Organ chip microflow control method and device based on hydrogel

Abstract

The invention relates to the technical field of organ chips, in particular to an organ chip microflow control method and device based on hydrogel, and the method comprises multi-mode stimulation cooperative regulation and control: at least two external stimulation signals are applied to a hydrogel microflow channel, and the stimulation signals are selected from temperature, light, enzyme concentration or pH change; the flow channel permeability, the shearing force or the solute diffusion rate are dynamically adjusted by changing the physical characteristics of the hydrogel; multi-parameter fusion feedback: synchronously detecting fluid dynamic parameters and cell physiological parameters in a flow channel, and generating multi-dimensional microenvironment state data; and closed-loop control: based on the multi-dimensional microenvironment state data, adjusting the intensity, time sequence or spatial distribution of stimulation signals through a dynamic algorithm to realize cross-organ chip metabolism and fluid mechanics cooperative regulation and control. The device comprises a multi-modal stimulation cooperative regulation and control module, a multi-parameter fusion feedback module and a closed-loop control module. The method has the effect of improving the simulation effect.

Description

Technical Field

[0001] The present application relates to the technical field of organ-on-a-chip, and particularly to a microfluidic control method and device for an organ-on-a-chip based on hydrogel. Background Art

[0002] Currently, an organ-on-a-chip is an in vitro research platform that simulates the physiological functions of human organs through microfluidic technology and has important value in drug development, toxicity testing, and disease modeling. Traditional organ-on-a-chip mostly uses inert materials such as polydimethylsiloxane (PDMS) to construct microfluidic channels, but PDMS lacks biological activity and is difficult to support the bionic function of the extracellular matrix (ECM). In recent years, hydrogel has become the core material of organ-on-a-chip due to its physicochemical properties similar to those of natural ECM (such as adjustable stiffness, permeability, and biocompatibility).

[0003] In the prior art, by utilizing the phase change characteristics of poly(N-isopropylacrylamide) (PNIPAM), the cross-sectional area of the flow channel is adjusted through an external temperature control device to control the fluid shear force. In a specific implementation, the temperature control device consists of discrete Peltier elements, and the flow channel width is changed through heating / cooling cycles (the cross-sectional area of the flow channel is 100 μm² at 25 °C and shrinks to 70 μm² at 37 °C), or a fixed flow channel design is adopted, relying on an external pump to drive the fluid circulation.

[0004] In the above technologies, relying on a single stimulus (such as temperature) with fixed parameters, it is impossible to dynamically adapt to pathophysiological changes (such as vascular leakage caused by inflammation); only single fluid parameters such as flow rate or pressure are monitored, ignoring the real-time feedback of cell metabolic states (such as oxygen partial pressure, lactic acid concentration); and it is impossible to simulate the dynamic metabolic coupling between organs and cannot dynamically simulate the changes of organs in response to various stimuli, resulting in poor simulation effects. Summary of the Invention

[0005] In order to improve the simulation effect, the present application provides a microfluidic control method and device for an organ-on-a-chip based on hydrogel.

[0006] In a first aspect, a microfluidic control method and device for an organ-on-a-chip based on hydrogel provided by the present application adopt the following technical solutions: A microfluidic control method for an organ-on-a-chip based on hydrogel includes the following steps: Multi-modal stimulus collaborative regulation: Applying at least two external stimulus signals to the hydrogel microfluidic channel, where the stimulus signals are selected from temperature, light, enzyme concentration, or pH change, and dynamically adjusting the flow channel permeability, shear force, or solute diffusion rate by changing the physical properties of the hydrogel; Multi-parameter fusion feedback: Synchronously detecting the hydrodynamic parameters and cell physiological parameters in the flow channel to generate multi-dimensional microenvironment state data; Closed-loop control: Based on the multi-dimensional microenvironment state data, the intensity, timing, or spatial distribution of the stimulation signal is adjusted through a dynamic algorithm to achieve coordinated regulation of metabolism and hydrodynamics across organ chips.

[0007] Traditional methods rely only on a single stimulus (such as temperature), with limited regulation dimensions and unable to simulate complex physiological scenarios; existing technologies usually only monitor flow rate or pressure, ignoring the feedback of cell metabolic status; organ chips operate independently (such as the Emulate liver chip), lacking cross-organ coordination logic, resulting in distorted pharmacokinetic simulations.

[0008] By adopting the above technical solution, temperature (25°C → 37°C) and light stimulation (405 nm laser, 50 mW / cm²) are applied to the liver chip, triggering the shrinkage of PNIPAM hydrogel (the cross-sectional area is reduced by 30%) and the increase in the local permeability of GelMA (the pore size is enlarged to 50 nm); enzyme (MMP-9, 0.5 μg / mL) and pH stimulation (pH 6.0) are applied to the kidney chip, and the permeability of the flow channel is enlarged to 5×10 -12 m² through enzymatic hydrolysis; a microelectrical impedance spectroscopy sensor (1 kHz - 100 kHz) measures the flow rate distribution (accuracy ±0.1 mm / s); a fluorescence probe (labeled with rhodamine B) detects the lactic acid concentration (resolution ±0.1 mM); a transendothelial electrical resistance (TEER) sensor evaluates the barrier integrity (frequency 1 MHz); a dynamic algorithm analyzes the flow rate-metabolism correlation data, adjusts the light stimulation intensity to 15 mW / cm², and reduces the enzyme injection rate to 0.3 μg / mL. The multi-stimulation coordination reduces the shear force control error and expands the permeability regulation range; through multi-dimensional data fusion, the differential fluid environments in the hepatic sinusoid region (shear force 1 - 5 dyn / cm²) and the central vein region (0.1 - 1 dyn / cm²) are simulated; cross-organ metabolic synchronization: the simulation error of the drug metabolism half-life between the liver and kidney is reduced from ±30% to ±5%, significantly improving the accuracy of drug toxicity testing; through multi-external stimulation simulation and simulating the dynamic metabolic coupling between organs, and being unable to dynamically simulate the changes of organs in response to various stimuli, the simulation effect is improved and it is more in line with the actual situation.

