Drug screening and disease model construction method for liver organoid

Abstract

The invention provides a method for drug screening and disease model construction of liver organoid, relates to the technical field of medical research, and adopts multi-modal data acquisition, fusion and intelligent analysis and quantum computing processing to optimize comprehensive application of data, so that the richness, accuracy and analysis efficiency of the data are remarkably improved, and the risk of drug screening and disease model construction of liver organoid is reduced. Multi-source data of genomics, transcriptomics and the like are collected comprehensively, an accurate research system of immune-hepatic disease interaction based on organoid is constructed, an in-vivo complex immune-hepatic disease microenvironment can be highly simulated, the proportion and activation state of immune cells are accurately regulated and controlled in an in-vitro model, functional and phenotypic changes of the immune cells are monitored in real time, and the immune-hepatic disease interaction accuracy is improved. According to the present invention, the interaction mechanism between the immune cells and the liver organ cells is deeply revealed, such that the accurate information basis is provided for the accurate determination of the drug action target, the development of the innovative drug for the immune check point or tumor microenvironment regulation is easily achieved, and the pertinence and the success rate of the drug research and development are substantially improved.

Description

Technical Field

[0001] The present invention relates to the field of medical research technology, and in particular to methods for drug screening and disease model construction of liver organoids. Background Art

[0002] According to the method and system for generating and utilizing analyte measurement values ​​disclosed in China Publication No. "CN118102976A". In certain aspects, a monitoring system includes: a continuous analyte sensor configured to generate an analyte measurement value associated with an analyte level of a patient; and a sensor electronics module coupled to the continuous analyte sensor and configured to receive and process the analyte measurement value.

[0003] According to a system and method for determining kinetic parameters associated with a kinetic model of an imaging agent in the liver disclosed in China Publication No. "CN111770760A". An image reconstruction device is capable of receiving radiotracer activity corresponding to a predetermined time period, for example, these radiotracer activities may include PET scan data corresponding to multiple time frames. The radiotracer activity can be used to determine a liver time activity curve and a cyclic input function. The liver time activity curve and the cyclic input function can be used together with a kinetic model of the liver to generate kinetic parameters. These kinetic parameters can be used to determine liver scores, such as liver steatosis scores, liver inflammation scores, and liver cirrhosis scores. These scores indicate liver diseases, including non-alcoholic fatty liver disease, non-alcoholic steatohepatitis, and liver fibrosis.

[0004] The above patent documents and prior art have the following technical problems when used: Problem 1: Traditional liver disease research and drug screening technologies often rely on a single or limited number of data types. The single data dimension leads to an incomplete description of the state of liver organoids under physiological, pathological and drug effects. Problem 2: Existing drug screening methods lack effective simulation of the immune microenvironment of liver diseases, making it difficult to accurately evaluate the combined effects of drugs on liver organoids and immune cells. This results in poor effectiveness and targeting of screened drugs in clinical applications, as well as long screening cycles and high costs. Question three: Previous research methods on the mechanisms of liver diseases have been unable to accurately analyze the complex molecular mechanisms of interaction between immune cells and liver organoid cells, resulting in many potential therapeutic targets being missed, treatment methods being limited, and treatment effects being unsatisfactory, leading to the inability of some diseases to receive targeted and effective treatment. Summary of the invention

[0005] Technical issues solved In view of the shortcomings of the prior art, the present invention provides a method for drug screening and disease model construction of liver organoids, which solves the following problems: 1. The single disease data collection in disease research leads to incomplete research on drug action mechanisms; 2. Address the problem of long treatment cycles caused by low drug screening efficiency and insufficient precision; 3. Insufficient depth in research on disease mechanisms and difficulty in discovering therapeutic targets result in poor treatment effects.

[0006] Technical Solution To achieve the above objectives, the present invention is implemented by the following technical solutions: a method for drug screening and disease model construction of liver organoids, wherein the model construction method is as follows: Sp1: Constructing an organoid-based precise research system for immune-liver disease interactions: When constructing liver organoids, different types of immune cells, including macrophages, T lymphocytes, natural killer cells, etc., are introduced according to specific proportions and precise activation states, and the expression of cell-cell interaction molecules is regulated to establish an in vitro model that simulates a variety of immune-related liver diseases, including autoimmune hepatitis, viral hepatitis, liver tumor immune microenvironment, etc. At the same time, develop immune cell function and phenotypic analysis technology for this co-culture system, use specific molecular probes or fluorescent labeling reagents, combined with single-cell sequencing, mass spectrometry flow cytometry and other advanced methods, to monitor the functional status of immune cells in real time, including cytokine secretion, immune synapse formation, cell killing activity, etc., and phenotypic changes, including immune cell subset differentiation, surface marker expression changes, etc., so as to deeply explore the mechanism of interaction between immune cells and liver organoids, determine key information related to the immune-liver cell axis, and provide basic data and target basis for subsequent steps; Sp2: Establish a multimodal data acquisition system: Based on the above-mentioned immune-liver organoid co-culture model, collect data in all aspects, including genomics, transcriptomics, proteomics, metabolomics, and biophysical properties. Use quantum sensing technology, such as quantum dots or superconducting quantum interference devices, to monitor biomolecular activities with ultra-high sensitivity and precision, and add it to multimodal data to obtain richer and microscopic information. Biomolecular activities include enzyme activity, ion channel opening and closing, free radical generation, etc. Biophysical properties include organoid elastic modulus, conductivity and other data; Sp3: Multimodal data fusion and intelligent analysis: Develop multimodal data fusion technology to integrate data from different levels and sources, build complex mathematical models and deep learning algorithms, collaboratively analyze these data, discover key drug targets and disease mechanism associations, build a dynamic knowledge graph based on organoid data, connect data nodes at different time points and drug treatment conditions, and associate existing biological knowledge, clinical data and drug mechanism information. Through intelligent query and reasoning, predict the synergistic or antagonistic effects of new drug combinations in liver organoids, and design personalized drug screening and combination drug regimens; Sp4: Using quantum computing to process and optimize data: With the help of quantum computing's powerful parallel computing capabilities and unique quantum algorithms, including quantum annealing algorithms and quantum gate circuit algorithms, the fused large-scale organoid data can be quickly processed and deeply mined to build a more accurate and comprehensive quantum model of liver disease, predict the long-term behavior and disease outcomes of liver organoids under different drug interventions, further optimize the drug screening and disease model building process, improve the accuracy and reliability of the model, and ultimately achieve liver organoid drug screening and disease model building based on data optimization.

[0007] Preferably, in the construction of the organoid-based precise research system for immune-liver disease interactions, when specific molecular probes or fluorescent labeling reagents are used, a synthetic aptamer sequence is designed according to specific markers on the surface of immune cells or specific functional molecular structures in cells, and fluorescent groups are connected to the aptamers. The binding affinity and specificity of the aptamers to the target molecules are adjusted through chemical modification, so that the aptamers can accurately identify and label target immune cells and their function-related molecules in the complex liver organoid-immune cell co-culture system, and the fluorescence signal intensity is linearly correlated with the target molecule concentration or the cell functional state, which is convenient for accurate quantitative analysis.

[0008] Preferably, in the multimodal data acquisition system, for genomic data acquisition, a single-cell whole genome amplification technology based on a microfluidic chip is used, which can efficiently and uniformly amplify the genome of a single liver organoid cell or immune cell in a nanoliter reaction system, avoiding the bias problem in traditional amplification methods, ensuring that the collected genomic information can accurately reflect the heterogeneity of the cell population, and the microfluidic chip has good compatibility with the entire multimodal data acquisition platform, and can realize automated, high-throughput sample processing and data acquisition.

[0009] Preferably, in the multimodal data acquisition system, quantum sensing technology is used to monitor the activities of biomolecules, and the use of superconducting quantum interference devices (SQUIDs) requires the use of specially designed nanoscale biosensor probes. The surface of the probes is modified with recognition elements that can specifically bind to liver organoid-related biomolecules, such as antibodies or nucleic acid aptamers. When the biomolecules bind to the recognition elements, they cause slight changes in the magnetic field near the probe. The SQUID can accurately detect such changes and convert them into electrical signals. After signal amplification and processing, relevant data on the activities of biomolecules are obtained. The probe is reusable and its detection activity for biomolecules can be restored through specific cleaning and regeneration procedures.

