect prediction unit is used for real-time prediction of the drug release effect using a machine learning prediction model; the error calculation unit is used for real-time calculation of the diffusion error and the effect error, and the use of the diffusion error and the effect error to construct a dynamic function to obtain the dynamic optimization point; the intelligent feedback unit is used for real-time feedback of the calculated dynamic optimization point to the drug release adaptive control model to update and optimize the model in real time.
4. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the computer instructions are executed, the steps of the microfluidic biochip data processing method based on drug release control as described in claim 1 are executed.