[0009] Improve the applicability of application scenarios: Drug liver toxicity testing: By dynamically adjusting the shear force of the liver chip (1 - 8 dyn / cm²) and the filtration rate of the kidney chip (permeability 1×10 -14 →5×10 -12 m²), the drug metabolism and excretion processes are accurately simulated; Inflammatory model construction: By synergistically degrading the hydrogel with enzyme + pH stimulation, the non-linear change of the vascular leakage rate is simulated (such as the time taken for the permeability to increase by 3 times is shortened from 20 minutes to 10 minutes).

[0010] Optionally, in the multi-modal stimulation coordinated regulation step, the application of the temperature stimulation adopts stepwise heating.

[0011] By adopting the above technical solutions, the prior art uses linear heating (1 °C / min), resulting in temperature overshoot (±5% error) and thermal stress damage to the hydrogel structure; the temperature control accuracy is improved: stepwise heating (initial 1 °C / min → subsequent 0.2 °C / min) reduces the cross-sectional area control error; reduces the fatigue of PNIPAM hydrogel caused by thermal stress, and the number of cycles is increased by stepwise heating, which can reduce cell thermal damage.

[0012] Optionally, in the closed-loop control step, the dynamic algorithm adopts a hierarchical control architecture. The bottom-layer PID controller adjusts the temperature / light stimulation intensity, and the top-layer reinforcement learning model optimizes the enzyme / pH stimulation strategy once per period; the reinforcement learning model is initialized by transfer learning, and the pre-trained dataset contains historical experimental data of several organ chips.

[0013] By adopting the above technical solutions, the traditional method uses a single PID or reinforcement learning, which cannot balance real-time response and long-term optimization; the new organ chip needs to be trained from scratch (time-consuming > 100 hours) and has a large data requirement; the control efficiency is improved: the PID quickly responds to shear force fluctuations (regulation time < 1 second), and the reinforcement learning optimizes cross-organ metabolic coupling (half-life error ±2%); transfer learning reduces the number of training iterations of the new organ chip.

[0014] Optionally, in the cross-organ collaborative regulation, there is a phase delay in the application of the stimulation signal between the upper chip and the lower chip. The temperature stimulation of the upper chip is triggered with a first phase delay time seconds, and the light stimulation of the lower chip is triggered with a second phase delay time seconds.

[0015] By adopting the above technical solutions, the liver chip (upper layer) and the kidney chip (lower layer) are connected in series through a microfluidic channel, and the physical distance is 10 mm; the stimulation timing is set: when the liver chip applies temperature stimulation (such as rising from 25 °C to 37 °C), the system sets the first phase delay time to 5 seconds, that is, the temperature stimulation of the liver chip is triggered 5 seconds in advance; the light stimulation of the kidney chip (such as 405 nm laser irradiation) is set with a second phase delay time of 3 seconds, that is, the light stimulation of the kidney chip starts 3 seconds after the liver stimulation is triggered.

[0016] Dynamic regulation logic: In the drug metabolism experiment, the metabolites in the liver-on-a-chip (such as drug intermediates) are transported through the flow channel to the kidney-on-a-chip with a delay of 5 seconds and 3 seconds, ensuring that after the metabolites are processed by the liver, the kidney-on-a-chip has adjusted the permeability through light stimulation to match the filtration demand; by delaying the triggering of the light stimulation of the kidney-on-a-chip, the time difference of real liver-kidney metabolism (physiological delay is about 3 - 5 seconds) is simulated, reducing the simulation error of the drug metabolism half-life from ±30% of the traditional fixed delay to ±10%; if the kidney-on-a-chip starts high-permeability filtration prematurely (without delay), it may lead to the accumulation of unmetabolized toxic substances, and the delay mechanism increases the cell survival rate from 70% to 85%. By setting the delay, it is more in line with the actual situation, improving the authenticity of the simulation and the simulation effect.

[0017] Optionally, the first phase delay time , is dynamically adjusted according to the hydrogel fatigue coefficient α and the metabolite concentration change rate β, and the calculation model is as follows: ; where is the physical transmission distance between the upper chip and the lower chip; represents the average diffusion rate of the drug or metabolite in the chip, represents the drug concentration gradient between the upper chip and the lower chip, is the reference concentration, is the time constant; is the number of hydrogel deformation cycles, the fatigue coefficient , β is the metabolite concentration change rate, which is detected in real time by a fluorescence probe; is the drug concentration feedback gain coefficient, ; the calculation model of the second phase delay time is as follows: , where is a set value.

[0018] By adopting the above technical solutions, the existing organ-on-a-chip uses a fixed delay (such as a 5-second liver-kidney delay), which cannot adapt to the changes in pathological conditions; this solution optimizes the metabolic synchronization, and the liver-kidney delay time is dynamically adjusted according to the metabolic load (such as shortening to 3 seconds when β is high), improving the metabolite clearance efficiency; material protection mechanism: when the hydrogel is fatigued ( ), the delay time is extended by 20%, reducing the material deformation caused by high-frequency stimulation and reducing structural damage; it can be used for the simulation of the kidney-on-a-chip. When the filtration rate of the kidney-on-a-chip decreases ( gradient increases), the delay of the liver-on-a-chip is automatically extended to 8 seconds, simulating the enhanced compensatory liver metabolism and accelerating the clearance of toxic metabolites, and the cell survival rate is increased.

[0019] Optionally, where Dynamic value acquisition, and the calculation model is as follows: .