[0010] Preferably, in the multimodal data fusion and intelligent analysis, multimodal data fusion technology is utilized, and the mathematical model based on data integration adopts a graph neural network-based architecture, in which different modal data are regarded as nodes in the graph, and the correlation between data is regarded as edges. Through the information transmission mechanism of the graph neural network, feature interaction and fusion between modal data are realized, thereby fully mining the intrinsic connections hidden in multi-source data. In addition, the model has adaptive learning capabilities and can continuously optimize its own parameters and structure according to new input data to adapt to the data fusion needs of liver organoids at different drug treatments or disease development stages.

[0011] Preferably, in the multimodal data fusion and intelligent analysis, a hybrid architecture combining a generative adversarial network (GAN) and a recurrent neural network (RNN) is adopted when analyzing multimodal data using a deep learning algorithm. GAN is used for data enhancement and feature extraction. Through adversarial training of the generator and the discriminator, virtual data samples similar to the real data distribution are generated to expand the data set size and mine potential data features. RNN is used to process the temporal information of the data, such as the dynamic change sequence of liver organoids during drug treatment. By learning and predicting the sequence data, the development trend and change law of the organoids under the action of drugs are analyzed. The two work together to improve the analysis accuracy and efficiency of multimodal data.

[0012] Preferably, in the process of constructing a dynamic knowledge graph based on organoid data in the multimodal data fusion and intelligent analysis, the construction and update of the knowledge graph adopts an automated information extraction and integration system, which can automatically extract entities related to liver organoids, immune cells, and drug action mechanisms, such as genes, proteins, cell types, drug names, etc., relationships, such as interactions, regulation, participation, etc., and attribute information from public biomedical databases, scientific research literature, and internal experimental data, and integrate this information into the knowledge graph through semantic analysis and data fusion technology, and continuously improve the structure and content of the graph using knowledge reasoning algorithms to ensure that the knowledge graph always keeps the latest biomedical research results incorporated and updated, and provides timely and comprehensive knowledge support for drug screening and disease mechanism research.

[0013] Preferably, the application of the quantum annealing algorithm for processing and optimizing data using quantum computing requires designing a special problem encoding method based on the characteristics of liver organoid data, converting optimization problems in drug screening and disease model construction, such as finding the best drug combination, determining key disease-related genes, etc., into an Ising model form that can be processed by the quantum annealing algorithm, and by controlling the state and interaction of quantum bits, using the quantum tunneling effect to quickly search for the optimal solution to the problem. In addition, the encoding method can adapt to data inputs of different types and sizes, thereby ensuring the effectiveness and accuracy of the quantum annealing algorithm in processing complex liver organoid data.

[0014] Preferably, the culture environment control of the liver organoid and immune cell co-culture system adopts an intelligent bioreactor system, which can monitor in real time and accurately regulate key parameters such as temperature, humidity, carbon dioxide concentration, nutrient concentration and cytokine concentration in the culture system, and can preset specific environmental change programs according to the requirements of different immune-liver disease models, simulate the dynamic changes of the microenvironment under physiological or pathological conditions in the body, provide stable and controllable in vitro environmental conditions for the growth, differentiation and interaction of liver organoids and immune cells, and ensure the reliability and repeatability of the experimental results.

[0015] Preferably, during the model construction process, distributed blockchain technology is used for data storage and management, and various types of data in the liver organoid experiment, including original data, processed data, analysis results, etc., are encrypted and stored. Each data block contains the hash value, timestamp and data source information of the data to ensure the integrity, non-tamperability and traceability of the data. Different research institutions or laboratories can securely share data resources through the blockchain network, promote data exchange and cooperation in the field of liver organoid research, and accelerate the progress of drug development and disease mechanism research.

[0016] Beneficial Effects The present invention provides a method for drug screening and disease model construction of liver organoids. It has the following beneficial effects: The present invention adopts the comprehensive application of multimodal data collection, fusion and intelligent analysis as well as quantum computing to optimize data processing, which significantly improves the richness, accuracy and analysis efficiency of data. From the comprehensive collection of multi-source data such as genomics and transcriptomics, and the use of quantum sensing technology to capture microscopic biological molecular activities, to the use of advanced graph neural network architecture to fuse data, adversarial generative networks and recurrent neural networks for collaborative analysis, and then to quantum annealing algorithms to process large-scale data to build an accurate liver disease quantum model, this series of steps realizes comprehensive, in-depth and rapid information mining of liver organoids in the process of drug screening and disease model construction, giving quantum computing powerful parallel computing capabilities. During drug screening, it can quickly and accurately predict the effects of different drug combinations in liver organoids, including synergistic or antagonistic effects, thereby efficiently screening out the most promising drug solutions, greatly shortening the new drug development cycle, reducing research and development costs, and at the same time improving the reliability of drug safety and effectiveness evaluation. The present invention constructs an organoid-based precise research system for immune-liver disease interactions, which can highly simulate the complex immune-liver disease microenvironment in the body. From the construction of an in vitro model of immune-liver disease, to the real-time monitoring of immune cells and organoids by specific molecular probes and advanced detection technologies, to the collection and analysis of multimodal data, it can quickly obtain a large amount of valuable data information, and predict drug effects through intelligent analysis. It can comprehensively evaluate the comprehensive effects of different drugs on liver organoids and immune cells, and design personalized drug screening programs based on the evaluation results. According to the disease characteristics and immune status of different patients, the most suitable drug combination can be screened out, which will help promote the development of precision medicine, improve the effectiveness and pertinence of drug treatment, reduce the use of ineffective drugs, reduce medical costs, greatly shorten the drug screening cycle in the process of new drug research and development, accelerate the speed of new drugs to market, and enable patients to benefit from new therapeutic drugs faster. 3. The present invention simulates in vitro models of various immune-related liver diseases, accurately regulates the interaction between immune cells and liver organoid cells, and combines advanced detection technology and data analysis methods to deeply analyze the molecular mechanisms in the occurrence and development of the disease. Based on the in-depth study of the immune-liver disease interaction mechanism, many potential therapeutic targets that are difficult to reach with traditional research methods can be discovered. Therapeutic drugs or intervention measures developed for these new targets are expected to break through the limitations of existing treatment methods and bring new breakthroughs to the treatment of liver diseases. During the treatment of the disease, the model can be used to monitor the state changes of liver organoids and immune cells in real time. By using multimodal data acquisition and analysis, the treatment effect can be accurately evaluated, and accurate disease monitoring research and management can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1A diagram of the model building method of the present invention; Figure 2 It is a diagram of the drug screening steps of the present invention; Figure 3 It is a step diagram of the pathological mechanism research of the present invention; Figure 4 It is a feature vector distribution diagram after multimodal data fusion of the graph neural network of the present invention; Figure 5 This is a visualization diagram of the quantum state of a single quantum bit after the rotation operation around the X-axis of the present invention; Figure 6 This is a graph showing the weight ratio of each modal data feature after multimodal data fusion of the present invention; Figure 7 This is a comparison chart of marker expression levels in different cell subpopulations of the present invention; Figure 8 It is a line graph showing changes in expression levels of specific genes in different cells of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Specific embodiment one: like Figure 1-Figure 8 As shown, the method of drug screening and disease model construction of liver organoids, the model construction method is as follows: Sp1: Constructing an organoid-based precise research system for immune-liver disease interactions: When constructing liver organoids, different types of immune cells, including macrophages, T lymphocytes, natural killer cells, etc., are introduced according to specific proportions and precise activation states, and the expression of cell-cell interaction molecules is regulated to establish an in vitro model that simulates a variety of immune-related liver diseases, including autoimmune hepatitis, viral hepatitis, liver tumor immune microenvironment, etc. At the same time, develop immune cell function and phenotypic analysis technology for this co-culture system, use specific molecular probes or fluorescent labeling reagents, combined with single-cell sequencing, mass spectrometry flow cytometry and other advanced methods, to monitor the functional status of immune cells in real time, including cytokine secretion, immune synapse formation, cell killing activity, etc., and phenotypic changes, including immune cell subset differentiation, surface marker expression changes, etc., so as to deeply explore the mechanism of interaction between immune cells and liver organoids, determine key information related to the immune-liver cell axis, and provide basic data and target basis for subsequent steps; Sp2: Establish a multimodal data acquisition system: Based on the above-mentioned immune-liver organoid co-culture model, collect data in all aspects, including genomics, transcriptomics, proteomics, metabolomics, and biophysical properties. Use quantum sensing technology, such as quantum dots or superconducting quantum interference devices, to monitor biomolecular activities with ultra-high sensitivity and precision, and add it to multimodal data to obtain richer and microscopic information. Biomolecular activities include enzyme activity, ion channel opening and closing, free radical generation, etc. Biophysical properties include organoid elastic modulus, conductivity and other data; Sp3: Multimodal data fusion and intelligent analysis: Develop multimodal data fusion technology to integrate data from different levels and sources, build complex mathematical models and deep learning algorithms, collaboratively analyze these data, discover key drug targets and disease mechanism associations, build a dynamic knowledge graph based on organoid data, connect data nodes at different time points and drug treatment conditions, and associate existing biological knowledge, clinical data and drug mechanism information. Through intelligent query and reasoning, predict the synergistic or antagonistic effects of new drug combinations in liver organoids, and design personalized drug screening and combination drug regimens; Sp4: Using quantum computing to process and optimize data: With the help of quantum computing's powerful parallel computing capabilities and unique quantum algorithms, including quantum annealing algorithms and quantum gate circuit algorithms, the fused large-scale organoid data can be quickly processed and deeply mined to build a more accurate and comprehensive quantum model of liver disease, predict the long-term behavior and disease outcomes of liver organoids under different drug interventions, further optimize the drug screening and disease model building process, improve the accuracy and reliability of the model, and ultimately achieve liver organoid drug screening and disease model building based on data optimization. Specific embodiment 2: like Figure 1-Figure 8 As shown, according to the contents in the above specific embodiments, the following contents are further disclosed: The method for constructing the above model further includes the following contents: In constructing an organoid-based precise research system for immune-liver disease interactions, when using specific molecular probes or fluorescent labeling reagents, the aptamer sequence is designed and synthesized according to the specific markers on the surface of immune cells or the specific functional molecular structure in the cells, and the fluorescent group is connected to the aptamer. The binding affinity and specificity of the aptamer to the target molecule are adjusted through chemical modification, so that it can accurately identify and label the target immune cells and their function-related molecules in the complex liver organoid-immune cell co-culture system, and the fluorescence signal intensity is linearly correlated with the target molecule concentration or the cell functional state, which is convenient for accurate quantitative analysis. By constructing an organoid-based precise research system for immune-liver disease interactions, it is possible to highly simulate the complex immune-liver disease microenvironment in vivo, accurately regulate the proportion and activation state of immune cells in in vitro models, and monitor their functional and phenotypic changes in real time, deeply revealing the interaction mechanism between immune cells and liver organoid cells, which provides an unprecedented accurate information basis for accurately determining drug targets. For example, in the simulation of the liver tumor immune microenvironment, a clear understanding of the interaction pattern between tumor-associated macrophages and T lymphocytes will help develop innovative drugs targeting immune checkpoints or tumor microenvironment regulation, greatly improving the targetedness and success rate of drug research and development, and is expected to break through the dilemma of poor drug efficacy caused by insufficient understanding of the immune mechanism of liver disease in traditional drug research and development.

