AI algorithm for evaluating growth state of mesenchymal stem cells

Abstract

The invention relates to the technical field of stem cell culture, in particular to an AI algorithm for evaluating the growth state of mesenchymal stem cells, which comprises the following steps: acquiring mesenchymal stem cell image data, and inputting related basic information; extracting morphological characteristic data of the cells from the image data by utilizing a stem cell culture technology and a deep learning algorithm, and judging whether the cells have pollution, degeneration or abnormal morphology; based on the cell morphological characteristic data, analyzing the growth stage of the cells, and adjusting culture conditions according to the characteristics of the stage; collecting epigenetics data of stem cells, predicting cell proliferation and differentiation trends by using a deep learning model in combination with cell morphological characteristic data and cell growth stage analysis results, and adjusting culture conditions; and monitoring metabolite data in the stem cell culture solution in real time, evaluating the health state of the cells by combining the metabolite data, the cell morphological characteristic data and the cell cycle analysis result, and judging whether the problem of abnormal metabolism or decline exists or not.

Description

Technical Field

[0001] This invention relates to the field of stem cell culture technology, and in particular to an AI algorithm for assessing the growth status of mesenchymal stem cells. Background Technology

[0002] Stem cell growth and differentiation are primarily assessed by monitoring the overall state of the cells to evaluate their health and by optimizing cell growth through adjustments to culture conditions. However, existing stem cell growth regulation methods generally neglect the impact of different phases of the cell cycle on stem cell proliferation, differentiation, and decline. The cell growth curve can be divided into the latent phase, exponential growth phase, plateau phase, and decline phase. Each phase has different cell activity and metabolic requirements. The exponential growth phase has the best cell state and is the ideal phase for use; the plateau phase is when cell numbers reach saturation, and the time point from the end of the exponential phase to the beginning of the plateau phase should be selected for use. However, current culture protocols often fail to precisely adjust to the actual growth stage of the cell population. Traditional methods rely on coarse overall assessments and fail to provide personalized and phased optimization plans for each growth stage.

[0003] Current technologies lack precise monitoring and phased regulation of cell growth processes, resulting in the inability to dynamically optimize cell culture conditions at different stages. Especially at certain critical stages of cell growth, due to the complexity of the cell cycle, simply relying on traditional overall state monitoring cannot meet the refined requirements of cell growth. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides an AI algorithm for assessing the growth status of mesenchymal stem cells, aiming to improve the problem that traditional overall status monitoring cannot meet the refined needs of cell growth.

[0005] In a first aspect, the present invention provides the following technical solution: an AI algorithm for evaluating the growth status of mesenchymal stem cells, comprising the following steps:

[0006] S1. Obtain mesenchymal stem cell image data and input relevant basic information;

[0007] S2. Using stem cell culture technology and deep learning algorithms, extract morphological feature data of cells from image data to determine whether the cells are contaminated, degenerated or have abnormal morphology.

[0008] S3. Based on cell morphology data, analyze the growth stage of the cells and adjust the culture conditions according to the characteristics of the stage.

[0009] S4. Collect epigenetic data of stem cells, and combine them with cell morphology data and cell growth stage analysis results to predict cell proliferation and differentiation trends using deep learning models and adjust culture conditions accordingly.

[0010] S5. Monitor the metabolite data in the stem cell culture medium in real time, and combine the metabolite data with cell morphology data and cell cycle analysis results to assess the health status of cells and determine whether there are metabolic abnormalities or degeneration problems.

[0011] S6. Based on the above analysis results, automatically generate culture optimization suggestions and adjust the culture program through a real-time feedback mechanism to ensure that the cells are in the best growth environment.

[0012] By adopting the above technical solution, it is possible to acquire cell image data in real time and analyze its morphological characteristics, accurately determine the growth stage of the cell by combining deep learning algorithms, and adjust the culture conditions according to the characteristics of each stage, thereby meeting the different needs of stem cell growth at different growth stages. It can provide customized optimization solutions at each stage of the cell cycle, improve the proliferation and differentiation efficiency of stem cells, and avoid the growth problems caused by neglecting the cycle characteristics of traditional methods.

[0013] Preferably, acquiring mesenchymal stem cell image data includes:

[0014] Mesenchymal stem cell image data are acquired periodically using a microscope, and the image acquisition frequency is set according to experimental needs.

[0015] Adjust the parameters of the microscope equipment, including magnification, focal length, and light source intensity;

[0016] The collected image data, along with the experiment number, collection time, sample number, and other relevant information, will be stored together.

[0017] The collected image data is screened for quality, and samples that do not meet the standards are removed.

[0018] Preferably, the extracted cell morphological feature data includes:

[0019] Cells are segmented using image processing algorithms, and the contour information of each cell is extracted;

[0020] Based on the extracted morphological features, the cell area, perimeter, aspect ratio, and morphological index parameters are quantified.

[0021] Analyze the area occupied by cells on the culture dish at different time points to confirm the growth stage of the cells;

[0022] Deep learning models are used to analyze the patterns of cell morphology and identify characteristic changes in cells.

[0023] Cell morphology characteristics are preprocessed to generate morphological feature data.