[0020] By adopting the above technical solution, at a high metabolic rate (such as in acute toxicity testing), the renal chip delay is shortened to 4.64 seconds, and the metabolite clearance rate is increased by 40%; Structural protection mechanism: When , automatically decreases, the light stimulation interval is extended, and the hydrogel fatigue fracture is prevented.

[0021] Optionally, the regulation results of the first phase delay time and the second phase delay time are input into the organ digital twin model, and the porous medium flow equation and solute diffusion equation are solved based on the finite element method to predict the metabolic trajectory after delay adjustment in real time. If the prediction error is greater than the set error, the control parameter reset is triggered.

[0022] By adopting the above technical solution, the prediction deviation of the metabolic trajectory is compressed from ±20% of the offline model to ±8%, avoiding control failure caused by model mismatch; Fast recovery ability: Parameter reset enables the system to recover to a steady state within 30 seconds, reducing the experimental interruption time.

[0023] Optionally, the digital twin model integrates a microchannel deformation prediction module, and the geometric changes of the microchannel are calculated in real time based on the viscoelastic constitutive equation of the hydrogel. If the predicted deformation exceeds the safety threshold, the global stimulation intensity is triggered to decrease.

[0024] By adopting the above technical solution, based on the Kelvin-Voigt model (elastic modulus E = 5 kPa, viscosity coefficient), the deformation of the microchannel cross-sectional area is predicted; Safety threshold judgment: If the predicted deformation > 35%, the global stimulation intensity is triggered to decrease; The temperature stimulation is reduced from 37 °C to 34 °C, and the light intensity is reduced from 50 mW / cm² to 25 mW / cm²; The prediction deviation of the metabolic trajectory is compressed from ±20% of the offline model to ±8%, avoiding control failure caused by model mismatch; Parameter reset enables the system to recover to a steady state within 30 seconds, reducing the experimental interruption time.

[0025] Deformation risk warning: The collapse risk (such as the cross-sectional area shrinking to the critical value) is predicted 5 minutes in advance, and the channel recovery rate after intervention is increased from 40% to 90%; Cross-organ protection: The reduction of global stimulation avoids chain organ damage (such as the cell necrosis rate of the renal chip due to microchannel blockage is reduced from 25% to 5%).

[0026] Optionally, in the viscoelastic constitutive equation of the digital twin model, a fatigue-metabolism coupling correction factor is introduced, and the calculation model is as follows: ; Among them, is the metabolic rate reference value, is the fatigue coefficient, is the real-time metabolite concentration change rate.

[0027] By adopting the above technical solution, through the introduction of the correction factor, the elastic modulus is adjusted, so that in the high-load experiment, the deformation prediction accuracy is improved; the service life is significantly extended: through the dynamic compensation of the γ factor, the chip still maintains 90% performance after 2000 cycles and extends the service life.

[0028] In the second aspect, a device provided by the present application adopts the following technical solution: A device includes the following modules: Multimodal stimulus co-regulation module: used to apply external stimuli and dynamically adjust the channel permeability, shear force or solute diffusion rate by changing the physical properties of the hydrogel; Multi-parameter fusion feedback module: The input end is connected to the output end of the multimodal stimulus co-regulation module, and is used to synchronously detect the hydrodynamic parameters and cell physiological parameters in the channel to generate multi-dimensional microenvironment state data; Closed-loop control module: The input end is connected to the output end of the multi-parameter fusion feedback module, and is used to adjust the intensity, timing or spatial distribution of the stimulus signal according to the multi-dimensional microenvironment state data to achieve the coordinated regulation of metabolism and hydrodynamics of the cross-organ chip.

[0029] By adopting the above technical solution.

[0030] In summary, the present application includes at least one of the following beneficial technical effects: 1. Through multi-external stimulus simulation and dynamic metabolic coupling between organs, and being unable to dynamically simulate the changes of organs in response to various stimuli, the simulation effect is improved and it is more in line with the actual situation; 2. By setting the delay, it is more in line with the actual situation, improving the authenticity of the simulation and the simulation effect. Specific embodiments

[0031] The following further details the present application.

[0032] This embodiment discloses a microfluidic control method for an organ chip based on hydrogel.

[0033] Embodiment 1: A microfluidic control method for an organ chip based on hydrogel, including the following steps: Multimodal stimulus co-regulation: Applying at least two external stimulus signals to the hydrogel microchannel, and the stimulus signals are selected from temperature, light, enzyme concentration or pH change, and dynamically adjusting the channel permeability, shear force or solute diffusion rate by changing the physical properties of the hydrogel; Specifically, stimulus signal selection and application: Temperature stimulus: A micro-thermocouple array (with a spacing of 200 μm) is used in combination with a PID temperature control module to locally heat / cool the thermosensitive hydrogel (PNIPAM) flow channel. The temperature range is 25 - 40 °C, with a resolution of ±0.1 °C. The application of the temperature stimulus adopts a stepped temperature increase.

[0034] Initial temperature increase stage: The temperature is increased from 25 °C to 35 °C at a rate of 1 °C / min, triggering the phase change and shrinkage of the PNIPAM hydrogel (the cross-sectional area shrinks from 100 μm² to 80 μm²); Slow speed maintenance stage: It is increased to the target temperature of 37 °C at a rate of 0.2 °C / min, and the cross-sectional area of the flow channel is kept stable at 70 μm² (error ±1.5%); reducing the structural stress damage of the hydrogel caused by sudden temperature increase, and the expression of cellular heat shock protein is reduced by 50%.