[0021] In the establishment of a multimodal data acquisition system, for genomic data acquisition, a single-cell whole genome amplification technology based on a microfluidic chip is used. It can efficiently and uniformly amplify the genome of a single liver organoid cell or immune cell in a nanoliter reaction system, avoiding the bias problem in traditional amplification methods, ensuring that the collected genomic information can accurately reflect the heterogeneity of the cell population, and the microfluidic chip has good compatibility with the entire multimodal data acquisition platform, which can realize automated, high-throughput sample processing and data acquisition. Quantum sensing technology is used to monitor the activity of biomolecules. The use of superconducting quantum interference devices (SQUIDs) requires the use of specially designed nanoscale biosensor probes with surface modification. There are recognition elements that can specifically bind to liver organoid-related biomolecules, such as antibodies or nucleic acid aptamers. When biomolecules bind to recognition elements, they will cause slight changes in the magnetic field near the probe. SQUID can accurately detect this change and convert it into an electrical signal. After signal amplification and processing, relevant data on biomolecule activity can be obtained. The probe is reusable and can be restored to its detection activity for biomolecules through specific cleaning and regeneration procedures. Through multimodal data acquisition, it integrates multi-dimensional information from macrogenomics, transcriptomics, proteomics, metabolomics to microscopic biophysical properties and biomolecule activities monitored by quantum sensing, providing an unprecedented comprehensive data portrait for the study of liver organoids. For example, quantum sensing technology's ultra-high sensitivity monitoring of microscopic processes such as enzyme activity and ion channel opening and closing can capture subtle changes that are difficult to detect with traditional technologies, allowing researchers to have a more accurate understanding of the physiological and pathological states of liver organoids.

[0022] In multimodal data fusion and intelligent analysis, multimodal data fusion technology is used. The mathematical model based on data integration adopts a graph neural network-based architecture, which regards different modal data as nodes in the graph and the correlation between data as edges. Through the information transmission mechanism of the graph neural network, the feature interaction and fusion between the modal data are realized, so as to fully explore the intrinsic connections hidden in the multi-source data. The model has adaptive learning capabilities and can continuously optimize its own parameters and structure according to the new input data to adapt to the data fusion needs of liver organoids at different drug treatments or disease development stages.

[0023] In multimodal data fusion and intelligent analysis, a hybrid architecture combining generative adversarial networks (GANs) and recurrent neural networks (RNNs) is used to analyze multimodal data using deep learning algorithms. GANs are used for data enhancement and feature extraction. Through adversarial training of generators and discriminators, virtual data samples similar to the real data distribution are generated to expand the data set size and mine potential data features. RNNs are used to process the temporal information of data, such as the dynamic change sequence of liver organoids during drug treatment. Through learning and predicting sequence data, the development trend and change law of organoids under drug action are analyzed. The two work together to improve the analysis accuracy and efficiency of multimodal data. Multimodal data fusion technology is combined with intelligent analysis, and feature interaction between data is realized based on advanced mathematical models such as graph neural networks, which can mine deep-level associations hidden in data from different sources. The collaborative work of generative adversarial networks and recurrent neural networks not only expands the data set size, but also effectively processes the temporal information of data, accurately analyzes the dynamic change law of liver organoids during drug treatment, and thus more accurately predicts the drug effect and disease development trend, providing a strong basis for drug screening and disease mechanism research.

[0024] In the process of building a dynamic knowledge graph based on organoid data in multimodal data fusion and intelligent analysis, the construction and update of the knowledge graph adopts an automated information extraction and integration system. This system can automatically extract entities related to liver organoids, immune cells, and drug mechanisms from public biomedical databases, scientific research literature, and internal experimental data, such as genes, proteins, cell types, drug names, etc., relationships such as interactions, regulation, participation, etc., and attribute information. This information is integrated into the knowledge graph through semantic analysis and data fusion technology, and the structure and content of the graph are continuously improved using knowledge reasoning algorithms to ensure that the knowledge graph always incorporates and updates the latest biomedical research results, providing timely and comprehensive knowledge support for drug screening and disease mechanism research.

