Cell culture optimization method and system based on brain-computer interface
By using a brain-computer interface-based cell culture optimization method, and leveraging artificial neural networks and multiple algorithms to generate chemical reagent ratios and physical environment control strategies, the low precision problem in existing adherent culture methods is solved, achieving efficient and stable cell culture suitable for brain-computer interface research.
Patent Information
- Application Number
- CN202511672072.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-14
- Publication Date
- 2026-02-10
AI Technical Summary
Existing adherent culture methods suffer from low precision and poor consistency in chemical reagent addition and light environment control, resulting in unstable cell quality and difficulty in meeting the demanding requirements of brain-computer interfaces and other applications.
A brain-computer interface-based cell culture optimization method is adopted, which utilizes artificial neural networks and multi-algorithm collaborative decision-making to generate chemical reagent ratios and physical environment control strategies. Automated control is achieved through a precise reagent ratio module and a closed-loop photoelectric multimodal control module, forming a closed-loop iterative optimization.
It achieves consistency in the concentration and reaction time of the chemical microenvironment, avoids cell stress caused by light fluctuations, ensures the activity of high-quality cells and batch-to-batch consistency, is suitable for brain-computer interface research, shortens the optimization cycle, and saves costs.
Smart Images
Figure CN121495824A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of cell culture automation and intelligent control, and particularly relates to a cell culture optimization method and system based on a brain-computer interface. BACKGROUND
[0002] Cell culture is one of the core supporting technologies for life science research and clinical medical development, which provides a basic platform for revealing cell physiological mechanisms, developing disease diagnosis and treatment schemes, and biological products. However, the cell growth process has high sensitivity, and its state is complexly regulated by various environmental factors such as temperature, nutrition, osmotic pressure, and the like. Therefore, how to achieve precise control and stable maintenance of conditions during the culture process is the key to obtaining reliable experimental data and promoting scientific research transformation.
[0003] In the prior art, the adherent culture method has become one of the most widely used methods in basic research and drug screening fields due to its simple operation, low cost, and good maintenance of cell function. A typical adherent culture process involves steps such as neuron separation and primary culture, immunofluorescence identification, lentivirus transfection, and blue light irradiation activation.
[0004] However, this method relies heavily on the experience of the operator, and has inherent defects in the following two key links, resulting in low culture efficiency and unstable cell quality: 1. Insufficient controllability of chemical microenvironment: The addition precision and consistency of chemical reagents (such as digestive enzymes, serum, antibodies, etc.) are difficult to guarantee. The slight deviation of reagent concentration and action time will directly lead to problems such as excessive or insufficient cell digestion, poor adhesion, uneven transfection efficiency, and even distorted identification results. For example, papain preparation error or FBS batch difference will directly affect cell survival and adhesion; improper concentration control of primary antibody / secondary antibody will interfere with cell purity determination.
[0005] 2. Lack of stability and precision of physical microenvironment (especially light): For cells such as neurons expressing light-sensitive proteins (such as ChR2) through lentivirus transfection, precise control of light parameters is crucial. The existing technology is difficult to effectively control: during the culture stage, the incubator is not well protected from light, or the environmental light introduced by daily observation may activate ChR2 in advance, inducing cell stress and damaging neurite structure; during the detection stage, the immunofluorescence sample may be quenched due to improper storage (such as light leakage), and during the patch clamp recording, uncontrolled environmental light will directly induce action potentials, interfering with the collection of normal electrophysiological signals.
[0006] In addition, the problem of synergistic interference of chemical and physical factors is particularly prominent in the transition from culture to subsequent functional detection. For example, if the pH and ion concentration of the extracellular fluid changed greatly after the patch clamp experiment, it would cause transient stress to the cells and affect the accuracy of the resting membrane potential recording.