[0024] Preferably, the determination of whether the cells are contaminated, degenerated, or have abnormal morphology includes:

[0025] By comparing the morphological characteristics with those of normal cells, it can be determined whether the changes in cell morphology exceed the normal range.

[0026] A classification model was used to distinguish between normal and abnormal cells, and the types of abnormal cells were identified.

[0027] By combining information on cell division status and cell membrane integrity, we can determine whether cells are contaminated or degenerated.

[0028] Preferably, the growth stage of the analyzed cells includes:

[0029] Key morphological features of cells, such as the cell nucleus division process, chromosome arrangement and morphological changes, are extracted using image analysis algorithms to identify cell cycle characteristics.

[0030] Deep learning models are used to analyze cell dynamics and the current stage is inferred by combining them with known cell cycle models.

[0031] Time series analysis was used to identify the cell growth stages.

[0032] Preferably, the adjustment of culture conditions includes:

[0033] Based on the analysis results of the cell growth stages, determine the current growth requirements of the cells;

[0034] Based on the morphological characteristics analysis results, the cell proliferation status and differentiation process can be determined;

[0035] Adjust the culture medium composition and environmental parameters to meet the needs of cells at different growth stages;

[0036] Real-time monitoring and dynamic adjustment of culture conditions.

[0037] Preferably, the epigenetic data collected from stem cells includes:

[0038] Collect stem cell gene expression data and screen epigenetic features related to cell proliferation, differentiation and cell cycle;

[0039] Bioinformatics tools were used to perform preliminary analysis of gene data to screen out genes that have a significant impact on cell growth.

[0040] By combining cell morphology characteristics and cell cycle analysis results, the effects of gene expression on cell growth were analyzed.

[0041] Preferably, the prediction of cell proliferation and differentiation trends includes:

[0042] Based on cell morphology data and epigenetic data, a predictive model for cell proliferation and differentiation was constructed.

[0043] Use deep learning or machine learning methods to analyze cell proliferation rate and differentiation direction;

[0044] By combining cell cycle, morphological characteristics, and epigenetic data, future proliferation and differentiation trends can be predicted.

[0045] We provide recommendations for adjusting culture conditions to optimize cell growth and differentiation processes.

[0046] Preferably, the assessment of cell health status includes:

[0047] By combining real-time monitoring data of metabolites, cell morphology characteristics, and cell cycle analysis results, the health status of cells can be assessed.

[0048] A cell health scoring system was established by utilizing changes in the concentration of metabolites to quantify the metabolic state of cells.

[0049] By comparing cell health data at different growth stages, metabolic abnormalities or degeneration problems can be identified.

[0050] Deep learning algorithms are used to extract key features from cell metabolic patterns, morphological characteristics, and proliferation status to predict health status.

[0051] Preferably, the adjustment of the culture program through a real-time feedback mechanism includes:

[0052] Based on the assessment results of cell health status, monitor the cell growth environment in real time;

[0053] Input metabolite data, morphological characteristic data, and cell cycle analysis results as feedback criteria;

[0054] Use control algorithms to optimize culture conditions, including culture medium composition, temperature and humidity, and oxygen concentration;

[0055] The training program is dynamically adjusted based on feedback information.

[0056] The present invention has the following beneficial effects:

[0057] 1. In this invention, by monitoring the cell cycle of stem cells in real time and dynamically adjusting cell culture conditions, the needs of different growth stages for cell growth can be accurately identified, and customized optimization schemes can be provided for each stage. By combining deep learning algorithms and time-series data analysis, changes in the cell cycle can be monitored in real time, and precise regulation can be carried out according to the needs of each stage, thereby effectively improving the proliferation and differentiation efficiency of stem cells and avoiding growth problems caused by ignoring cell cycle characteristics.

[0058] 2. In this invention, by analyzing high-throughput epigenetic data and combining it with the cell proliferation and differentiation status, a deep learning method is used to predict the future proliferation and differentiation trends of stem cells. This model can not only monitor the growth status of stem cells but also identify key epigenetic markers, thereby providing a scientific basis for adjusting culture conditions and optimizing stem cell proliferation and differentiation.

[0059] 3. In this invention, the relationship between metabolite concentration and cell proliferation, differentiation state, and health status is modeled to assess cell health in real time. Based on this real-time data, culture conditions can be dynamically adjusted to optimize the cell growth environment, prevent cell degeneration due to metabolic abnormalities, and effectively improve cell growth and differentiation efficiency. Attached Figure Description

[0060] Figure 1 This is a flowchart of an AI algorithm for evaluating the growth status of mesenchymal stem cells proposed in this invention. Detailed Implementation

[0061] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1:

[0063] In a first embodiment of the present invention, the present invention provides an AI algorithm for assessing the growth status of mesenchymal stem cells, such as... Figure 1 As shown, it includes the following steps:

[0064] S1. Obtain mesenchymal stem cell image data and input relevant basic information;

[0065] Furthermore, obtaining mesenchymal stem cell image data includes:

[0066] Mesenchymal stem cell image data are acquired periodically using a microscope, and the image acquisition frequency is set according to experimental needs.