[0025] Using quantum computing to process and optimize data, the application of quantum annealing algorithm needs to design a special problem encoding method according to the characteristics of liver organoid data, and convert the optimization problems in drug screening and disease model construction, such as finding the best drug combination and determining key disease-related genes, into the Ising model form that can be processed by quantum annealing algorithm. By controlling the state and interaction of quantum bits, the optimal solution of the problem can be quickly searched using the quantum tunneling effect. Moreover, this encoding method can adapt to data input of different types and scales, ensuring the effectiveness and accuracy of quantum annealing algorithm in processing complex data of liver organoids. Quantum computing has powerful parallel computing capabilities and unique algorithms, such as quantum annealing algorithm for drug screening and disease model construction optimization problems, and can quickly search for the optimal solution of complex problems. Its encoding method that can adapt to data input of different types and scales ensures the efficiency and accuracy in processing massive and complex data of liver organoids, greatly shortens the data processing time, accelerates the entire research process, and helps to convert research results into clinical applications faster.

[0026] The culture environment control of the liver organoid and immune cell co-culture system adopts an intelligent bioreactor system, which can monitor and accurately control key parameters such as temperature, humidity, carbon dioxide concentration, nutrient concentration and cytokine concentration in the culture system in real time. It can also preset specific environmental change programs according to the needs of different immune-liver disease models, simulate the dynamic changes of the microenvironment under physiological or pathological conditions in the body, and provide stable and controllable in vitro environmental conditions for the growth, differentiation and interaction of liver organoids and immune cells, ensuring the reliability and repeatability of experimental results.

[0027] During the model construction process, distributed blockchain technology is used for data storage and management. Various data in the liver organoid experiment, including raw data, processed data, analysis results, etc., are encrypted and stored. Each data block contains the data hash value, timestamp and data source information to ensure the integrity, non-tamperability and traceability of the data. Different research institutions or laboratories can share data resources securely through the blockchain network, promote data exchange and cooperation in the field of liver organoid research, accelerate the progress of drug development and disease mechanism research, and the intelligent bioreactor system accurately controls the culture environment of the liver organoid and immune cell co-culture system. The distributed blockchain technology can be used to The management of data provides a stable, reliable and shareable support environment for the entire research process. The bioreactor simulates the dynamic changes of the in vivo microenvironment to ensure the repeatability and scientificity of the experimental results, which enhances the comparability of data between different batches of experiments and is conducive to the verification and promotion of research results. Blockchain technology ensures the safe sharing of data and promotes exchanges and cooperation in the field of liver organoid research on a global scale. Different research institutions can conduct more in-depth joint analysis and knowledge expansion based on shared data, avoiding duplicate research and data islands, and accelerating the technological progress and innovative development of the entire liver organoid field for drug screening and disease model construction, providing a more powerful research synergy for conquering liver diseases. Specific embodiment three: like Figure 1-Figure 8 As shown, based on the contents in the above specific embodiments, the following contents are further disclosed:.

[0029] In the entire model building method, the algorithm content of each step is as follows: Regarding single-cell sequencing and mass cytometry in the construction of an organoid-based precision research system for immune-liver disease interactions, the following content is further disclosed: The implementation steps of single-cell sequencing are as follows: First, single immune cells or liver organoid cells need to be isolated from the co-culture system. Generally, this is done through specific cell sorting techniques, such as fluorescence-activated cell sorting (FACS) or microfluidic chip sorting. The dispersed single cells are captured into corresponding reaction containers (such as 96-well plates, etc.). The cells are then lysed to release nucleic acids (mainly DNA or RNA, depending on the sequencing type, DNA for genome sequencing and RNA for transcriptome sequencing). The corresponding library construction kit is then used to construct the library. This process involves adding various adapters, primers, etc., amplifying the nucleic acids, etc., so that they meet the subsequent loading requirements of the sequencer. Finally, the constructed library is placed in the sequencer for sequencing.

[0030] Operation logic: The sequencer identifies the order of bases on the nucleic acid chain by detecting the fluorescent signals (different bases carry different fluorescent markers) or electrical signals (such as nanopore sequencing technology by detecting the change in current when nucleic acid molecules pass through nanopores) generated during the sequencing process, and then obtains the genome or transcriptome information of the cell. For example, in the second-generation sequencing technology based on sequencing by synthesis, each time a new fluorescently labeled base is added, the sequencer will record the corresponding fluorescent signal, and after a series of complex image recognition and data analysis, it will be converted into base sequence information.

[0031] Beneficial effects: Cell heterogeneity can be analyzed at the single-cell level, that is, the differences in gene expression, gene mutation, etc. between different cells. The characteristics of each immune cell or liver organoid cell can be accurately known, which is helpful for in-depth exploration of the interaction mechanism between immune cells and liver organoid cells, such as finding out which cells have up-regulated or down-regulated expression of specific genes during the interaction process, which provides a basis for the subsequent determination of key information related to the immune-liver cell axis The relevant implementation steps of mass spectrometry flow cytometry are as follows: first, the cells in the co-culture system are collected, washed, and pre-treated to remove impurities and excess culture medium components, and then the markers on the cell surface or inside the cell are labeled with specific antibodies with metal tags. These antibodies can specifically bind to the corresponding markers, and the antibodies corresponding to different markers carry different metal labels (such as lanthanide elements, etc.). After labeling, the cells are made into a single cell suspension and passed through the detection area one by one through the injection system of the flow cytometer. In the detection area, the cells are excited by the laser, and the markers bound by the metal-tagged antibodies will emit specific ion signals. The mass spectrometry detection system in the mass spectrometry flow cytometer will detect and analyze these ion signals.

[0032] Operation logic: When cells pass through the detection area one by one and are irradiated by laser, the metal tags will be ionized. Different metal elements have their specific mass-to-charge ratio (ratio of mass to charge) in the mass spectrometer. The mass spectrometer can distinguish different markers and their corresponding expression levels by detecting the mass-to-charge ratio and signal intensity of these ions. According to the pre-set relationship between markers and cell functions and phenotypes, the functional status and phenotypic changes of cells can be analyzed.

[0033] Beneficial effects: It can detect multiple markers (dozens or even hundreds) at the same time, and can obtain richer cell information than traditional flow cytometry. It can comprehensively analyze the functional status of immune cells in the co-culture system (such as changes in markers corresponding to cytokine secretion, etc.), phenotypic changes (such as the expression of different markers corresponding to the differentiation of immune cell subpopulations), etc., so as to more accurately understand the interaction between immune cells and liver organoid cells, and provide detailed data support for subsequent research.

[0034] The steps to monitor the activities of biomolecules using a superconducting quantum interference device (SQUID) are as follows: First, a special nanoscale biosensor probe must be designed. The surface of the probe must be modified with recognition elements (such as antibodies or nucleic acid aptamers) that can specifically bind to liver organoid-related biomolecules. The modified probe must be placed in a co-culture system or a sample environment containing target biomolecules. When the biomolecule binds to the recognition element, it will cause slight changes in the magnetic field near the probe. The SQUID device, through its structural characteristics such as superconducting rings and Josephson junctions, can accurately detect this change in the magnetic field and convert it into an electrical signal. Finally, the converted electrical signal is amplified and processed to obtain relevant data on the activities of biomolecules (such as indirect reflections of enzyme activity, ion channel opening and closing, free radical production, etc.).

[0035] Operation logic: Based on the principle of superconducting quantum interference, the superconducting ring in the SQUID is extremely sensitive to changes in the magnetic field. When the magnetic field at the probe changes due to the binding of biological molecules to the recognition element, the superconducting current in the superconducting ring will change accordingly. Through the quantum characteristics of the Josephson junction, this current change will be converted into a measurable voltage signal or other electrical signal form, and the activity of the biological molecule can be inferred based on the characteristics of the electrical signal (such as amplitude, frequency, etc.). For example, if the activity change of an enzyme causes a change in the concentration of the substrate or product during its catalytic reaction, which in turn causes a change in the magnetic field near the biosensor probe that specifically binds to the substrate or product, the change in the electrical signal detected by the SQUID can indirectly reflect the change in enzyme activity.