[0007] In summary, the existing adherent culture method has the core problems of low precision and poor consistency in chemical agent addition and light environment control, making it difficult to obtain high-quality cells with consistent activity and complete function stably and efficiently. Especially for brain-computer interface and other downstream applications with strict requirements for cell state, it has become a major technical bottleneck. SUMMARY
[0008] The purpose of the present application is to provide a brain-computer interface cell culture optimization method and system to solve the core problems of low cell quality, large batch-to-batch difference, and difficulty in adapting to downstream application requirements such as brain-computer interface caused by manual operation errors, inaccurate chemical microenvironment control, and unstable physical light environment in existing adherent culture technology.
[0009] To achieve the above-mentioned purpose, the present application provides a brain-computer interface cell culture optimization method, comprising the following steps: Step S1, input the cell type and culture target, and call the culture basic parameters and historical experimental data in the multi-cell type parameter database; Step S2, based on the cell type, culture target and historical experimental data in step S1, the feature learning is performed through the artificial neural network module, and the culture basic parameter range of the cell type is output; Step S3, taking the culture basic parameter range output in step S2 as the optimization boundary, the multi-algorithm optimization module is processed to generate an executable chemical agent proportioning scheme and a physical environment control strategy; Step S4, the chemical agent proportioning scheme generated in step S3 is received and executed by the agent precise proportioning module, and the automatic weighing, dilution, pH adjustment and timed and quantitative addition of the agent are completed; Step S5, the physical environment control strategy generated in step S3 is received and executed by the closed-loop light-electricity multi-modal control module, the light parameters are monitored and dynamically adjusted in real time, and the stability of temperature, humidity and CO2 concentration is maintained; Step S6, the cell culture effect after step S4 and step S5 is detected by the result verification and feedback module, and the detection data is obtained; Step S7, the detection data obtained in step S6 is fed back to the multi-cell type parameter database and the intelligent decision unit as new historical experimental data, which is used to optimize the model parameters of the artificial neural network module and the multi-algorithm optimization module, forming a closed-loop iterative optimization.
[0010] Preferably, in step S2, the artificial neural network module uses NIL. PLS The TDC hybrid architecture performs feature learning on the input data and outputs a range of basic cell culture parameters; NIL PLS The TDC hybrid architecture includes a nonlinear feature interaction learning module, a partial least squares feature dimensionality reduction module, and a time-series dynamic correction module, which are used to process the input cell type, culture target, and historical experimental data to obtain the basic parameter range used to guide subsequent optimization.
[0011] Preferably, in step S3, the multi-algorithm optimization module uses NSGA. II and TOPSIS algorithms are used to jointly optimize chemical reagent parameters, and deep reinforcement learning and random forest algorithms are used to jointly optimize physical environment parameters.
[0012] Preferably, in step S4, the precise drug proportioning module weighs the drug through the automatic weighing module, dilutes and adjusts the pH through the precise dilution and mixing submodule, and adds the drug according to the set time and dosage through the quantitative addition and timing submodule.
[0013] Preferably, the precise drug mixing module includes one or more of the following: a high-precision electronic balance, a liquid flow meter, an automatic acid / alkali dispensing device, a magnetic stirrer, a pH sensor, a pipette, and a peristaltic pump.
[0014] Preferably, in step S5, the closed-loop photoelectric multimodal control module collects light data in real time through the light parameter monitoring submodule, shields harmful wavelengths and precisely controls the light source through the light dynamic adjustment submodule, and maintains the stability of temperature, humidity and CO2 concentration through the comprehensive environmental stabilization submodule.
[0015] Preferably, the closed-loop photoelectric multimodal control module includes a light sensor, an LED light source controller, a light-shielding device, and temperature, humidity, and CO2 concentration sensors.
[0016] Preferably, step S7 includes the following steps: Step S71: Classify and store the detection data obtained in step S6; Step S72: Compare and analyze the detection data with historical data to identify state deviations during the cultivation process; Step S73: Update the model parameters of the artificial neural network module and the multi-algorithm optimization module using the detection data; Step S74: Based on the updated model parameters, adjust the optimization strategy of the intelligent decision-making unit; Step S75: Apply the optimized strategy to the subsequent cell culture process to achieve closed-loop iterative optimization.