[0067] Adjust the parameters of the microscope equipment, including magnification, focal length, and light source intensity;

[0068] The collected image data, along with the experiment number, collection time, sample number, and other relevant information, will be stored together.

[0069] The collected image data is screened for quality, and samples that do not meet the standards are removed.

[0070] Specifically, acquiring image data of mesenchymal stem cells involves multiple steps. First, image data of mesenchymal stem cells is acquired periodically using a microscope. During this process, the acquisition frequency of the microscope is set according to experimental requirements to ensure that cell image data can be obtained periodically or in real-time during the experiment. The acquisition frequency depends on the specific experimental objectives and is usually optimized based on the cyclical changes in cell growth and the time window for data acquisition. The selection of the acquisition frequency can be quantified using the following formula: ;

[0071] in, Image acquisition frequency, The time period required for the experiment, This refers to the required amount of image data. By adjusting these parameters, a suitable image data acquisition frequency can be obtained.

[0072] During image data acquisition, precise adjustments to the microscope's parameters are necessary to ensure image quality meets experimental requirements. Specifically, the microscope's magnification, focal length, and light source intensity need to be optimized based on the cell size, morphology, and the target of observation. The choice of magnification directly affects the resolution and detail of the cell image; excessively high or low magnification may lead to blurring or distortion. Adjusting the focal length ensures precise focusing of the observation area, avoiding image blurring. Adjusting the light source intensity affects the brightness and contrast of the cell image; excessively strong or weak light sources will affect detail capture. These microscope parameter adjustments not only need to consider the properties of the cells themselves but also require dynamic adjustments based on the experimental environmental conditions.

[0073] After image data acquisition, all acquired image data is stored along with the experiment number, acquisition time, sample number, and other relevant information. The storage process employs systematic management to ensure that image data is systematically categorized and archived according to different experiment numbers, sample numbers, and other information, facilitating subsequent data retrieval and analysis. Specifically, each image is tagged during storage with information including the image acquisition time, microscope parameter settings, and experimental group, ensuring the integrity and traceability of the experimental data.

[0074] After image data is stored, it needs to undergo quality screening. The purpose of quality screening is to remove image samples that do not meet the standards to ensure the accuracy of data in subsequent analysis. Image quality screening is evaluated using preset standards, mainly including indicators such as image sharpness, contrast, and noise level. Images that do not meet the standards are automatically marked and removed by the system to prevent them from interfering with subsequent analysis. Quality screening can use image processing algorithms to automatically evaluate images, and specific methods include quantitative analysis based on image contrast, sharpness, and texture features. The removal criteria can be described by the following formula: ;

[0075] in, Indicates image quality score. For image contrast, For image clarity, This represents the image noise level. (If the quality score...) If the image quality is below the set threshold, it is considered that the image quality does not meet the standard and must be rejected.

[0076] Through the above process, the acquisition, processing, and storage of mesenchymal stem cell image data can be managed efficiently and systematically, ensuring the reliability and accuracy of data for subsequent analysis.

[0077] S2. Using stem cell culture technology and deep learning algorithms, extract morphological feature data of cells from image data to determine whether the cells are contaminated, degenerated or have abnormal morphology.

[0078] Furthermore, the extraction of cellular morphological feature data includes:

[0079] Cells are segmented using image processing algorithms, and the contour information of each cell is extracted;

[0080] Based on the extracted morphological features, the cell area, perimeter, aspect ratio, and morphological index parameters are quantified.

[0081] Analyze the area occupied by cells on the culture dish at different time points to confirm the growth stage of the cells;

[0082] Deep learning models are used to analyze the patterns of cell morphology and identify characteristic changes in cells.

[0083] Cell morphology characteristics are preprocessed to generate morphological feature data.

[0084] Furthermore, determining whether cells exhibit contamination, degeneration, or abnormal morphology includes:

[0085] By comparing the morphological characteristics with those of normal cells, it can be determined whether the changes in cell morphology exceed the normal range.

[0086] A classification model was used to distinguish between normal and abnormal cells, and the types of abnormal cells were identified.

[0087] By combining information on cell division status and cell membrane integrity, we can determine whether cells are contaminated or degenerated.

[0088] Specifically, the cell image data is segmented using image processing algorithms to extract the contour information of each cell. This process is typically achieved through edge detection algorithms, which can identify cell boundaries in the image using gradient information. Image segmentation techniques can accurately distinguish the relationship between each cell and the background, thereby extracting the contour line of each cell, which provides the basic data for subsequent morphological analysis. Common segmentation algorithms include threshold-based segmentation, region growing, and deep learning-based image segmentation methods, such as the U-Net network. These methods obtain the pixel region of each cell in the image, which serves as input for subsequent processing.