[0036] Beneficial effects: It can monitor the activities of biomolecules with ultra-high sensitivity and precision, capture very small changes at the biomolecular level, and supplement the multimodal data to make the acquired information richer and deeper into the microscopic level. It helps to more accurately understand the intracellular biochemical processes, intermolecular interactions, etc. in liver organoid research, and provide more detailed basic data for drug screening and disease model construction. For example, it can more accurately discover the early changes of biomolecules in liver organoid cells under the action of drugs, thereby providing strong clues for the study of drug action mechanisms.

[0037] The multimodal data fusion technology based on graph neural network further includes the following contents: Suppose we have data of different modalities represented as , consider them as nodes in the graph, nodes in the graph neural network The eigenvector of , in The update formula of the layer can be roughly expressed as: ; in: Representation Node In the The feature vector of the layer, for example, if Corresponding to a feature node of transcriptomics data, this vector contains the numerical representation of the transcriptome-related features. The initial value can be the original data or encoded data of the feature after preprocessing. Is a node The set of neighboring nodes, that is, the node Other nodes that are connected (data are associated with each other) reflect the mutual connection between different modal data. For example, if there is an association between the transcriptomics data node and the proteomics data node, then the corresponding nodes are in each other's neighborhood set; Usually it is a normalization coefficient, which is used to perform reasonable weight normalization when aggregating neighborhood node information to avoid excessive influence of some nodes. The calculation method may be related to factors such as the degree of the node (the number of connected edges); It is The learnable weight matrix of the layer continuously adjusts its value through training to find the best way to fuse the features of different modal data. Different dimensions correspond to the weighted processing of different elements of the input feature vector, which determines the importance of the feature in the fusion process.

[0038] It is The bias vector of the layer is used to adjust the offset of the feature so that the network can better fit the data features.

[0039] It is an activation function, such as the commonly used ReLU, which introduces nonlinear factors into the network so that the network can learn more complex data relationships.

[0040] Implementation steps: First, preprocess the data of different modalities (genomics, transcriptomics, proteomics, metabolomics, etc.), including data normalization, encoding and other operations, so that it can be represented in the form of appropriate feature vectors and input into the graph neural network as nodes of the graph; construct the graph structure, determine the connection relationship between different nodes (different modal data) based on existing biological knowledge, prior data associations, etc., that is, determine the edge situation, and form the initial graph structure; initialize the parameters of the graph neural network, including the weight matrix and bias vector of each layer, usually using random initialization; convert the feature vector of the data node into Input the network, perform multi-layer feature propagation and update according to the above formula, and continuously update the feature vector of the node by aggregating the features of the neighboring nodes and processing them through activation functions at each layer, so as to realize the interactive fusion of features of data of different modalities; through supervised (for example, the accurate effects of some known drug treatments are used as labels) or unsupervised (such as based on the similarity of the data itself) learning methods, use the loss function (such as mean square error loss, cross entropy loss, etc., selected according to the specific task) to measure the difference between the fused features and the target, back-propagate and update the network parameters, and after multiple rounds of training, the network can achieve a better fusion effect.

[0041] Operation logic: At each layer of the network, each node collects the feature information of its neighboring nodes, performs weighted summation through the weight matrix, adds the bias vector, and performs nonlinear transformation through the activation function to update its own feature vector. As the number of layers increases, the features of different modal data will gradually merge, and the originally isolated features of each modal data will penetrate and integrate with each other, excavating the inherent connections hidden in the multi-source data, and finally outputting the fused feature representation, which can comprehensively reflect the key information of each modal data, and can be used for subsequent disease mechanism analysis, drug target mining and other tasks.

[0042] Beneficial effects: It can make full use of the correlation between data of different modalities, effectively integrate data from multiple levels such as genomics and transcriptomics, overcome the limitations of analyzing data of a certain modality alone, and mine more comprehensive and in-depth key information, such as discovering key drug targets or disease-related mechanisms that are reflected in gene expression, protein changes, and metabolic levels at the same time, providing more powerful data support for subsequent precise drug screening and disease model construction.

[0043] A hybrid architecture combining Generative Adversarial Networks (GANs) and Recurrent Neural Networks (RNNs) for analyzing multimodal data Generative Adversarial Network (GAN) part: Generator The goal is to generate data that is as similar to the real data distribution as possible. , is a random noise vector that follows a priori distribution (such as normal distribution is the identity matrix; Discriminator The task is to distinguish real data And the data generated by the generator , the discriminator output or Indicates the probability that the input data is real data or generated data. Generally, a Sigmoid function is used to map the output from 0 to 1 to an interval, for example: ; ; in It is a function of the discriminator, usually composed of multiple neural network layers (such as fully connected layers, etc.), which extracts features and processes input data. is the Sigmoid function , convert the output into a probability value.

[0044] The training goal of GAN is to achieve the effect that the data generated by the generator is difficult to distinguish from the real data by optimizing the following loss function (taking minimization as an example): ; in: Represents the distribution of real data Data , calculate the expectation of the logarithm of the probability that the discriminator judges it to be true data, that is, we hope that the discriminator can accurately judge the real data as true, and the larger this expectation, the better (when minimizing the overall goal, this item will play a corresponding role in the optimization process such as derivation); Represents the pair generated by the generator according to random noise , obeys the distribution ) generated data , calculate the logarithm expectation of the probability that the discriminator judges it to be false, that is, we hope that the data generated by the generator can deceive the discriminator and make the discriminator think it is real data. The smaller this expectation is, the better (it also has a corresponding effect in the optimization process).

[0045] Recurrent Neural Network (RNN) part: For a time series data is the sequence length, at each time step , hidden state The update formula can be expressed as: ; in: is the time step The input data, such as the multimodal data feature vector at a certain time point during the drug treatment of liver organoids, may be comprehensive features such as gene expression and protein content.

[0046] It is the weight matrix input to the hidden layer, which determines the weight of the input data on the hidden state. Different elements correspond to the weighting of different dimensions of the input vector. It is the weight matrix from the previous hidden state to the current hidden state, reflecting the importance and influence of historical information at the current moment.

[0047] is the bias vector of the hidden layer, which adjusts the offset of the hidden state; It is an activation function (such as the commonly used tanh function, etc.), which introduces nonlinearity into the network so that it can handle complex time series changes.

[0048] Final RNN output (The output layer structure can be set according to the specific task, such as predicting the features of the next time step, etc.) can be expressed as: ; in, is the activation function of the output layer (selected according to the task, such as linear function, Sigmoid, etc.), is the weight matrix from hidden state to output layer, is the bias vector of the output layer.

[0049] Implementation steps: GAN part: Sp1: Initialize the generator and the discriminator The network parameters are usually randomly initialized to set the network structure (for example, the generator uses a multi-layer deconvolution network to generate a structure similar to real data from random noise, and the discriminator uses a multi-layer convolution network to extract and judge features); Sp2: Extract a batch of real data from the real data set , and generate a batch of random noise vectors by sampling from a random noise distribution (such as a normal distribution) ; Sp3: Real data Input Discriminator , calculate the discriminator's judgment output on the real data , random noise Input Generator Get the generated data , and then Input Discriminator get ; Sp4: Calculate the loss value according to the above loss function , through the back propagation algorithm, the discriminator and generator The parameters of the model are updated, the discriminator is optimized to make it more accurate in distinguishing between real and generated data, and the generator is optimized to make the data it generates more able to deceive the discriminator. This process is repeated for multiple rounds until the data generated by the generator is similar enough to the real data in distribution.

[0050] RNN part: Sp1: Preprocess the multimodal data of the time series, such as normalization, to make it meet the network input requirements, determine the time series length of the input data and the feature dimension of each time step, etc.; Sp2: Initialize the network parameters of the RNN (weight matrix , , and the bias vector ,wait); Sp3: Input the time series data into the RNN in time steps, and calculate the hidden state update formula at each time step , and based on According to the output formula, the corresponding output is obtained ; Sp4: According to the loss function set for the specific task (such as the mean square error loss can be used to predict the next time step feature, etc., to compare the difference between the predicted value and the true value), the network parameters of the RNN are updated through the back propagation algorithm. After multiple rounds of training, the RNN can better process and predict the rules in the time series data; Combination part: The data generated by GAN (virtual data samples after data enhancement and feature extraction) are organized together with the original real data in a reasonable way such as time series order as the input of RNN. The ability of RNN to process the temporal information of data is used to analyze the dynamic change sequence of liver organoids under the action of drugs at different time points, and to explore the development trends and change laws.