[0035] Light stimulus: A digital micromirror device (DMD) is used to generate dynamic light spots (wavelength 405 nm / 650 nm, power 10 - 50 mW / cm²) to irradiate the photosensitive hydrogel (GelMA) area; Shrinkage regulation: The 405 nm light spot (15 mW / cm²) locally irradiates the photosensitive GelMA area, the pore size shrinks to 30 nm, and the permeability drops to 1×10 -14 m²; Expansion regulation: The 650 nm light spot (20 mW / cm²) expands the pore size to 50 nm, and the permeability rises to 3×10 -12 m²; Technical effect: Light intensity gradient control realizes the shear force distribution (0.1 - 10 dyn / cm²), simulating the turbulent effect of blood vessel branching.

[0036] Enzyme / pH stimulus: Matrix metalloproteinase (MMP-9, concentration 0.1 - 1 μg / mL) and acidic buffer solution (pH 5.0 - 6.5) are injected through a microfluidic valve to adjust the cross-linking density of the hydrogel. Enzyme / pH stimulus: Enzymatic degradation regulation: Inject MMP-9 (0.5 μg / mL) to degrade the cross-linking points of the hydrogel, and the permeability increases by 5 times (1×10 -14 →5×10 -12 m²) within 10 minutes; pH synergy: The pH 5.5 buffer solution enhances the enzyme activity, and the rate of decrease in cross-linking density is increased by 2 times; Technical effect: Simulating vascular leakage under inflammatory conditions (non-linear change in permeability, R² > 0.95); Dynamic response mechanism: Temperature response: PNIPAM undergoes phase transition and shrinkage at 37°C, with the cross-sectional area of the flow channel reduced by 30% (e.g., from 100 μm² to 70 μm²), and the shear force increasing from 1 dyn / cm² to 3 dyn / cm². Light response: 405 nm light irradiation triggers the contraction of GelMA (pore size reduced to 20 nm), and 650 nm light irradiation triggers the expansion (pore size enlarged to 50 nm), with the permeability changing in the range of ±5%. Enzyme / pH response: MMP-9 enzymatically degrades the hydrogel cross-linking points, combined with a decrease in pH (pH 6.0 → 5.5), and the permeability increases by 5 times (1×10 -14 →5×10 -12 m²).

[0037] Multi-parameter fusion feedback: Synchronously detect the hydrodynamic parameters and cell physiological parameters in the flow channel to generate multi-dimensional microenvironment state data; Specifically, hydrodynamic parameter detection: Micro-particle image velocimetry (μPIV): Fluorescent nanoparticles (200 nm, excitation wavelength 488 nm) are injected into the flow channel, and a high-speed CMOS camera (frame rate 1000 fps) captures the particle trajectories to calculate the flow velocity distribution (accuracy ±0.1 mm / s) and shear force gradient.

[0038] Micro-impedance spectroscopy sensing: Interdigitated electrodes (spacing 50 μm) measure the impedance changes in the frequency band of 1 kHz - 1 MHz to invert the local shear force in the flow channel (accuracy ±0.05 dyn / cm²).

[0039] Cell physiological parameter detection: Trans-endothelial electrical resistance (TEER): 1 MHz high-frequency impedance is used to detect the integrity of the endothelial barrier, and a 10% decrease in impedance triggers an alarm.

[0040] Fluorescent metabolic probe: Glucose labeled with rhodamine B (excitation / emission wavelength 560 nm / 585 nm) is used to detect the metabolic rate (resolution ±0.1 mM, sampling rate 10 Hz).

[0041] Data fusion and synchronization: Hardware: The FPGA chip synchronizes multi-sensor data in real time (time deviation < 1 ms); Software: Construct a multi-dimensional state matrix (flow velocity, shear force, metabolite concentration, TEER), and eliminate noise through Kalman filtering.

[0042] Closed-loop control: Based on the multi-dimensional microenvironment state data, adjust the intensity, timing, or spatial distribution of the stimulation signal through a dynamic algorithm to achieve coordinated regulation of metabolism and hydrodynamics across the organ chip.

[0043] Specifically, hierarchical control architecture: Bottom layer PID control: Temperature regulation: Adjust the thermocouple power at a frequency of 10 Hz (±0.1 °C), and respond to shear force fluctuations (e.g., from 1.5 → 2.0 dyn / cm², regulation time < 1 second). Light intensity regulation: Dynamically match the target permeability, and the light intensity adjustment step is ±2 mW / cm².

[0044] Top layer reinforcement learning optimization: Policy generation: Based on indicators such as metabolite clearance rate, cell survival rate, and barrier integrity every 5 minutes, optimize enzyme / pH stimulation parameters (e.g., MMP-9 concentration from 0.5 → 0.3 μg / mL).

[0045] Transfer learning initialization: The pre-trained dataset contains 100 sets of historical experimental data of liver chips (shear force, metabolic rate, number of deformation cycles), and the number of training iterations is reduced by 70% after being transferred to the kidney chip.

[0046] Cross-organ collaborative logic: Dynamic phase delay: The temperature stimulation of the liver chip is triggered 5 seconds in advance, that is, the temperature stimulation is started 5 seconds before the drug injection is completed; the light stimulation of the kidney chip is delayed by 4.5 seconds, that is, the light stimulation is applied 4.5 seconds after the temperature stimulation is applied, ensuring that the metabolite transport and filtration rates match.

[0047] Digital twin verification: Predict the metabolic trajectory based on the Brinkman equation and Fick's law, and reset the control parameters if the error > 10%; Cross-organ collaboration: Ensure that when the metabolites of the liver chip reach the kidney chip, the permeability of the kidney flow channel has been adjusted to the target value (e.g., 3×10 -12 m²).

[0048] For example, chip preparation and initialization: Materials: Upper layer liver chip: PNIPAM hydrogel (LCST = 32 °C), flow channel cross-sectional area 100 μm², and primary hepatocytes are planted on the inner wall; Lower layer kidney chip: GelMA hydrogel (photo-crosslinked), and the flow channel is modified with renal tubular endothelial cells.