[0089] After extracting cell contour information, the next step is to quantify the cell morphology based on this contour data. Specifically, this involves calculating morphological features such as cell area, perimeter, and aspect ratio to describe the cell's geometry. Cell area and perimeter are important indicators of cell size and shape, while parameters such as aspect ratio and morphological index reveal morphological characteristics, such as whether the cell exhibits a regular shape or abnormal extension or deformation. For each extracted cell, its morphological features are quantified using the following formula: ;

[0090] in, The area of ​​the cell, This represents the pixel region of the cell outline; the perimeter can be calculated using the boundary length of the outline, using the following formula: ;

[0091] in, For the perimeter, The boundary line of the cell outline. Indicates the arc length of the element on the boundary.

[0092] After quantifying the morphological characteristics of cells, deep learning models are further used to analyze the morphological data to identify patterns in cell morphological changes. Deep learning models, such as convolutional neural networks (CNNs), can automatically learn high-level representations of cell morphological features from large-scale cell morphological data. By training a CNN model, the system can identify feature patterns corresponding to different cell morphological changes and perform predictions and classifications. The core of this process is that through learning from a large amount of labeled data, the model can adaptively capture the potential patterns of cell morphological changes and identify phenomena such as cell proliferation, differentiation, contamination, or degeneration.

[0093] After preprocessing the cell morphology data, this data will serve as the foundation for analyzing and determining cell state. At this stage, the extracted morphology data undergoes standardization, noise reduction, and smoothing to ensure accuracy and consistency. This preprocessed morphology data will then be used for subsequent cell state assessments to ensure the precision of the results.

[0094] Cellular morphological data is used not only to determine cell proliferation and differentiation status, but also to identify contamination, degeneration, or abnormal morphology. By comparing the current morphological characteristics of cells with the standard morphological characteristics of normal cells, the system can detect deviations in cell morphology and determine whether they exceed the normal range. This process can employ machine learning-based classification models, such as Support Vector Machines (SVM) or deep neural networks, to distinguish between normal and abnormal cells and identify the type of abnormal cells. Abnormal cell types include contamination, degeneration, and morphological variations.

[0095] Besides morphological comparisons, cell division status and membrane integrity are also important parameters for assessing cell health. Cells exhibit unique morphological characteristics during division, such as chromosome arrangement and nucleus changes, which help determine whether a cell is in the division phase. Membrane integrity is crucial for determining whether a cell is contaminated or degenerated; cell membrane integrity can be assessed through its thickness, elasticity, and fluidity. Combining this information, the system can determine whether cells have exhibited contamination or degeneration.

[0096] By combining the above techniques, we can efficiently and accurately extract morphological features of cells from image data, detect contamination, degradation, and abnormal morphology, and provide an important basis for subsequent cell health assessment and adjustment of culture conditions.

[0097] S3. Based on cell morphology data, analyze the growth stage of the cells and adjust the culture conditions according to the characteristics of the stage.

[0098] Further analysis of the cell's growth stage includes:

[0099] Key morphological features of cells, such as the cell nucleus division process, chromosome arrangement and morphological changes, are extracted using image analysis algorithms to identify cell cycle characteristics.

[0100] Deep learning models are used to analyze cell dynamics and the current stage is inferred by combining them with known cell cycle models.

[0101] Time series analysis was used to identify the cell growth stages.

[0102] Further adjustments to the cultivation conditions include:

[0103] Based on the analysis results of the cell growth stages, determine the current growth requirements of the cells;

[0104] Based on the morphological characteristics analysis results, the cell proliferation status and differentiation process can be determined;

[0105] Adjust the culture medium composition and environmental parameters to meet the needs of cells at different growth stages;

[0106] Real-time monitoring and dynamic adjustment of culture conditions.

[0107] Specifically, key morphological features of cells are extracted using image analysis algorithms. These features include the cell division process, chromosome arrangement, and morphological changes. Each stage of the cell cycle, such as G1, S, G2, and M phases, exhibits unique morphological characteristics. For example, in the M phase, the cell division process is clearly visible, chromosomes align, and significant changes occur in the cell membrane and cytoplasm. In the S phase, DNA synthesis and replication activities within the cell nucleus intensify, manifested morphologically as a change in the nucleocytoplasmic ratio. In the G1 phase, the cell structure is relatively stable, the nucleus is relatively intact, and there are no signs of division. By utilizing these morphological features and combining them with known patterns of the cell cycle, the system can accurately extract cell cycle characteristics from image data.

[0108] Deep learning models are employed to further analyze the dynamic changes of cells. Trained on large amounts of labeled data, these models adaptively learn the criteria for determining growth stages from the evolution of cell morphology. Commonly used deep learning algorithms include image classification and regression models such as Convolutional Neural Networks (CNNs). By using extracted cell morphological features as input, the deep learning model can infer the cell's growth stage based on morphological changes. The model predicts different cell images based on the changing patterns of cell morphological features at each stage, determining which stage of the cell cycle it is in. This process is not limited to the analysis of static images but also incorporates the dynamic changes of the cell cycle, considering time-series data to further accurately predict the cell's growth stage.

[0109] After inferring the cell growth stage, time series analysis is used to further confirm the cell's growth stage. Time series analysis observes morphological changes in cells at different time points, tracks the life cycle evolution of each cell, and uses machine learning algorithms such as recurrent neural networks (RNNs) or long short-term memory networks (LSTMs) to extract dynamic features of the cell cycle from the time series data. Based on these analyses, more accurate growth stage determination can be achieved, improving the accuracy and stability of the determination. Finally, through comprehensive analysis, the current growth stage of the cell is determined.