[0017] This invention also provides a brain-computer interface-based cell culture optimization system, comprising: The input and data unit includes an input module for receiving cell types and culture objectives, and a multi-cell type parameter database for storing basic culture parameters and historical experimental data; The intelligent decision-making unit includes an artificial neural network module connected to the input and data unit and a multi-algorithm optimization module. The artificial neural network module is used to predict the range of basic parameters based on the input data, and the multi-algorithm optimization module is used to generate chemical reagent ratio schemes and physical environment control strategies with the range of basic parameters as the boundary. The execution unit includes a precise drug ratio module and a closed-loop photo-electric multimodal control module connected to the intelligent decision-making unit, which are used to execute the chemical drug ratio scheme and the physical environment control strategy, respectively. The result verification and feedback unit, connected to the execution unit and the input and data unit, is used to detect the cell culture effect and feed the detection data back to the multi-cell type parameter database and the intelligent decision-making unit to form a closed-loop optimization.
[0018] Preferably, the result verification and feedback unit includes: The basic state detection submodule is used to detect cell viability, adhesion rate, and purity; the functional integrity detection submodule is used to detect cell action potentials and light stimulation response. The deviation analysis and parameter adjustment submodule is used to automatically analyze the causes of cell state deviations and generate real-time adjustment instructions; The data feedback and model iteration submodule is used to feed back the monitored and adjusted new data to the multi-cell type parameter database, and to retrain the artificial neural network module and optimize the multi-algorithm optimization module using the new data.
[0019] Therefore, the present invention employs the above-described brain-computer interface-based cell culture optimization method and system, and the beneficial technical effects are as follows: (1) This invention automatically generates the optimal chemical reagent ratio and physical environment parameters through collaborative decision-making using artificial neural networks and multiple algorithms; it utilizes a precise reagent ratio module to achieve automatic and high-precision addition of digestive enzymes, antibodies, and other reagents, fundamentally eliminating human operation errors and ensuring the consistency of chemical microenvironment concentration and action time; simultaneously, it monitors and dynamically adjusts illumination parameters in real time through a closed-loop photo-electric multimodal regulation module, effectively shielding harmful wavelengths and avoiding photosensitive cell stress, fluorescence quenching, or functional damage caused by light fluctuations. This series of technical features work synergistically to directly solve the core problems of insufficient controllability of the chemical microenvironment and lack of stability of the physical microenvironment in existing technologies, enabling the stable production of high-activity, functionally complete, and batch-to-batch consistent high-quality cells, which are particularly suitable for research such as brain-computer interfaces.
[0020] (2) This invention integrates intelligent decision-making based on historical data, automated and precise execution, result verification based on multi-dimensional indicators, and model optimization based on feedback data into an organic whole. The system can automatically analyze deviations and iteratively optimize the algorithm model and culture parameters based on data such as cell viability and electrophysiological function collected by the result verification and feedback module. This closed-loop technical feature enables the system to continuously improve itself, get rid of excessive reliance on the personal experience of operators, avoid the lengthy trial and error process of repeated experiments in traditional methods, thereby shortening the optimization cycle, saving reagent and time costs, and realizing the standardization and intelligence of the culture process.
[0021] (3) The optimization strategy of this invention can adapt to the specific needs of different adherent cells (such as epithelial cells and light-sensitive neurons) and generate personalized culture programs. More importantly, the system takes into account the needs of subsequent functional testing (such as patch clamp) when making decisions. By controlling the extracellular fluid parameters (pH, ion concentration) and the light intensity of the detection environment in advance, it ensures a smooth transition of cells from the culture stage to the functional testing stage, effectively preventing the problems of instantaneous cell stress and electrophysiological signal interference mentioned in the background art. It provides reliable and stable cell materials for downstream applications such as drug screening and disease mechanism research, and expands the application scope and value of the technology. Attached Figure Description
[0022] Figure 1 This is a diagram of the overall system framework. Figure 2 A flowchart for an intelligent decision-making unit; Figure 3 A flowchart for the execution unit; Figure 4 This is a flowchart for the results verification and feedback unit. Detailed Implementation
[0023] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.