[0049] Sensor integration: μPIV imaging window, TEER electrode array, fluorescence probe microfluidic injection channel.

[0050] Multi-modal stimulation application and feedback: Liver chip regulation: Gradually heat up to 37 °C, the flow channel cross-sectional area is reduced to 70 μm², and the shear force is increased to 3 dyn / cm²; Local irradiation with a 405 nm laser (15 mW / cm²), and the permeability is reduced to 1×10 -14 m².

[0051] Kidney chip regulation: Inject MMP-9 (0.5 μg / mL, pH 6.0), and the permeability is increased to 3×10 -12 m²; Expand the filtration pore size to 50 nm with a 650 nm laser (20 mW / cm²).

[0052] Data acquisition: The flow velocity detected by μPIV is 1.2 ± 0.1 mm / s, the TEER impedance drops by 15%, and the measured lactic acid concentration is 5.2 mM.

[0053] Closed-loop control and optimization: PID real-time adjustment: The temperature drops to 34 °C, the cross-sectional area recovers to 85 μm², and the shear stress stabilizes at 1.5 dyn / cm²; the light intensity of the kidney chip is increased to 35 mW / cm², and the permeability rises to 5×10 -12 m². Optimization by reinforcement learning: Generate an MMP-9 injection strategy (0.3 μg / mL) based on historical data, and adjust the pH to 5.8.

[0054] Metabolic synchronization: The simulated drug half-life is 12 hours, and the measured value is 12.3 hours (error ±2.5%); Cell viability: The viability of hepatocytes is 92% (68% by traditional method), and the viability of renal cells is 89%; System stability: After 1000 cycles of testing, the attenuation rate of the flow channel cross-sectional area <5% (traditional chip >30%).

[0055] Through dynamic regulation, the control error of shear stress is reduced, the regulation range of permeability is expanded, combined with cross-organ coordination, the error of metabolic half-life is reduced, and the cell viability is improved. By adopting digital twin, the material life is extended.

[0056] Example 2: In cross-organ coordinated regulation, there is a phase delay in the application of stimulation signals between the upper chip and the lower chip. The temperature stimulation of the upper chip is triggered with a first phase delay time seconds, and the light stimulation of the lower chip is triggered with a second phase delay time seconds; The first phase delay time is dynamically adjusted according to the hydrogel fatigue coefficient α and the metabolite concentration change rate β. The calculation model is as follows: ; where is the physical transmission distance between the upper chip and the lower chip; represents the average diffusion rate of the drug or metabolite in the chip, represents the drug concentration gradient between the upper chip and the lower chip, is the reference concentration (1 μg / mL), is the time constant (such as 60 s); is the number of hydrogel deformation cycles, and the fatigue coefficient , β is the metabolite concentration change rate, which is detected in real time by a fluorescent probe; is the drug concentration feedback gain coefficient, ; The calculation model of the second phase delay time is as follows: , where is the set value.

[0057] Specifically, chip layout and distance measurement: Measure the physical transmission distance of the flow channel between the upper layer (liver chip) and the lower layer (kidney chip) , for example, set ; Calibration: Verify the geometric parameters of the flow channel through the microfluidic chip design drawing to ensure that the spacing error < 1%.

[0058] Metabolic rate calibration: Equipment: Fluorescent tracer (such as rhodamine B-labeled drug) and high-speed imaging system. Inject the fluorescently labeled drug (concentration 1 μM), and track the diffusion process of the drug from the liver chip to the kidney chip through microscopic imaging (frame rate 1000 fps).

[0059] Calculate the average diffusion rate , for example, if the measured transmission time is 6.67 seconds, then ; Sensor calibration: Fluorescent probe: Calibrate the detection range (0 - 10 mM / s) and resolution (±0.1 mM) of the metabolite concentration change rate (β). Pressure sensor: Record the number of deformation cycles of the hydrogel , with an accuracy of ±1 time.

[0060] Real-time parameter acquisition: Metabolite concentration change rate (β): The fluorescent probe measures the lactate concentration every 10 seconds and calculates the change rate (for example, β = 0.1 mM / s).

[0061] Hydrogel fatigue degree ( ): The cumulative number of deformations of the pressure sensor (for example ), and the fatigue degree coefficient α = 0.002.

[0062] Drug concentration gradient ( ): The microelectrode measures the concentration difference between the outlet of the liver chip and the inlet of the kidney chip (for example ).

[0063] Calculate the first phase delay time , ; Set ; Then .

[0064] Upper layer liver chip: Trigger the temperature stimulation (25 °C → 37 °C) 7.377 seconds before drug injection, reduce the cross-sectional area of the flow channel to 70 μm², and increase the shear force to 3 dyn / cm².

[0065] Lower-layer kidney chip: Trigger the light stimulation (405 nm, 15 mW / cm²) with a 5.377-second delay after the temperature stimulation is applied, and adjust the permeability to 3×10 -12 m².

[0066] Example of dynamic adjustment: Scenario 1 (high metabolic load): If β rises to 4 mM / s, recalculate , , with limited impact on metabolism.

[0067] Scenario 2 (material fatigue): If , , , extend the stimulation interval to protect the hydrogel.

[0068] Verification and feedback: Verification of metabolic synchronization: Equipment: A liquid chromatography - mass spectrometry system (LC - MS) to detect the metabolites in the liver chip (such as drug intermediates) and the filtrate of the kidney chip. Index: Half - life error (target value 12 hours, measured value 12.3 hours, error ±2.5%). Monitoring of cell physiological state: Detection of viability: Calcein - AM / PI double - staining method, hepatocyte viability 92% (control group 68%). Barrier function: TEER impedance value maintained > 200 Ω·cm² (damage threshold < 150 Ω·cm²).