[0110] Based on the analysis of cell growth stages, the system can determine the cells' growth needs and adjust culture conditions accordingly to meet these needs at different stages. Different stages of the cell cycle have different requirements for culture conditions. For example, in the S phase, cells have a high demand for DNA synthesis and repair, therefore sufficient nutrients for nucleic acid and protein synthesis should be provided. In the G1 phase, cells rely more on cell metabolism and energy synthesis, thus requiring a greater supply of glucose and energy sources in the culture medium. In the M phase, energy consumption during cell division increases, therefore conditions such as ion concentration and pH in the culture medium need to be adjusted to promote cell division.

[0111] By combining morphological characteristic analysis results, the system can further determine the cell's proliferation status and differentiation process. Cells in the proliferative phase require sufficient nutrients and growth factors, while cells in the differentiation phase require specific growth factors and lower levels of proliferation factors. Based on the cell's proliferation status and differentiation process, the system can dynamically adjust the composition of the culture medium to ensure that the cells are in an optimal growth environment.

[0112] Besides the composition of the culture medium, other parameters in the cell growth environment, such as temperature, humidity, and oxygen concentration, also need to be adjusted appropriately according to the cell growth stage. Cells have different environmental requirements at different growth stages. For example, cells consume more oxygen during division, thus requiring an increase in oxygen concentration, while during proliferation, cells are more sensitive to temperature, and the culture temperature may need to be adjusted to improve cell viability.

[0113] During implementation, the system monitors cell growth in real time and dynamically adjusts culture conditions based on the analysis results of each growth stage. Through sensors and a feedback mechanism, changes in cell state are fed back to the control system in real time. The system then adjusts culture conditions, such as pH, oxygen concentration, temperature, and humidity, to ensure that the cell's growth needs at different stages are met. This feedback mechanism ensures dynamic optimization of culture conditions, enabling the system to adapt to changes in cell requirements during growth, thereby improving the efficiency and quality of cell culture.

[0114] This process combines cell cycle analysis with adjustments to culture conditions to provide precise support for cell growth and differentiation, thereby ensuring the efficiency and stability of the cell culture process.

[0115] S4. Collect epigenetic data of stem cells, and combine them with cell morphology data and cell growth stage analysis results to predict cell proliferation and differentiation trends using deep learning models and adjust culture conditions accordingly.

[0116] Further, the collection of epigenetic data from stem cells includes:

[0117] Collect stem cell gene expression data and screen epigenetic features related to cell proliferation, differentiation and cell cycle;

[0118] Bioinformatics tools were used to perform preliminary analysis of gene data to screen out genes that have a significant impact on cell growth.

[0119] By combining cell morphology characteristics and cell cycle analysis results, the effects of gene expression on cell growth were analyzed.

[0120] Furthermore, predicting cell proliferation and differentiation trends includes:

[0121] Based on cell morphology data and epigenetic data, a predictive model for cell proliferation and differentiation was constructed.

[0122] Use deep learning or machine learning methods to analyze cell proliferation rate and differentiation direction;

[0123] By combining cell cycle, morphological characteristics, and epigenetic data, future proliferation and differentiation trends can be predicted.

[0124] We provide recommendations for adjusting culture conditions to optimize cell growth and differentiation processes.

[0125] Specifically, the first step is to collect epigenetic data from stem cells to provide foundational data for predicting cell proliferation and differentiation trends. This process involves acquiring stem cell gene expression data and combining it with cell morphology data and cell growth stage analysis results to construct a comprehensive predictive model. This process begins with preliminary analysis of the gene data using bioinformatics tools to screen for epigenetic features related to cell proliferation, differentiation, and cell cycle. Gene expression data can be obtained through high-throughput RNA sequencing technology to obtain gene expression profiles of stem cells under different culture conditions. Then, based on pre-defined biological hypotheses or literature, genes related to cell proliferation and differentiation are screened, such as Cyclin family genes related to proliferation and transcription factors related to differentiation. The expression levels of these genes can serve as important inputs for subsequent analyses.

[0126] After preliminary analysis, the selected gene data is combined with the results of cell morphology and growth stage analyses to further study the impact of gene expression on cell growth. Cell morphology and growth stage analyses can be performed using the aforementioned image processing and deep learning methods to obtain information such as cell morphology indicators and cell cycle progression. By combining this data with gene expression data, a multi-dimensional dataset can be formed, containing information on gene expression, cell morphology, and cell cycle status. This dataset provides sufficient biological basis for subsequent prediction of cell proliferation and differentiation trends.

[0127] Based on the aforementioned multi-dimensional data, deep learning models are used to predict cell proliferation and differentiation trends. Deep learning models play a central role in this process, accurately predicting cell proliferation rates and differentiation directions by learning the complex relationships between large-scale gene expression data, cell morphology features, and cell growth stage data. To achieve this goal, deep learning models such as Convolutional Neural Networks (CNNs) or Long Short-Term Memory Networks (LSTMs) can be employed. CNNs can extract morphological features from cell images, while LSTMs can predict the proliferation and differentiation status of cells at a future time point through time series analysis combined with cell growth stage data. Through backpropagation algorithms, the model can continuously optimize the prediction results, thereby improving the accuracy of cell proliferation and differentiation predictions.