[0051] Operation logic: GAN part: The discriminator and the generator compete with each other during the training process. The discriminator continuously learns the characteristics of real data and tries to accurately distinguish between real and generated data, while the generator continuously adjusts the distribution of the data it generates to make it closer to the real data. Through this adversarial training, the generator can mine the potential feature distribution in the real data and generate virtual data samples similar to the real data distribution, which expands the size of the data set and can extract some features that were originally hidden deeper, such as discovering new data feature patterns that may be related to drug effects.

[0052] RNN part: When processing time series data, by continuously updating the hidden state and integrating historical information into the current calculation, it can capture the dependencies and changing patterns of data in the time dimension. For example, in the process of drug treatment of liver organoids, how multimodal data such as gene expression and metabolites change dynamically over time, so as to predict the data situation at future time points and analyze the development trend of organoids under the action of drugs.

[0053] Combination part: The data generated by GAN supplements the deficiencies of the original data and enriches the data content. RNN uses its own ability to process time series to analyze the time series containing the expanded data. The two work together to more comprehensively and accurately analyze the information contained in multimodal data, taking into account both the data characteristics themselves and the dynamic changes of the data over time.

[0054] Quantum Annealing Algorithm For a The Hamiltonian (operator describing the energy of a system) can be expressed as: ; in: is the Hamiltonian of the system, which is a parameter that changes with time (or during annealing) ) changes, Usually the value range is between 0 and 1, which is used to control the annealing process, such as represents the initial state, represents the final state; and are coefficient functions, and they also vary with change, Usually at the beginning of annealing ) value is large, ensuring that the system is in an initial state that is easier to control (similar to the classic disordered state). Increase gradually decrease; On the contrary, it is small at the beginning and then The increase gradually increases, so that the system gradually changes to the target state that reflects the interaction, and the two satisfy (A common setting method); Pauli Operator, acting on On qubits, its eigenvalue is , used to describe the two basic states of a quantum bit (similar to 0 and 1 of a classical bit), and represents the state information of a quantum bit in quantum computing; It means and The parameter of the interaction strength (coupling strength) between quantum bits is It means that there is no interaction between the two qubits, and the non-zero value reflects the correlation between them. For example, when converting the optimization problem in drug screening and disease model construction into the Ising model form, It can represent the mutual influence relationship between different drug targets, disease-related genes and other factors; For all pairs of quantum bits that interact with each other, Perform the summation, that is, consider the coupling between all relevant quantum bits in the whole system; The entire quantum annealing algorithm is to find this Hamiltonian The ground state (the state with the lowest energy) corresponds to the optimal solution to optimization problems (such as finding the best drug combination, identifying key disease-related genes, etc.).

[0055] Implementation steps: Sp1: Problem encoding: Based on the characteristics of liver organoid data and the optimization problems to be solved in drug screening and disease model construction, a special problem encoding method is designed to convert the actual problem into the above-mentioned Ising model form and determine the number of quantum bits , the interaction strength between quantum bits Equivalent parameters, that is, to construct the corresponding Hamiltonian Expression, so that the quantum annealing algorithm can handle the problem. For example, if you want to find the best combination of three drugs, you may use three quantum bits to represent whether these three drugs are selected (the quantum bit state corresponds to Indicates whether to select or not select), and other parameters are determined by analyzing the synergistic or antagonistic interactions between drugs; Sp2: Initialize the system: Initialize the quantum bits in the quantum annealing device (such as a specific quantum computer or quantum simulator) to a suitable state, generally a uniform and relatively disordered state, corresponding to The situation at this time Large, making the system mainly affected This effect facilitates the start of the subsequent annealing operation; Sp3: Annealing process: According to the set annealing schedule, gradually change the value of the parameter to make Gradually decrease, Gradually increases, in this process, the state of the quantum bit will gradually evolve from the initial disordered state to a lower energy state under the influence of quantum properties such as the quantum tunneling effect, that is, towards the Hamiltonian The system continuously searches for the possible optimal solution. Sp4: Read result: When the annealing process is completed ( ), read the final state of the quantum bit, and convert the state of the quantum bit ( The data (values) are converted back into solutions to actual problems, such as determining the final drug combination, finding key disease-related genes, etc.

[0056] Quantum gate circuit algorithm For the rotation operation of a single quantum bit (such as rotation around the X-axis), the following mathematical representation corresponding to the unitary operator quantum gate operation can be used to ensure the normalization of the quantum state before and after the operation and other properties: ; in: Just around The axis rotation angle is The operator corresponding to the quantum gate operation, which acts on the state vector of the quantum bit and changes the state of the quantum bit; is the rotation angle parameter, by changing The value of can achieve different degrees of quantum bit state rotation, such as It is a common specific rotation operation that realizes a specific change in the state of the quantum bit; Pauli operator, which is a The matrix ( , is the state of the quantum bit in operator for the direction-dependent properties, and Combined with the realization of quantum bit winding The rotation operation of the axis acts on the state vector of the quantum bit (usually represented by a two-dimensional complex vector, such as ,satisfy ) on, changing its state; For the two-qubit controlled-NOT, CNOT gate, a commonly used two-bit gate operation, its operator is expressed as: ; The way it works is: if the control bit (the first bit) is 0, the target bit (the second bit) remains unchanged; if the control bit is 1, the target bit state flips (0 to 1, 1 to 0). Mathematically, this The matrix of is multiplied with the composite state vector (a complex vector with a dimension of 4) representing the state of the two quantum bits to realize the corresponding state change; In actual quantum computing tasks, complex quantum circuits are constructed by combining multiple such basic quantum gate operations such as single-bit gates and two-bit gates to achieve specific computing tasks, such as specific processing and analysis operations on large-scale data after liver organoid fusion. The operator corresponding to the overall quantum circuit can be expressed as the product of multiple quantum gate operators (in the order of quantum gate operations), for example: ; in, It can be various single-bit gates, two-bit gates and other quantum gate operators. Through such sequential operations, a series of transformations are performed on the input initial quantum state (representing the encoding form of data and other information), and finally the quantum state corresponding to the calculation result is obtained, which is then converted into an actual readable result through measurement and other means.

[0057] Implementation steps: Sp1: Data encoding: First, the data related to liver organoids (such as multimodal data after fusion, etc.) is encoded and converted into the initial state vector form of quantum bits. This process requires the design of a suitable encoding method based on the characteristics of the data and the problem to be solved. For example, different eigenvalues ​​of the data can be mapped to different states of quantum bits or the combined states of multiple quantum bits. Sp2: Quantum circuit construction: According to the specific computing task (such as specific feature extraction and relationship analysis of data for drug screening and disease model construction), select and combine appropriate quantum gates (single-bit gates, two-bit gates, etc.) to build quantum circuits, determine the type of quantum gates, parameters (such as rotation angle, etc.) and operation sequence, just like writing a program for a classical computer, except that quantum gates are used to build circuits to implement computing logic; Sp3: Quantum state evolution: The encoded quantum bit initial state is input into the constructed quantum circuit, and the quantum state is operated in sequence according to the operation order of the quantum gate. Each quantum gate performs matrix multiplication and other operations with the quantum state vector through the corresponding operator to make the quantum state evolve continuously and realize quantum computing processing of data. This process uses the superposition and entanglement characteristics of quantum bits to perform parallel computing and complex information processing; Sp4: Result measurement and interpretation: After the quantum circuit operation is completed, the final quantum state is measured. The measurement operation will cause the quantum state to collapse to a certain eigenstate (for example, the quantum bit changes from a superposition state to a certain state of or). Through the pre-designed encoding and decoding rules, the measured quantum bit state is converted into actual meaningful data results, such as recommendations on drug screening, analysis results of key parameters in disease models, etc.