[0024] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.
[0025] Example 1 like Figures 1-4 A brain-computer interface-based cell culture optimization method includes the following steps: Step S1: Input cell type and culture objective, and call up basic culture parameters and historical experimental data from the multi-cell type parameter database.
[0026] Step S2: Based on the cell type, culture objective, and historical experimental data in Step S1, feature learning is performed through an artificial neural network module to output the basic range of culture parameters for the cell type.
[0027] The artificial neural network module uses NIL PLS The TDC hybrid architecture performs feature learning on the input data and outputs a range of basic cell culture parameters.
[0028] NIL PLS The TDC hybrid architecture sequentially includes a nonlinear feature interaction learning module, a partial least squares feature dimensionality reduction module, and a time-series dynamic correction module. These modules are used to process the input cell type, culture objective, and historical experimental data to obtain the basic parameter range used to guide subsequent optimization. The specific process is as follows: Input data preprocessing.
[0029] (1) Perform one-hot encoding on the classification data (including cell type and light-sensitive properties); Let the set of cell types be (like Neuron, (cells), for any cell type The encoded vector is: .
[0030] (2) Mining the numerical data (including drug concentration and light intensity). Max normalization maps the original values to the interval [0, 1].
[0031] Example: If the original value of the papain concentration numerical characteristic is 0.2%, the minimum characteristic value is 0.1%, and the maximum characteristic value is 0.3%, then the normalized value is 0.5.
[0032] (3) Based on 3 The principle is to detect and remove outliers from historical data.
[0033] NIL module (Nonlinear Feature Interaction Learning).
[0034] (1) Construct a set of second-order interaction terms among the preprocessed basic features.
[0035] (2) Calculate the covariance weights of interaction terms and training objectives based on the attention mechanism, retain highly correlated interaction terms (attention weight greater than 80%), and output the weighted interaction feature vector.
[0036] PLS module (Partial Least Squares Feature Reduction).
[0037] (1) Construct the independent variable matrix (integrating basic features and weighted interaction features) and the dependent variable matrix (cell state indicators).
[0038] (2) Iteratively extract the principal components of the independent and dependent variables, maximize their covariance, and update the residual matrix until the cumulative variance contribution rate reaches the set threshold.
[0039] Iterative solution of principal components of independent variables With the principal components of the dependent variable ,satisfy: .
[0040] First round of component calculation: ; ; ; ; Residual matrix update: ; ; in, Represents the principal components of the independent variable. Indicates drip Round of iterative extraction principal component, Represents the matrix of independent variables. Represents the dependent variable matrix. This represents the weight vector of the principal components of the independent variables. This represents the weight vector of the principal components of the dependent variable. Indicates the principal components of the dependent variable. Indicates the first round The weight vector of the principal components, Indicates the first round The weight vector of the principal components, Indicates the first Wheel independent variable load vector, Indicates the first Wheel dependent variable load vector, Indicates the first The western bright matrix during round iteration, Indicates the first The dependent variable matrix during round iteration, Indicates the first The residual matrix of the independent variables, Indicates the first The residual matrix of the dependent variable.
[0041] (3) Based on the independent variable loading vector ,reserve The features are analyzed, and a low-dimensional key feature matrix is output. .
[0042] TDC module (Time Dynamic Correction).
[0043] (1) Based on historical time series data ( The temporal deviation between the culture stage (e.g., digestion period / adherence period) and the LSTM output is represented, and the temporal correction coefficient for each culture stage is calculated. ; in, Represents the timing correction coefficient. This indicates the timing skew of the LSTM output. This represents the historical average of the characteristics of this stage.