[0128] To further improve prediction accuracy, cell growth stages, morphological characteristics, and epigenetic data can be weighted within the model and fused according to their different levels of importance. Through this multi-level data fusion, deep learning models can more accurately capture the growth needs of cells at different stages. For example, some genes may have high expression levels during cell division but decrease during differentiation; therefore, the model needs to dynamically adjust gene expression according to different stages.

[0129] After the model completes training and generates predictions of cell proliferation and differentiation trends, the system can generate targeted recommendations for adjusting culture conditions based on these predictions. These recommendations primarily include adjustments to the culture medium composition, changes in temperature and humidity, and regulation of oxygen concentration. For example, when cells are in the proliferation phase, growth factors or other substances that promote cell division can be appropriately increased in the culture medium; while when cells are in the differentiation phase, it may be necessary to adjust the composition of the culture medium, reducing the concentration of certain growth factors and providing specific factors that promote differentiation. By monitoring changes in cell state in real time, the system can dynamically adjust these culture conditions to ensure that cells grow and differentiate in an optimal environment.

[0130] In this process, quantifying the adjustments to culture conditions using formulas is also crucial. For example, the concentration of a specific factor in the culture medium, such as the concentration of a growth factor, can be calculated using the following formula:

[0131] in, This is the adjusted factor concentration. This is the current factor concentration. It refers to the required concentration increment or reduction based on the predicted results of cell proliferation and differentiation trends.

[0132] Furthermore, cell proliferation rate and differentiation tendency can be characterized by the following prediction formula: ;

[0133] in, It is the cell proliferation rate. and These are the predicted cell counts at the beginning and end of the cycle, respectively. and These represent the start and end points of the cell cycle, respectively. For predicting the direction of differentiation, a similar formula can be used for quantitative assessment by analyzing changes in cell surface markers.

[0134] In practical applications, this series of deep learning analyses and model predictions can not only accurately predict cell growth and differentiation trends, but also adjust the cell culture environment in real time based on the prediction results, optimize culture conditions, and thus improve the efficiency and quality of cell culture.

[0135] S5. Monitor the metabolite data in the stem cell culture medium in real time, and combine the metabolite data with cell morphology data and cell cycle analysis results to assess the health status of cells and determine whether there are metabolic abnormalities or degeneration problems.

[0136] Further assessment of cellular health includes:

[0137] By combining real-time monitoring data of metabolites, cell morphology characteristics, and cell cycle analysis results, the health status of cells can be assessed.

[0138] A cell health scoring system was established by utilizing changes in the concentration of metabolites to quantify the metabolic state of cells.

[0139] By comparing cell health data at different growth stages, metabolic abnormalities or degeneration problems can be identified.

[0140] Deep learning algorithms are used to extract key features from cell metabolic patterns, morphological characteristics, and proliferation status to predict health status.

[0141] Specifically, the health status of cells is assessed by real-time monitoring of metabolite data in the stem cell culture medium. To achieve this, the system continuously collects concentration data of key metabolites in the stem cell culture medium, such as lactate, glucose, amino acids, and ATP, using sensors or metabolite monitoring devices. Changes in these metabolite concentrations reflect cellular metabolic activity, thus providing important evidence for assessing cell health. By monitoring real-time changes in metabolite concentrations, metabolic characteristics of cells during proliferation, differentiation, or dormancy can be identified, allowing for the timely detection of metabolic abnormalities or signs of cell degeneration.

[0142] Building upon this foundation, and combining the aforementioned cell morphology data and cell cycle analysis results, a more comprehensive assessment of cell health status is achieved. Cell morphology features, such as cell area, perimeter, and morphological indices, are extracted using image processing techniques and deep learning models. The cell cycle is derived through dynamic analysis of cell cycle-related features, such as DNA synthesis and division phases. By combining this data, the system can construct a multi-dimensional cell health status assessment model. In this process, the interrelationships between cell morphology features, cell cycle data, and metabolite concentration data are crucial. For example, when cell proliferation or differentiation changes, the concentration of metabolites changes accordingly, and the cell morphology also exhibits corresponding characteristics. Through comprehensive analysis of this data, the system can make a more accurate judgment on cell health status.

[0143] To quantify the metabolic state of cells, the system establishes a cell health scoring system based on changes in metabolite concentrations. This scoring system relies on the temporal changes in metabolite concentrations, quantifying cell health by setting thresholds and reference standards. For example, a scoring formula based on changes in metabolite concentrations can be defined:

[0144] in, Score cell health. , , , These represent the concentrations of lactic acid, glucose, ATP, and amino acids, respectively. , , , These are the weighting coefficients for the impact of each metabolite on cell health. The weighting coefficients can be set based on experimental or historical data through training and optimization, enabling the scoring system to reflect the actual health level of cells under different conditions.