[0058] Operation logic: The quantum gate circuit algorithm uses the characteristics of quantum bits that are different from classical bits, such as superposition (a quantum bit can be in a superposition state of and at the same time, and multiple quantum bits can form more complex superposition and entangled states). In quantum circuits, when quantum states are operated through quantum gates, all possible superposition states are actually processed in parallel at the same time, which is equivalent to performing calculations in multiple data state spaces at the same time, greatly improving the computing efficiency. Moreover, by cleverly constructing quantum circuits and using the combination and operation sequence of different quantum gates, various complex computing logics can be realized, and deep relationships and features hidden in the data can be excavated. For processing large-scale and complex data of liver organoids, more comprehensive and accurate analysis can be performed, providing strong support for drug screening and disease model construction.

[0059] Beneficial effects: When processing complex data of liver organoids, the quantum gate circuit algorithm can quickly process and deeply mine large-scale data with its powerful parallel computing capabilities and unique ability to operate on quantum states. For example, it can more efficiently analyze the complex correlations between liver organoid data under different drug interventions, build a more accurate and comprehensive quantum model of liver disease, and accurately predict the long-term behavior and disease outcomes of liver organoids under different drug interventions. Compared with classical data analysis and calculation methods, it can better optimize the drug screening and disease model construction process, improve the accuracy and reliability of the entire research, and help discover more effective drug treatment options and a deeper understanding of the occurrence and development of the disease. Specific embodiment four: like Figure 1-Figure 8 As shown, according to the contents in the above specific embodiments, the following contents are further disclosed: According to step Sp1 in the above specific embodiment 1, a precise research system for immune-liver disease interaction based on organoids is constructed: when constructing liver organoids, different types of immune cells are introduced according to specific proportions and precise activation states, wherein the specific proportions for different immune cells are as follows: For the simulation of autoimmune hepatitis model: Macrophages can account for about 10%-20% of the total number of liver organoid cells, among which the ratio of M1 macrophages (pro-inflammatory type) to M2 macrophages (anti-inflammatory and repair type) is about 3:1 to 4:1. M1 macrophages initiate the immune inflammatory response of liver organoids by secreting a large number of pro-inflammatory cytokines such as TNF-α, IL-1β, IL-6, etc. at the initial stage of the disease, while M2 macrophages participate in subsequent tissue repair attempts and inflammation regulation to a certain extent; Among T lymphocytes, the ratio of CD4+T cells to CD8+T cells is roughly 2:1 to 3:1, and Th17 cells (secreting pro-inflammatory factors such as IL-17) account for 15%-25% of the total CD4+T cells, and Th1 cells (secreting IFN-γ, etc.) account for about 30%-40%. These T lymphocyte subsets work synergistically to recognize abnormal antigens on the surface of liver organoid cells and launch immune attacks, leading to liver cell damage and continued inflammation; Natural killer cells account for a relatively small proportion, approximately 5%-10% of the total number of liver organoid cells. Their main function is to kill virally infected or cancerous liver organoid cells by releasing perforins, granzymes, etc., while secreting cytokines such as IFN-γ to regulate the intensity of the immune response. In the autoimmune hepatitis model, their activity is complexly regulated by other immune cells and the liver microenvironment.

[0061] For viral hepatitis models: The proportion of macrophages is about 15%-25%, and in the early stage of viral infection, M1 macrophages respond quickly, and the proportion can temporarily increase to about 60%-70%. As the infection progresses, M2 macrophages gradually increase to balance the inflammatory response. In the later stage, the ratio of the two may be close to 1:1; In terms of T lymphocytes, the proportion of CD8+T cells is relatively high, accounting for 40%-50% of the total number of T lymphocytes. They mainly kill liver organoid cells infected by viruses. Among CD4+T cells, the proportion of Th1 cells increases to about 50%-60%, mainly assisting CD8+T cells to function and regulate the overall immune response. The proportion of natural killer cells is about 8%-12%. They actively participate in antiviral responses in the early stages of viral infection. Their activity and number are regulated by viral infection signals and cytokines secreted by other immune cells, such as IFN-α and IFN-β, which can enhance their antiviral activity.

[0062] For the liver tumor immune microenvironment model: The proportion of macrophages can be as high as 30%-40%, among which tumor-associated macrophages (TAMs) are mainly M2 type, accounting for about 70%-80% of the total number of macrophages. This type of macrophage can secrete a variety of growth factors (such as VEGF, TGF-β, etc.) and immunosuppressive cytokines (such as IL-10, etc.), promoting tumor angiogenesis, tumor cell proliferation and immune escape; Among T lymphocytes, CD8+T cell function is often suppressed, and its proportion in T lymphocytes may drop to 20%-30%. There are a large number of exhausted CD8+T cells, which are characterized by high expression of immunosuppressive receptors such as PD-1 and CTLA-4. Among CD4+T cells, the proportion of regulatory T cells (Treg) increases, accounting for 30%-40% of the total CD4+T cells. They suppress anti-tumor immune responses by secreting inhibitory cytokines such as IL-10 and TGF-β. The proportion of natural killer cells is about 3%-6%. Their anti-tumor activity is inhibited by various factors in the tumor microenvironment, such as immunosuppressive factors secreted by tumor cells and the action of TAM cells. However, they still retain a certain immune surveillance function and can be reactivated to exert anti-tumor effects under certain immunotherapy interventions.

[0063] These proportions are not absolutely fixed and will fluctuate according to the different stages of the disease, individual differences and subtle changes in experimental conditions. In the actual process of model construction, preliminary experiments and fine-tuning are needed to determine the immune cell proportion combination that is most suitable for simulating specific immune-liver disease interactions. Specific embodiment five: like Figure 1-Figure 8 As shown, according to the contents in the above specific embodiments, the following contents are further disclosed: The steps of the entire model method in practical application in drug screening and disease mechanism research are as follows: The drug screening steps are as follows: Sp1: Data preparation and model initialization: Input the basic information of the drug to be screened (chemical structure, preliminary speculation on the target, etc.) into the dynamic knowledge graph and quantum model built based on liver organoid data. At the same time, prepare the initial state monitoring data of the liver organoid and immune cell co-culture system, including various data obtained by the multimodal data acquisition system and the baseline value of the biomolecular activity data monitored by quantum sensing technology; Sp2: Drug treatment and data collection: Add the drug to be screened to the co-culture system of liver organoids and immune cells according to the set concentration gradient, and continuously use the multimodal data acquisition system and quantum sensing technology to collect multi-source data of the treated liver organoids at predetermined time intervals (such as every hour, every 6 hours, etc., determined according to the characteristics of the drug action and research needs), including dynamic changes in genomics, transcriptomics, proteomics, metabolomics, and biophysical properties data, as well as real-time data on biomolecular activities monitored by quantum sensors; Sp3: Data fusion and analysis: Use multimodal data fusion technology to integrate various types of data collected after drug treatment, use the constructed complex mathematical model and deep learning algorithm for collaborative analysis, and combine quantum computing to deeply mine the data and quickly extract key information, such as the effect of drugs on intracellular signaling pathways in liver organoids, changes in immune cell function and phenotype, and changing trends in biophysical properties of organoids. Through intelligent interaction and reasoning with dynamic knowledge graphs, predict the mechanism of action, possible efficacy, and potential toxicity risks of the drug in liver organoids, compare the data changes under different drug concentrations, and determine the dose-effect relationship of the drug: Sp4: Result evaluation and screening decision: Based on the data analysis results, comprehensively evaluate the effectiveness and safety indicators of the drug. For example, if the drug can significantly regulate the expression of specific genes related to the disease, improve the function of liver organoids, and the regulation of immune cells is beneficial to the improvement of the disease without obvious adverse changes in the activity of biological molecules or damage to the structure and function of organoids, it is judged as a potential drug. Based on the evaluation results, screen out drugs with better efficacy and safety characteristics to enter further in vivo experiments or more in-depth preclinical research stages; The steps for studying the pathological mechanism are as follows: Sp1: Disease model construction and data collection: Based on the previously constructed liver organoid disease model containing specific immune cells (such as the autoimmune hepatitis model), without drug intervention, the multimodal data acquisition system and quantum sensing technology are used to continuously collect various data of liver organoids during the natural development of the disease, including multi-source data from the accumulation of gene mutations at the genomic level, gene expression disorders in transcriptomics, abnormal protein expression and modification in proteomics, metabolite disorders in metabolomics, and changes in biophysical properties. At the same time, the abnormal data of biomolecular activities monitored by quantum sensors are obtained, such as abnormally increased activity of inflammation-related enzymes and excessive production of free radicals; Sp2: Data integration and in-depth analysis: Use multimodal data fusion technology to integrate different types of data, use deep learning algorithms and quantum computing to deeply mine the integrated data, and build a cross-scale liver organoid mathematical model (from molecular to organ level) to analyze the intrinsic connections and interaction mechanisms between data at all levels during the development of the disease. For example, study how specific gene mutations affect protein expression and metabolic processes through signal pathways, thereby leading to changes in the structure and function of liver organoids and abnormal interactions with immune cells. Use dynamic knowledge graphs based on organoid data to associate existing clinical data and biological knowledge on liver diseases, and deeply explore the key pathological links and molecular markers of the disease. For example, determine the specific molecular bridges between the imbalance of immune cell subsets and liver organoid cell damage in the autoimmune hepatitis model, as well as the key metabolic nodes in the disease progression process; Sp3: Mechanism verification and new target discovery: Based on the data analysis results, select key molecular targets or cell behavior processes for experimental verification. For example, use gene editing technology to knock down or knock out putative key genes in liver organoids to observe whether the disease phenotype is improved; or use specific inhibitors or activators to intervene in specific signaling pathways or cellular processes to verify their role in the occurrence and development of the disease.