[0044] (2) The characteristic mean of PLS output Perform corrections and constrain the standard deviation after correction to ensure that the parameter fluctuation range meets the set requirements.
[0045] Corrected mean : ; Corrected standard deviation constraints : ; The output integrates decision-making and basic parameter ranges.
[0046] (1) Constructing the fusion feature vector Includes TDC-corrected features With NIL attention weights ; ; in, This indicates transpose.
[0047] (2) Perform Sigmoid activation mapping on the fused feature vector.
[0048] (3) Combine the mean of the corrected features Standard deviation and cell physiological tolerance threshold Calculate and output the final range of basic parameters to ensure batch compliance rate.
[0049] ; ; in, Indicates a safe fluctuation range to ensure Batch meets standards , These represent the lower and upper limits of the final calculated range of basic parameters, respectively.
[0050] Step S3: Using the range of basic cultivation parameters output in step S2 as the optimization boundary, the system is processed through a multi-algorithm optimization module to generate an executable chemical reagent ratio scheme and physical environment control strategy.
[0051] The multi-algorithm optimization module uses NSGA II and TOPSIS algorithms are used to jointly optimize chemical reagent parameters, and deep reinforcement learning and random forest algorithms are used to jointly optimize physical environment parameters.
[0052] NSGA The specific process for co-optimizing chemical reagent parameters using II and TOPSIS is as follows: Optimization goals and parameter boundaries were set as follows: The main optimization goals were cell viability ≥90%, adhesion rate ≥85%, and single-batch digestive enzyme usage ≤0.5g. The optimization parameter was limited to a papain concentration of 0.19%. 0.23%, FBS concentration 12.5% 13.5%, primary antibody dilution ratio 1:180 1:220.
[0053] Multi-objective initial screening based on NSGA-II algorithm: Using a multi-objective genetic algorithm, under the preset population size, number of iterations, crossover and mutation probabilities, the performance of each parameter combination is quantitatively evaluated using a nonlinear fitness function to screen out several sets of non-dominated solutions; further, candidate solution sets that meet the conditions are retained according to the crowding threshold.
[0054] The optimal parameter combination is determined based on the TOPSIS algorithm: a decision matrix is constructed for the candidate solution set, and weights are assigned in combination with factors such as cell viability, adhesion rate and cost. Data standardization is performed and positive and negative ideal solutions are calculated. Finally, the optimal parameter combination is determined based on the proximity (e.g., papain 0.21%, FBS 13.1%, primary antibody 1:200).
[0055] The specific process of DRL and random forest co-optimizing physical environment parameters is as follows: Setting optimization targets and parameter boundaries: The targets are a light responsivity ≥ 85% and a lactic acid concentration ≤ 2 mmol / L. The optimization parameters are limited to the range of the basic parameters output in step S2 (including the illumination wavelength of 468 nm). 472nm, light intensity 0.17 0.22mW / mm 2 CO2 concentration 4.8% 5.2%.
[0056] Pre-screening of parameter range based on random forest algorithm: Train random forest model using historical data, and pre-shrink parameter range according to feature importance (such as Gini coefficient) to improve search efficiency of subsequent reinforcement learning.
[0057] Dynamic optimization based on deep reinforcement learning (DRL): Within the DQN framework, the agent adjusts its actions based on environmental state output parameters, and the environment provides comprehensive reward signals (including photoresponse rate, lactate concentration, and system stability). By setting an experience replay mechanism and learning rate, the agent gradually converges to the optimal parameter combination with stable rewards during training (e.g., light wavelength 470nm, light intensity 0.20mW / mm²). 2 CO2 concentration 5.0%.
[0058] Step S4: Receive and execute the chemical reagent ratio scheme generated in step S3 through the reagent precise ratio module to complete the automatic weighing, dilution, pH adjustment and timed quantitative addition of the reagent.
[0059] The precise drug formulation module weighs the drug through an automatic weighing module, dilutes and adjusts the pH through a precise dilution and mixing submodule, and adds the drug according to the set time and dosage through a quantitative addition and timing submodule.