[0145] By comparing cell health data at different growth stages, the system can identify metabolic abnormalities or cell decline. For example, the concentration of cellular metabolites should exhibit certain regularities at different growth stages. When metabolite concentrations fluctuate abnormally or deviate significantly from normal reference values, the system will detect signs of metabolic abnormalities using a health scoring model, indicating potential cell decline or other health problems. In this case, the system can alert operators to intervene, such as adjusting culture conditions, updating the culture medium formulation, or changing the culture temperature and humidity, thereby optimizing the cell growth environment and preventing metabolic imbalances or cell decline.

[0146] To further improve the accuracy of health status prediction, the system uses deep learning algorithms to extract key features from cell metabolic patterns, morphological characteristics, and proliferation status, and combines these features to predict cell health status. Deep learning models can discover complex patterns and relationships in massive datasets, further optimizing health status prediction by comparing metabolic characteristics across different time periods and cell growth stages. For example, recurrent neural networks (RNNs) or long short-term memory networks (LSTMs) can be used to process time-series data on metabolite concentrations, analyze their trends, and combine this with data on cell morphology and growth stage for a comprehensive assessment of health status. By training deep learning models, the system can automatically identify changes in health status and issue early warnings, helping to improve cell quality control during stem cell culture.

[0147] In practical applications, the system can continuously monitor and analyze the combination of metabolite data, cell morphology characteristics, and cell cycle data to assess the health status of stem cells in real time. When problems are detected, timely measures can be taken to make adjustments, thereby ensuring the optimization of the cell culture environment, improving the culture effect of stem cells, and reducing the occurrence of metabolic abnormalities and decline.

[0148] S6. Based on the above analysis results, automatically generate culture optimization suggestions and adjust the culture program through a real-time feedback mechanism to ensure that the cells are in the best growth environment.

[0149] Furthermore, adjusting the training program through a real-time feedback mechanism includes:

[0150] Based on the assessment results of cell health status, monitor the cell growth environment in real time;

[0151] Input metabolite data, morphological characteristic data, and cell cycle analysis results as feedback criteria;

[0152] Use control algorithms to optimize culture conditions, including culture medium composition, temperature and humidity, and oxygen concentration;

[0153] The training program is dynamically adjusted based on feedback information.

[0154] Specifically, by analyzing cell health status, metabolite data, cell morphology data, and cell cycle analysis results, the system generates culture optimization suggestions and dynamically adjusts the culture protocol based on a real-time feedback mechanism to ensure cells are in the optimal growth environment. First, the system acquires key information about cell culture status by monitoring various parameters of the cell growth environment in real time. The monitored environmental parameters include temperature, humidity, oxygen concentration, carbon dioxide concentration, and pH value, all of which directly affect cell growth and proliferation. By combining this environmental data with metabolite data, cell morphology data, and cell cycle data, the system can assess cell health status in real time.

[0155] The assessment of cell health status is based on previous analyses of metabolite concentrations, morphological characteristics, and cell cycle data. These data provide essential feedback for adjusting culture protocols. Specifically, metabolite data, such as lactate, glucose, and ATP concentrations, reflect cellular metabolic activity and physiological state; morphological data provide information on changes in cell morphology during growth; and cell cycle data reveals the cell's proliferative state and life cycle stage. By integrating this data, the system can comprehensively analyze the current state of cells and identify culture environment parameters that may require adjustment.

[0156] Based on the health status assessment results, the system uses a control algorithm to optimize culture conditions. This control algorithm, based on machine learning or deep learning models, combines multi-dimensional input data, including metabolite data, morphological feature data, and cell cycle analysis results, to adjust various culture environment parameters. For example, if the system detects that cell proliferation is too slow, it may promote cell growth by increasing the concentration of nutrients in the culture medium or optimizing gas concentrations, such as increasing oxygen concentration. If the system identifies an abnormally high concentration of metabolites, such as excessive lactic acid, it may automatically adjust the composition of the culture medium or the ambient temperature to reduce the metabolic stress on the cells.

[0157] In this feedback mechanism, adjusted culture conditions take effect immediately. The system continuously monitors changes in cell status and dynamically evaluates the effectiveness of adjustments based on real-time feedback data. When cell health improves or stabilizes, the system updates its culture protocol and continues operating. If cell status does not improve effectively, the system further adjusts parameters to ensure the culture environment always meets the cells' needs. For example, if cell proliferation fails to meet expectations, the system may reassess the temperature, pH, and culture medium formulation and make appropriate modifications. Through this cyclical feedback process, the system ensures that cells are always in an optimal growth environment.

[0158] To achieve this process, the system can employ automatic control algorithms, including PID control, to adjust the cultivation parameters based on feedback from real-time data. The formula can be expressed as: ;

[0159] in, This indicates the adjustment amount of the control parameter. , , These are the gain parameters for the PID controller. This includes current errors such as temperature and pH errors. The control algorithm automatically adjusts the culture environment parameters by optimizing the regulation of these parameters, ensuring that the needs of the cells are fully met at each stage.