[0065] Through the above steps, new potential pathological mechanism-related targets or biomarkers were discovered, providing a theoretical basis for the development of new diagnostic methods and treatment strategies. At the same time, the liver organoid disease model was further improved to more accurately simulate the real disease pathological process, so as to carry out more targeted drug screening and treatment research in the future.

[0066] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprising a reference structure" do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0067] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for drug screening and disease model construction of liver organoids, characterized in that: The model construction method is as follows: Sp1: Construct an organoid-based precise research system for immune-liver disease interactions: When constructing liver organoids, introduce different types of immune cells in specific proportions and precise activation states, regulate the expression of cell-cell interaction molecules, establish an in vitro model that simulates a variety of immune-related liver diseases, develop immune cell function and phenotypic analysis technology for this co-culture system, use specific molecular probes or fluorescent labeling reagents to monitor the functional state and phenotypic changes of immune cells in real time, and deeply explore the mechanism of interaction between immune cells and liver organoids, determine key information related to the immune-liver cell axis, and provide basic data and target basis for subsequent steps; Sp2: Establish a multimodal data acquisition system: Based on the above immune-liver organoid co-culture model, collect data from all aspects, including genomics, transcriptomics, proteomics, metabolomics, and biophysical properties. Use quantum sensing technology to monitor biomolecular activities with ultra-high sensitivity and precision, and supplement it with multimodal data to obtain richer and microscopic information. Sp3: Multimodal data fusion and intelligent analysis: Develop multimodal data fusion technology to integrate data from different levels and sources, build complex mathematical models and deep learning algorithms, collaboratively analyze these data, discover key drug targets and disease mechanism associations, build a dynamic knowledge graph based on organoid data, connect data nodes at different time points and drug treatment conditions, and associate existing biological knowledge, clinical data and drug mechanism information. Through intelligent query and reasoning, predict the synergistic or antagonistic effects of new drug combinations in liver organoids, and design personalized drug screening and combination drug regimens; Sp4: Using quantum computing to process and optimize data: With the help of quantum computing's powerful parallel computing capabilities and unique quantum algorithms, the fused large-scale organoid data can be quickly processed and deeply mined to build a more accurate and comprehensive quantum model of liver disease, predict the long-term behavior and disease outcomes of liver organoids under different drug interventions, further optimize the drug screening and disease model building process, improve the accuracy and reliability of the model, and ultimately achieve liver organoid drug screening and disease model building based on data optimization.

2. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: In the above-mentioned organoid-based precise research system for immune-liver disease interactions, when specific molecular probes or fluorescent labeling reagents are used, aptamer sequences are designed and synthesized according to specific markers on the surface of immune cells or specific functional molecular structures in cells, fluorescent groups are connected to the aptamers, and the binding affinity and specificity of the aptamers to the target molecules are adjusted through chemical modification, so that the aptamers can accurately identify and label target immune cells and their function-related molecules in the complex liver organoid-immune cell co-culture system, and the fluorescence signal intensity is linearly correlated with the target molecule concentration or the cell functional state, which is convenient for accurate quantitative analysis.

3. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: In the multimodal data acquisition system, for genomic data acquisition, a single-cell whole genome amplification technology based on a microfluidic chip is used, which can efficiently and uniformly amplify the genome of a single liver organoid cell or immune cell in a nanoliter reaction system.

4. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: In the multimodal data acquisition system established above, quantum sensing technology is used to monitor the activities of biomolecules. The use of superconducting quantum interference devices (SQUIDs) requires the use of specially designed nanoscale biosensor probes, the surface of which is modified with recognition elements that can specifically bind to liver organoid-related biomolecules. The probe is reusable and its detection activity for biomolecules can be restored through specific cleaning and regeneration procedures.

5. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: In the multimodal data fusion and intelligent analysis, multimodal data fusion technology is utilized, and the mathematical model based on data integration adopts a graph neural network-based architecture. Different modal data are regarded as nodes in the graph, and the correlation between data is regarded as edges. The feature interaction and fusion between modal data are realized through the information transmission mechanism of the graph neural network.

6. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: In the multimodal data fusion and intelligent analysis, a hybrid architecture combining a generative adversarial network (GAN) and a recurrent neural network (RNN) is adopted when analyzing multimodal data using a deep learning algorithm. GAN is used for data enhancement and feature extraction. Through adversarial training of the generator and the discriminator, virtual data samples similar to the real data distribution are generated to expand the data set size and mine potential data features. RNN is used to process the temporal information of the data. Through learning and predicting the sequence data, the development trend and change law of organoids under the action of drugs are analyzed. The two work together to improve the analysis accuracy and efficiency of multimodal data.

7. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: In the multimodal data fusion and intelligent analysis, in the process of constructing a dynamic knowledge graph based on organoid data, the construction and update of the knowledge graph adopts an automated information extraction and integration system, which can automatically extract entities, relationships and attribute information related to liver organoids, immune cells and drug action mechanisms from public biomedical databases, scientific research literature and internal experimental data, integrate this information into the knowledge graph through semantic analysis and data fusion technology, and continuously improve the structure and content of the graph using knowledge reasoning algorithms to ensure that the knowledge graph always keeps the latest biomedical research results incorporated and updated, and provides timely and comprehensive knowledge support for drug screening and disease mechanism research.

8. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: The application of quantum annealing algorithm in the data processing and optimization using quantum computing requires designing a special problem encoding method based on the characteristics of liver organoid data, converting the optimization problems in drug screening and disease model construction into an Ising model form that can be processed by the quantum annealing algorithm, and controlling the state and interaction of quantum bits. The quantum tunneling effect is used to quickly search for the optimal solution to the problem. In addition, the encoding method can adapt to data inputs of different types and sizes, ensuring the effectiveness and accuracy of the quantum annealing algorithm in processing complex liver organoid data.

9. The method for drug screening and disease model construction of liver organoids according to claim 1, characterized in that: During the model construction process, distributed blockchain technology is used for data storage and management. Various data in the liver organoid experiment are encrypted and stored, and different research institutions or laboratories can share data resources securely through the blockchain network.

Images