[0069] Evaluation of material performance: Fatigue test: After 1000 cycles of cyclic loading, the cross - sectional area decay rate of the hydrogel < 5% (traditional material > 30%).

[0070] Specific implementation case: Drug metabolism simulation of liver - kidney chip Drug: Test compound X (a hepatotoxic drug in terms of metabolism, EC 50 = 10 μM).

[0071] Chip parameters: Liver chip: PNIPAM flow channel, initial cross - sectional area 100 μm², seeded with primary hepatocytes; Kidney chip: GelMA flow channel, initial permeability 1×10 -14 m², seeded with renal tubular epithelial cells.

[0072] Phase - delay regulation process: Parameter initialization: , , α = 0.003, k = 0.1.

[0073] Dynamic calculation and stimulation trigger: Liver chip: Heat up to 37 °C 7 seconds in advance, and the cross - sectional area shrinks to 70 μm²; Kidney chip: Start the light stimulation with a 5 - second delay, and the permeability rises to 3×10 -12 m².

[0074] Data acquisition: μPIV shows that the shear stress of the liver chip is 3.0 ± 0.2 dyn / cm²; The filtration rate of the kidney chip detected by the fluorescence probe is 1.2 μL / min.

[0075] Metabolic synchrony: The drug intermediate reaches the kidney chip 7 seconds after being processed by the liver chip and is filtered within 5 seconds (total delay of 12 seconds), simulating the physiological liver-kidney delay (10 - 15 seconds).

[0076] Toxicity control: The residual amount of unmetabolized drugs is reduced from 25% of the traditional method to 5%, and the cell necrosis rate is reduced from 30% to 8%; System stability: After 1000 cycles of testing, the flow channel function remains intact (cross-sectional area > 95 μm²).

[0077] The delay time is adjusted in real time through the fatigue degree (α) and the metabolic rate (β) to improve the accuracy of pathological simulation; The phase delay matches the liver-kidney metabolism time sequence, and the half-life error is compressed to ±2.5%; The fatigue compensation mechanism extends the chip life to 3 times that of the traditional method.

[0078] In other embodiments, wherein Dynamic value taking, the calculation model is as follows: .

[0079] Specifically, parameter acquisition: Real-time detection , , ; Calculation , if , take ; The above calculation is 0.635 > 0.5, directly adopt the calculated value.

[0080] By dynamically adjusting , reduce the possibility of excessive delay, reduce the possibility of too long delay time, optimize the response speed of the system, the basic transmission time, the fatigue attenuation factor, reflect the influence of material fatigue on the delay, remove the influence of β, solve the unit conflict, and focus on the coupling effect of fatigue and transmission time.

[0081] In other embodiments, the first phase delay time and the second phase delay time The regulatory results are input into the organ digital twin model. Based on the finite element method, the porous medium flow equation and the solute diffusion equation are solved to predict the metabolomic trajectory after delay adjustment in real time. If the prediction error is greater than the set error, the control parameter reset is triggered; the digital twin model integrates a microchannel deformation prediction module, and the geometric changes of the flow channel are calculated in real time based on the viscoelastic constitutive equation of the hydrogel. If the predicted deformation exceeds the safety threshold, the global stimulation intensity is triggered to decrease; in the viscoelastic constitutive equation of the digital twin model, a fatigue-metabolism coupling correction factor is introduced , and the calculation model is as follows: ; Among them, is the metabolic rate reference value, is the fatigue coefficient, is the real-time metabolite concentration change rate.

[0082] Specifically, a high-performance computing cluster is deployed, equipped with a GPU acceleration module for real-time finite element calculation; a biosensor array is integrated to collect organ metabolite concentration (β), fluid pressure, and microchannel deformation data in real time.

[0083] Software architecture: Build an organ digital twin model platform, integrating a multi-physics field solver (COMSOL / ANSYS), a viscoelastic constitutive equation solver module, and a control logic unit.

[0084] Digital twin model construction: Geometric modeling: Reconstruct the three-dimensional structure of the target organ through medical imaging data (CT / MRI), divide the microchannel network and the porous medium area, and generate unstructured tetrahedral meshes.

[0085] Physical field coupling: Porous medium flow equation: The Brinkman equation is used to describe fluid motion: , where is the fluid viscosity, is the permeability, is the pressure field.

[0086] Solute diffusion equation: Coupling convection-diffusion effects: ; among them, is the diffusion coefficient, is the metabolic source term.

[0087] Phase delay regulation and metabolomic trajectory prediction: Input parameters: Input the first phase delay time (single-channel regulation) and the second phase delay time (cross-channel collaborative regulation) into the model.

[0088] Finite element solution: Time discretization: The implicit Euler method is adopted, and the time step is adaptively adjusted (Δt = 1~10 ms).

[0089] Boundary conditions: According to the set pulsating pressure waveform at the inlet of the flow channel, and the outlet is set to free outflow.

[0090] Real-time solution: The flow velocity field 、concentration field and metabolite distribution .

[0091] Error feedback mechanism: Calculate the predicted metabolic trajectory and the measured value of the root mean square error (RMSE): ; If RMSE > the set threshold (such as 5%), trigger the reset of control parameters: go back to the parameters of the previous stable state and start the genetic algorithm optimization and .

[0092] Flow channel deformation prediction and stimulation intensity regulation: Viscoelastic constitutive equation: The generalized Maxwell model is used to describe the rheological properties of the hydrogel: ; where and are the relaxation modulus and viscosity, which are calibrated through dynamic mechanical analysis (DMA) experiments.

[0093] Real-time calculation of deformation: Update the displacement δ of the flow channel wall at each step. If the local deformation (safety threshold), it is judged as a risk of structural failure.

