Data analysis method and system for stem cell culture

Abstract

The invention discloses a data analysis method and system for stem cell culture, and relates to the technical field of stem cells, and the system performs real-time feature extraction and high-dimensional feature space mapping through a high-precision data acquisition module in combination with various cell growth feature data. And through a nonlinear mapping function phi in the feature analysis module, the complex biological feature data are converted into a comprehensive feature vector Fcom, and then a growth prediction model is constructed. The model is based on a historical comprehensive feature vector Fcom, an integrated learning method is utilized, a multiple decision tree model is constructed, and the nonlinear relation in the stem cell growth process is effectively captured. Through the precise prediction mechanism, researchers can predict the growth dynamic state of the stem cells at the future moment in advance, a scientific basis is provided for supporting decision and adjustment in stem cell culture, and the prediction precision and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the field of stem cell technology, and in particular to a data analysis method and system for stem cell culture. Background Art

[0002] Stem cell research is an important branch of biomedical engineering, especially in regenerative medicine and tissue engineering, where stem cell culture technology is widely used. Stem cells have the ability to self-renew and multidirectionally differentiate, and are important materials for studying disease model construction, drug screening, and tissue repair. In this field, the monitoring of stem cell growth patterns and culture environments is crucial, directly affecting the culture effect and the reliability of research results. Therefore, how to effectively analyze and predict the growth pattern of stem cells has become one of the core issues in stem cell culture technology.

[0003] At present, stem cell culture technology mainly relies on static monitoring and manual experience guidance. This method has many shortcomings. First, data collection during the culture process mainly relies on regular manual recording, which leads to insufficient real-time performance and difficulty in detecting abnormal situations in a timely manner. Secondly, existing analysis methods are often based on a single factor or a simple statistical model, ignoring the complexity of the stem cell growth environment and the interaction of multiple factors, resulting in inaccurate prediction results. In actual operation, this deficiency may lead to abnormal stem cell growth, such as overgrowth or growth stagnation, thus affecting the repeatability and reliability of the experiment. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a data analysis method and system for stem cell culture, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: including a data acquisition module, a feature analysis module, a stem cell growth prediction module, an adaptive and real-time prediction module and an iterative optimization module;

[0006] The data acquisition module is used to collect growth data of stem cells, transmit the growth data to the central processing center, and perform feature extraction to obtain a growth feature set;

[0007] The feature analysis module is used to define a nonlinear mapping function Φ, and map the original feature space of the growth feature set to a high-dimensional feature space, and perform combined calculations to obtain a comprehensive feature vector Fcom;

[0008] The stem cell growth prediction module constructs a growth prediction model based on an integrated learning method by integrating the feature vector Fcom, and sets a prediction model formula to calculate and obtain a growth prediction modulus value G(t+n);

[0009] The adaptive and real-time prediction module constructs an adaptive adjustment formula to dynamically adjust the dynamic weight γ of the growth prediction model formula. i (t+1), and then calculate the actual comprehensive feature vector F according to the real-time extracted growth feature set com In real-time, the growth prediction model is used to calculate the actual prediction value G. real-time The first growth threshold G1 and the second growth threshold G2 are preset and the actual prediction value G real-time Conduct preliminary comparative assessments and generate preliminary labeling suggestions;

[0010] The iterative optimization module is used to receive the label prompts generated by the preliminary evaluation in real time, adjust and optimize the growth environment of the stem cells according to the label prompts, and re-execute the system calculation after optimization to obtain the optimized prediction value G opt , and then compared with the standard prediction value G actual , perform error calculation to obtain the prediction error value MSE, and preset the optimization threshold E for secondary comparative evaluation.