[0160] The entire process continuously optimizes the cell culture environment and improves cell growth and differentiation efficiency through real-time data feedback and intelligent algorithm adjustments. Ultimately, through continuous optimization and feedback mechanisms, the system can not only make real-time adjustments during cell culture but also generate personalized culture optimization suggestions based on the health status of the cells, thereby ensuring that the cells are under optimal growth conditions and achieving the goal of improving cell culture efficiency and stability.

[0161] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AI algorithm for assessing the growth status of mesenchymal stem cells, characterized in that, Includes the following steps: S1. Obtain mesenchymal stem cell image data and input relevant basic information; S2. Using stem cell culture technology and deep learning algorithms, extract morphological feature data of cells from image data to determine whether the cells are contaminated, degenerated or have abnormal morphology. S3. Based on cell morphology data, analyze the growth stage of the cells and adjust the culture conditions according to the characteristics of the stage. S4. Collect epigenetic data of stem cells, and combine them with cell morphology data and cell growth stage analysis results to predict cell proliferation and differentiation trends using deep learning models and adjust culture conditions accordingly. S5. Monitor the metabolite data in the stem cell culture medium in real time, and combine the metabolite data with cell morphology data and cell cycle analysis results to assess the cell health status and determine whether there are metabolic abnormalities or degeneration problems. S6. Based on the above analysis results, automatically generate culture optimization suggestions and adjust the culture program through a real-time feedback mechanism to ensure that the cells are in the best growth environment.

2. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The acquisition of mesenchymal stem cell image data includes: Mesenchymal stem cell image data are acquired periodically using a microscope, and the image acquisition frequency is set according to experimental needs. Adjust the parameters of the microscope equipment, including magnification, focal length, and light source intensity; The collected image data, along with the experiment number, collection time, sample number, and other relevant information, will be stored together. The collected image data is screened for quality, and samples that do not meet the standards are removed.

3. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The extracted cell morphological feature data includes: Cells are segmented using image processing algorithms, and the contour information of each cell is extracted; Based on the extracted morphological features, the cell area, perimeter, aspect ratio, and morphological index parameters are quantified. Analyze the area occupied by cells on the culture dish at different time points to confirm the growth stage of the cells; Deep learning models are used to analyze the patterns of cell morphology and identify characteristic changes in cells. Cell morphology characteristics are preprocessed to generate morphological feature data.

4. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The determination of whether cells are contaminated, degenerated, or have abnormal morphology includes: By comparing the morphological characteristics with those of normal cells, it can be determined whether the changes in cell morphology exceed the normal range. A classification model was used to distinguish between normal and abnormal cells, and the types of abnormal cells were identified. By combining information on cell division status and cell membrane integrity, we can determine whether cells are contaminated or degenerated.

5. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The growth stages of the analyzed cells include: Key morphological features of cells, such as the cell nucleus division process, chromosome arrangement and morphological changes, are extracted using image analysis algorithms to identify cell cycle characteristics. Deep learning models are used to analyze cell dynamics and combined with known cell cycle models to infer the current stage. Time series analysis was used to identify the cell growth stages.

6. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The adjustment of culture conditions includes: Based on the analysis results of the cell growth stages, determine the current growth requirements of the cells; Based on the morphological characteristics analysis results, the cell proliferation status and differentiation process can be determined; Adjust the culture medium composition and environmental parameters to meet the needs of cells at different growth stages; Real-time monitoring and dynamic adjustment of culture conditions.

7. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The epigenetic data collected from the stem cells include: Collect stem cell gene expression data and screen epigenetic features related to cell proliferation, differentiation and cell cycle; Bioinformatics tools were used to perform preliminary analysis of gene data to screen out genes that have a significant impact on cell growth. By combining cell morphology characteristics and cell cycle analysis results, the effects of gene expression on cell growth were analyzed.

8. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The predicted cell proliferation and differentiation trends include: Based on cell morphology data and epigenetic data, a predictive model for cell proliferation and differentiation was constructed. Use deep learning or machine learning methods to analyze cell proliferation rate and differentiation direction; By combining cell cycle, morphological characteristics, and epigenetic data, future proliferation and differentiation trends can be predicted. We provide recommendations for adjusting culture conditions to optimize cell growth and differentiation processes.

9. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The assessment of cell health status includes: By combining real-time monitoring data of metabolites, cell morphology characteristics, and cell cycle analysis results, the health status of cells can be assessed. A cell health scoring system was established by utilizing changes in the concentration of metabolites to quantify the metabolic state of cells. By comparing cell health data at different growth stages, metabolic abnormalities or degeneration problems can be identified. Deep learning algorithms are used to extract key features from cell metabolic patterns, morphological characteristics, and proliferation status to predict health status.

10. The AI ​​algorithm for assessing the growth status of mesenchymal stem cells according to claim 1, characterized in that, The adjustment of the culture program through a real-time feedback mechanism includes: Based on the assessment results of cell health status, monitor the cell growth environment in real time; Input metabolite data, morphological characteristic data, and cell cycle analysis results as feedback criteria; Use control algorithms to optimize culture conditions, including culture medium composition, temperature and humidity, and oxygen concentration; The training program is dynamically adjusted based on feedback information.

Images