[0094] Trigger the attenuation of the stimulation intensity: The global stimulation voltage / current is dynamically reduced according to (k is the attenuation coefficient).

[0095] Integration of fatigue-metabolism coupling correction factor: Dynamic update of parameters: : Statistically count the number of pressure cycles in the microchannel (ΔP > 10 kPa is one cycle).

[0096] : Obtained from the benchmark value of the metabolic rate under static culture.

[0097] Real-time metabolic rate , which is obtained through sliding window filtering of sensor data.

[0098] Calculation of the correction factor γ: Update the value of γ every 5 minutes. If γ > 2.0, trigger an alarm and reduce the α weight coefficient (α = α × 0.9).

[0099] Embed γ into the constitutive equation and correct the relaxation modulus: .

[0100] Safety protection and data interaction: Dual redundant monitoring: Run two sets of solvers in parallel. When the result difference exceeds 3%, switch to the standby solver; API interface: Provide a RESTful API to interact with external medical devices, support the transmission of control parameters (such as , I) and alarm signals.

[0101] The embodiments of the present application also disclose a device, including the following modules: Multimodal stimulus collaborative regulation module: Used to apply external stimuli to dynamically adjust the flow channel permeability, shear force or solute diffusion rate by changing the physical properties of the hydrogel; Specifically, temperature control unit: Micro thermocouple array (spacing 200μm, made of nickel-chromium alloy): Embedded in the side wall of the hydrogel flow channel for local heating / cooling.

[0102] PID temperature control circuit (accuracy ±0.1°C): Integrated with MAX31855 chip, supports 10Hz feedback regulation, and matches the physiological temperature range (25 - 40°C).

[0103] Application scenario: Trigger the phase change of thermosensitive hydrogel (such as PNIPAM) to adjust the cross-sectional area of the flow channel (such as from 100μm² to 70μm²).

[0104] Light regulation unit: Digital micromirror device (DMD) (DLPLightCrafter4500): Generate dynamic light spots (wavelength 405nm / 650nm), light intensity range 1 - 50mW / cm² (adjustable).

[0105] Optical lens group (NA = 0.3): Focus the light spot to the target area (diameter 50μm) to reduce light scattering loss.

[0106] Application scenario: 405nm light shrinks the GelMA pore size (30nm → 20nm), 650nm light expands the pore size (20nm → 50nm).

[0107] Enzyme / pH infusion unit: Microfluidic injection pump (HarvardApparatusPHDUltra): Nanoscale flow control (0.1 - 10μL / min), integrated with a micro solenoid valve (response time <10ms).

[0108] Multi-channel mixer: Real-time mixing of MMP-9 (0.1 - 1 μg / mL) and buffer (pH 5.0 - 6.5) in the microchannel with a gradient accuracy of ±0.1 pH. Application scenario: Simulate the inflammatory microenvironment (e.g., MMP-9 degrades crosslinking points, and the permeability increases by 5 times). Multi-parameter fusion feedback module: The input end is connected to the output end of the multi-modal stimulation co-regulation module, used to synchronously detect the hydrodynamic parameters and cell physiological parameters in the flow channel, and generate multi-dimensional microenvironment state data; Specifically, functions: Real-time detection of hydrodynamic parameters (flow rate, shear force) and cell physiological parameters (metabolism, barrier function), generating a multi-dimensional data matrix.

[0109] Hydrodynamic sensor: Micro-particle image velocimetry system (μPIV): Fluorescent nanoparticles (200 nm, FluoSpheres™): Injected into the flow channel, excitation wavelength 488 nm (CNI laser).

[0110] High-speed CMOS camera (PhantomVEO410L): Frame rate 1000 fps, spatial resolution 1 μm / pixel, analyzing the flow velocity field through MATLAB PIVlab.

[0111] Micro-impedance spectrometer (Agilent 4294A): Interdigitated electrodes (spacing 50 μm, platinum material): Measuring impedance from 1 kHz to 1 MHz, and inverting local shear force (accuracy ±0.05 dyn / cm²).

[0112] Trans-endothelial electrical resistance (TEER) detection module: High-frequency impedance analyzer (ECIS Zθ): 1 MHz AC signal, detecting the integrity of the endothelial barrier (impedance drop > 10% triggers an alarm).

[0113] Fluorescent metabolism probe system: Glucose labeled with rhodamine B (Ex / Em = 560 / 585 nm): Fiber optic spectrometer (OceanInsight FLAME) detecting the metabolic rate (resolution ±0.1 mM).

[0114] Microfluidic probe injection channel: Integrated piezoelectric micropump (Bartels mp6), sampling rate 10 Hz.

[0115] Data synchronization unit: FPGA controller (Xilinx Zynq-7000): Real-time synchronization of multi-sensor data (time deviation < 1 ms), uploading to the host computer through the PCIe interface.

[0116] Closed-loop control module: The input end is connected to the output end of the multi-parameter fusion feedback module, and is used to adjust the intensity, timing or spatial distribution of the stimulation signal through a dynamic algorithm according to the multi-dimensional microenvironment state data, so as to realize the coordinated regulation of metabolism and hydrodynamics of the cross-organ chip.

[0117] Specifically, the stimulation parameters are dynamically adjusted based on the feedback data to achieve coordinated regulation of metabolism-hydrodynamics.

[0118] Underlying PID controller: Temperature regulation: STM32F4 microcontroller, updating the thermocouple power at 10Hz (PWM duty cycle ±5%).

[0119] Light intensity regulation: DMD driver board (DLPC3478), light intensity step ±2mW / cm², matching the target permeability (such as 1×10 -14 →3×10 -12 m²).

[0120] Top-level reinforcement learning unit: NVIDIA Jetson AGX Xavier: Runs the PyTorch reinforcement learning model, optimizing enzyme / pH parameters every 5 minutes (such as MMP-9 concentration 0.5→0.3μg / mL).