[0060] The precise drug mixing module includes one or more of the following: a high-precision electronic balance, a liquid flow meter, an automatic acid / alkali dispensing device, a magnetic stirrer, a pH sensor, a pipette, and a peristaltic pump.
[0061] Step S5: Receive and execute the physical environment control strategy generated in step S3 through the closed-loop photoelectric multimodal control module, monitor and dynamically adjust the illumination parameters in real time, and maintain the stability of temperature, humidity and CO2 concentration.
[0062] The closed-loop photoelectric multimodal control module collects real-time illumination data through the illumination parameter monitoring submodule, shields harmful wavelengths and precisely controls the light source through the illumination dynamic adjustment submodule, and maintains the stability of temperature, humidity and CO2 concentration through the comprehensive environmental stabilization submodule.
[0063] The closed-loop photoelectric multimodal control module includes a light sensor, an LED light source controller, a light-shielding device, and temperature, humidity, and CO2 concentration sensors.
[0064] Light-shielding device: The observation window of the incubator is equipped with an openable and closable metal light-shielding structure. The LED light source outlet is equipped with an adjustable aperture mechanism and a 470nm narrow-band filter element. Each independent well of the multi-well culture plate is equipped with a dedicated light-shielding cover.
[0065] Step S6: Detect the cell culture effect after processing in steps S4 and S5 through the result verification and feedback module, and obtain the detection data.
[0066] Step S7: Feed the detection data obtained in step S6 back to the multi-cell type parameter database and intelligent decision-making unit as new historical experimental data to optimize the model parameters of the artificial neural network module and the multi-algorithm optimization module, forming a closed-loop iterative optimization.
[0067] Step S71: Classify and store the detection data obtained in step S6; Step S72: Compare and analyze the detection data with historical data to identify state deviations during the cultivation process; Step S73: Update the model parameters of the artificial neural network module and the multi-algorithm optimization module using the detection data; Step S74: Based on the updated model parameters, adjust the optimization strategy of the intelligent decision-making unit; Step S75: Apply the optimized strategy to the subsequent cell culture process to achieve closed-loop iterative optimization.
[0068] The invention will be further illustrated below with specific examples.
[0069] S101, Information Entry and Data Retrieval.
[0070] HeLa cells (cell number) were entered. HeLa 001) and its cultivation objectives, namely, 48-hour survival rate ≥90% and adhesion rate ≥85%; The database was used to retrieve 30 sets of relevant historical data to obtain the optimal parameter range, including a pancreatic enzyme concentration of 0.2%. 0.22%, FBS concentration 10%, incubation temperature 37°C 0.1℃.
[0071] S102, Parameter prediction and optimization scheme generation.
[0072] The basic parameter range, including a pancreatic enzyme concentration of 0.20%, was obtained based on predictions using artificial neural networks. 0.22%, incubation temperature 36.8°C 37.2℃; Through multi-algorithm collaborative optimization, the final execution plan is output, including trypsin concentration of 0.21%, FBS concentration of 10.5%, culture temperature of 37.0℃, and light conditions of 450nm / 40lux.
[0073] S103, Reagent formulation and environmental control.
[0074] The drug preparation module automatically executes the following steps according to the plan: weigh pancreatic enzyme (final concentration 0.21%) using an electronic balance, add FBS (final concentration 10.5%) quantitatively using a peristaltic pump, and mix thoroughly with magnetic stirring for 5 minutes. The environmental control module synchronously adjusts the temperature and CO2 concentration to 37.0℃ and 5.0% by closing the light shield and loading a 450nm filter. It will automatically alarm when the temperature exceeds the limit.
[0075] S104. Effect detection and model iterative optimization.
[0076] Cell status was assessed after 24 hours, with a viability of 89% and an adhesion rate of 84%. The reason for this was attributed to the low culture temperature (36.9℃). After adjusting the culture temperature to 37.0℃, the viability increased to 91% and the adhesion rate increased to 86% after retesting. The experimental data was stored in the database, the model was retrained and the parameter weights were optimized (e.g., the weight of the effect of temperature on the adhesion rate was increased from 0.35 to 0.4), and the algorithm parameters were updated to improve subsequent decision-making.
[0077] Example 2 A brain-computer interface-based cell culture optimization system includes: The input and data unit includes an input module for receiving cell types and culture objectives, and a multi-cell type parameter database for storing basic culture parameters and historical experimental data; The intelligent decision-making unit includes an artificial neural network module connected to the input and data unit and a multi-algorithm optimization module. The artificial neural network module is used to predict the range of basic parameters based on the input data, and the multi-algorithm optimization module is used to generate chemical reagent ratio schemes and physical environment control strategies with the range of basic parameters as the boundary. The execution unit includes a precise drug ratio module and a closed-loop photo-electric multimodal control module connected to the intelligent decision-making unit, which are used to execute the chemical drug ratio scheme and the physical environment control strategy, respectively. The result verification and feedback unit, connected to the execution unit and the input and data unit, is used to detect the cell culture effect and feed the detection data back to the multi-cell type parameter database and the intelligent decision-making unit to form a closed-loop optimization.
[0078] The results verification and feedback unit includes: The basic state detection submodule is used to detect cell viability, adhesion rate, and purity; the functional integrity detection submodule is used to detect cell action potentials and light stimulation response. The deviation analysis and parameter adjustment submodule is used to automatically analyze the causes of cell state deviations and generate real-time adjustment instructions; The data feedback and model iteration submodule is used to feed back the monitored and adjusted new data to the multi-cell type parameter database, and to retrain the artificial neural network module and optimize the multi-algorithm optimization module using the new data.
[0079] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.
[0080] Therefore, the present invention adopts the above-mentioned brain-computer interface-based cell culture optimization method and system, which can overcome the excessive reliance on human experience in the prior art. Through the deep integration of intelligent decision-making and automated execution, it achieves precise and stable control of the chemical microenvironment and physical light environment during cell culture, thereby improving the quality consistency and process efficiency of cell culture, and providing reliable technical support for brain-computer interface research and other high-precision cell experiments.
[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.
Claims
1. An optimization method for cell culture based on brain-computer interface, characterized in that, Includes the following steps: Step S1: Input cell type and culture objective, and call up basic culture parameters and historical experimental data from the multi-cell type parameter database; Step S2: Based on the cell type, culture objective, and historical experimental data in Step S1, feature learning is performed through an artificial neural network module to output the range of basic culture parameters for the cell type. Step S3: Using the range of basic cultivation parameters output in step S2 as the optimization boundary, the parameters are processed through a multi-algorithm optimization module to generate an executable chemical reagent ratio scheme and physical environment control strategy. Step S4: Receive and execute the chemical reagent ratio scheme generated in step S3 through the reagent precise ratio module to complete the automatic weighing, dilution, pH adjustment and timed quantitative addition of the reagent; Step S5: Receive and execute the physical environment control strategy generated in step S3 through the closed-loop photoelectric multimodal control module, monitor and dynamically adjust the illumination parameters in real time, and maintain the stability of temperature, humidity and CO2 concentration. Step S6: Detect the cell culture effect after processing in steps S4 and S5 through the result verification and feedback module, and obtain the detection data; Step S7: Feed the detection data obtained in step S6 back to the multi-cell type parameter database and intelligent decision-making unit as new historical experimental data to optimize the model parameters of the artificial neural network module and the multi-algorithm optimization module, forming a closed-loop iterative optimization.
2. The method for optimizing cell culture based on brain-computer interface according to claim 1, characterized in that, In step S2, the artificial neural network module uses NIL. PLS The TDC hybrid architecture performs feature learning on the input data and outputs a range of basic cell culture parameters; NIL PLS The TDC hybrid architecture includes a nonlinear feature interaction learning module, a partial least squares feature dimensionality reduction module, and a time-series dynamic correction module, which are used to process the input cell type, culture target, and historical experimental data to obtain the basic parameter range used to guide subsequent optimization.