[0121] Transfer learning initialization: The pre-trained dataset (100 groups of liver / kidney chip data) is stored in the SSD (Samsung 870 EVO 1TB).

[0122] Dynamic phase delay controller: Real-time clock module (DS3231): Calibrates the timing of the stimulation signal, with a phase delay accuracy of ±0.1s (such as the liver chip triggering temperature control 7.395s in advance).

[0123] Digital potentiometer (AD5293): Adjusts the opening of the microfluidic valve to match the metabolite transport rate (such as =3mm / s).

[0124] Digital twin verification system: Edge computing server (Intel NUC11): GPU acceleration module (NVIDIA RTX A4000): Real-time solution of the Brinkman equation (COMSOL LiveLink), predicting the metabolic trajectory.

[0125] Safety threshold monitoring: If the deformation δ / δ_0>1.2, trigger the attenuation of the stimulation intensity through GPIO (such as reducing the light intensity to 50%).

[0126] Dual-redundancy communication module: CAN bus (PCA82C250): Transmits control instructions to the actuators (such as thermocouples, micropumps); Bluetooth 5.0 (TICC2640R2F): Remotely monitors the alarm signal (such as pushing to the mobile APP when RMSE>5%).

[0127] The above are all preferred embodiments of this application, and do not limit the protection scope of this application accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of this application shall be covered within the protection scope of this application.

Claims

1. A hydrogel-based organ-on-a-chip microfluidic control method, characterized in that: Comprising the following steps: Multi-modal stimulus co-regulation: Applying at least two external stimulus signals to the hydrogel microchannel, where the stimulus signals are selected from temperature, light, enzyme concentration or pH change, and dynamically adjusting the channel permeability, shear force or solute diffusion rate by changing the physical properties of the hydrogel; Multi-parameter fusion feedback: Synchronously detecting the hydrodynamic parameters and cell physiological parameters in the channel to generate multi-dimensional microenvironment state data; Closed-loop control: Based on the multi-dimensional microenvironment state data, adjusting the intensity, timing or spatial distribution of the stimulus signals through a dynamic algorithm to achieve the co-regulation of metabolism and hydrodynamics across the organ-on-a-chip.

2. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 1, characterized in that: In the multi-modal stimulus co-regulation step, the application of the temperature stimulus adopts a stepped temperature increase.

3. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 1, wherein: In the closed-loop control step, the dynamic algorithm adopts a hierarchical control architecture. The bottom-layer PID controller adjusts the temperature / light stimulus intensity, and the top-layer reinforcement learning model optimizes the enzyme / pH stimulus strategy once per cycle; the reinforcement learning model is initialized by transfer learning, and the pre-training dataset contains the historical experimental data of several organ-on-a-chips.

4. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 1, wherein: In the cross-organ collaborative regulation, there is a phase delay in the application of the stimulation signals between the upper chip and the lower chip. The temperature stimulation of the upper chip is triggered with a first phase delay time seconds earlier, and the light stimulation of the lower chip is triggered with a second phase delay time seconds earlier.

5. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 4, characterized in that: The first phase delay time , which is dynamically adjusted according to the hydrogel fatigue coefficient α and the metabolite concentration change rate β. The calculation model is as follows: Among them, is the physical transmission distance between the upper chip and the lower chip; represents the average diffusion rate of the drug or metabolite in the chip, represents the drug concentration gradient between the upper chip and the lower chip, is the reference concentration, is the time constant; is the number of hydrogel deformation cycles, fatigue coefficient , β is the metabolite concentration change rate, detected in real time by a fluorescence probe; is the drug concentration feedback gain coefficient, ; The calculation model of the second phase delay time is as follows: , where is a set value.

6. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 5, characterized in that: Among them Dynamically obtain values, and the calculation model is as follows: 。 7. The hydrogel-based organ-on-a-chip microfluidic control method according to any one of claims 4 to 6, characterized in that: The first phase delay time and the second phase delay time The regulation results are input into the digital twin model of the organ. Based on the finite element method, the porous medium flow equation and the solute diffusion equation are solved to predict the metabolic trajectory after delay adjustment in real time. If the prediction error is greater than the set error, the control parameter reset is triggered.

8. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 7, wherein: The digital twin model integrates a microchannel deformation prediction module, calculates the geometric changes of the channel in real time based on the viscoelastic constitutive equation of the hydrogel, and if the predicted deformation exceeds the safety threshold, triggers a reduction in the global stimulus intensity.

9. The hydrogel-based organ-on-a-chip microfluidic control method according to claim 8, characterized in that: In the viscoelastic constitutive equation of the digital twin model, a fatigue degree-metabolism coupling correction factor is introduced , and the calculation model is as follows: ； wherein, is the metabolic rate reference value, is the fatigue coefficient, is the real-time metabolite concentration change rate.

10. A microfluidic control device, characterized in that: For implementing the hydrogel-based organ-on-a-chip microfluidic control method according to any one of claims 1-9, comprising the following modules: Multi-modal stimulus co-regulation module: Used to apply external stimuli and dynamically adjust the channel permeability, shear force or solute diffusion rate by changing the physical properties of the hydrogel; Multi-parameter fusion feedback module: The input end is connected to the output end of the multi-modal stimulus co-regulation module, and is used to synchronously detect the hydrodynamic parameters and cell physiological parameters in the channel to generate multi-dimensional microenvironment state data; Closed-loop control module: The input end is connected to the output end of the multi-parameter fusion feedback module, and is used to adjust the intensity, timing or spatial distribution of the stimulus signals according to the multi-dimensional microenvironment state data to achieve the co-regulation of metabolism and hydrodynamics across the organ-on-a-chip.

Images