3. The method for optimizing cell culture based on brain-computer interface according to claim 1, characterized in that, In step S3, the multi-algorithm optimization module uses NSGA. II and TOPSIS algorithms are used to jointly optimize chemical reagent parameters, and deep reinforcement learning and random forest algorithms are used to jointly optimize physical environment parameters.
4. The method for optimizing cell culture based on brain-computer interface according to claim 1, characterized in that, In step S4, the precise drug ratio module weighs the drug through the automatic weighing module, dilutes and adjusts the pH through the precise dilution and mixing submodule, and adds the drug according to the set time and dosage through the quantitative addition and timing submodule.
5. The method for optimizing cell culture based on brain-computer interface according to claim 4, characterized in that, The precise drug mixing module includes one or more of the following: a high-precision electronic balance, a liquid flow meter, an automatic acid / alkali dispensing device, a magnetic stirrer, a pH sensor, a pipette, and a peristaltic pump.
6. The method for optimizing cell culture based on brain-computer interface according to claim 1, characterized in that, In step S5, the closed-loop photoelectric multimodal control module collects real-time illumination data through the illumination parameter monitoring submodule, shields harmful wavelengths and controls the light source through the illumination dynamic adjustment submodule, and maintains the stability of temperature, humidity and CO2 concentration through the comprehensive environmental stabilization submodule.
7. The method for optimizing cell culture based on brain-computer interface according to claim 6, characterized in that, The closed-loop photoelectric multimodal control module includes a light sensor, an LED light source controller, a light-shielding device, and temperature, humidity, and CO2 concentration sensors.
8. The method for optimizing cell culture based on brain-computer interface according to claim 1, characterized in that, Step S7 includes the following steps: Step S71: Classify and store the detection data obtained in step S6; Step S72: Compare and analyze the detection data with historical data to identify state deviations during the cultivation process; Step S73: Update the model parameters of the artificial neural network module and the multi-algorithm optimization module using the detection data; Step S74: Based on the updated model parameters, adjust the optimization strategy of the intelligent decision-making unit; Step S75: Apply the optimized strategy to the subsequent cell culture process to achieve closed-loop iterative optimization.
9. A brain-computer interface-based cell culture optimization system, characterized in that, include: The input and data unit includes an input module for receiving cell types and culture objectives, and a multi-cell type parameter database for storing basic culture parameters and historical experimental data; The intelligent decision-making unit includes an artificial neural network module connected to the input and data unit and a multi-algorithm optimization module. The artificial neural network module is used to predict the range of basic parameters based on the input data, and the multi-algorithm optimization module is used to generate chemical reagent ratio schemes and physical environment control strategies with the range of basic parameters as the boundary. The execution unit includes a precise drug ratio module and a closed-loop photo-electric multimodal control module connected to the intelligent decision-making unit, which are used to execute the chemical drug ratio scheme and the physical environment control strategy, respectively. The result verification and feedback unit, connected to the execution unit and the input and data unit, is used to detect the cell culture effect and feed the detection data back to the multi-cell type parameter database and the intelligent decision-making unit to form a closed-loop optimization.
10. The brain-computer interface-based cell culture optimization system according to claim 9, characterized in that, The results verification and feedback unit includes: The basic state detection submodule is used to detect cell viability, adhesion rate, and purity; the functional integrity detection submodule is used to detect cell action potentials and light stimulation response. The deviation analysis and parameter adjustment submodule is used to automatically analyze the causes of cell state deviations and generate real-time adjustment instructions; The data feedback and model iteration submodule is used to feed back the monitored and adjusted new data to the multi-cell type parameter database, and to retrain the artificial neural network module and optimize the multi-algorithm optimization module using the new data.