[0011] Preferably, the data acquisition module includes a data acquisition unit, a data transmission unit and a feature extraction unit;

[0012] The data acquisition unit is used to install stem cell detection equipment around the stem cell culture dish to collect stem cell growth data in real time; the stem cell detection equipment includes a microelectrode, an AFM probe, a micro pH sensor, an enzyme-linked immunosorbent assay, fluorescence resonance energy transfer and an optical microscope;

[0013] The growth data include the stem cell internal potential value Vin, the stem cell external potential value Vout, the probe force response value Fpr, the extracellular matrix contact area Aecm, the pH value ph, the cell volume Vtj and the signal molecule concentration Csi;

[0014] The stem cell internal potential value Vin and the stem cell external potential value Vout are located by using a microscope and inserting a microelectrode into the stem cell, while another motor is placed in a culture dish to record the stem cell internal and external potential value data;

[0015] The probe force response value Fpr and the extracellular matrix contact area Aecm are obtained by gently contacting the ECM surface with the AFM probe, applying force and recording the force response of the probe at different positions;

[0016] The pH value ph measures the pH value of different regions of the stem cells through a micro pH sensor;

[0017] The cell volume Vtj and signal molecule concentration Csi are detected by enzyme-linked immunosorbent assay to measure the concentration of cytokines or other secretions, and the physical interaction of intercellular signal molecules is detected by fluorescence resonance energy transfer, and then the spatial distribution and intensity of intercellular signals are observed in real time using a spectral microscope;

[0018] The data transmission unit is used to send the collected growth data to the central processing center through a wireless transmission mechanism;

[0019] The feature extraction unit is used to extract the growth data in the central processing center in real time, and extract the features of the growth data to obtain a growth feature set, which includes the potential difference ΔV between the inside and outside of the cell. cell , extracellular matrix tension T ECM , pH gradient and intercellular communication signal strength S comm ;

[0020] The potential difference between the inside and outside of the cell is △V cell The specific extraction method △V is obtained by performing difference processing on the stem cell internal potential value Vin and the stem cell external potential value Vou in the growth data. cell =Vin-Vout;

[0021] The extracellular matrix tension T ECM The feature extraction is performed by extracting the needle force response value Fpr and the extracellular matrix contact area Aecm in the growth data. The specific extraction method is

[0022] The pH gradient The pH values ​​of different regions of stem cells obtained in the growth data are obtained by partial derivative calculation. The specific extraction method is In the formula, x represents the horizontal coordinate of the stem cell, and y represents the vertical coordinate of the stem cell. represents partial derivative;

[0023] The intercellular communication signal strength S comm By extracting the cell volume Vtj and signal molecule concentration Csi obtained from the growth data, the specific extraction method is

[0024] Preferably, the feature analysis module includes a mapping function construction unit and a comprehensive feature analysis unit;

[0025] The mapping function construction unit is used to construct a nonlinear mapping function Φ using a radial basis function RBF kernel to map the acquired growth feature set to a high-dimensional feature space;

[0026] The nonlinear mapping function Φ is mapped by the following algorithm formula;

[0027]

[0028] The comprehensive feature analysis unit is used to construct a comprehensive feature vector algorithm formula according to the mapped growth characteristics, and perform calculation to obtain the comprehensive feature vector Fcom;

[0029] The comprehensive feature vector Fcom is calculated and obtained by the following algorithm formula:

[0030]

[0031] In the formula, a1 represents the potential difference between the inside and outside of the cell △V cell and extracellular matrix tension T ECM The weight value of the interaction term, a2 represents the PH gradient and intercellular communication signal strength S comm The weight value of the interaction term, β represents the weight value of the sum of the square terms of all related features in the growth feature set; the specific value is set by the user;

[0032] a1·(ΔV cell ×T ECM ) represents the potential difference between the inside and outside of the binding cell △V cell and extracellular matrix tension T ECM , reflecting the effect of the interaction between the potential difference between the inside and outside of the cell membrane and the mechanical force on cell behavior;

[0033] Indicates the combination of pH gradient and intercellular communication signal strength S comm , reflecting the effects of microenvironmental chemical and signaling regulation on cell metabolism and group behavior;

[0034] It represents the comprehensive analysis of the nonlinear effects of the parameters in the growth feature set, reflects the contribution of each parameter to the system through the square term, and captures its potential nonlinear relationship.

[0035] Preferably, the stem cell growth prediction module adopts an integrated learning method and uses the comprehensive feature vector Fcom to construct a growth prediction model, and sets a prediction model formula to obtain a growth prediction model value G(t+n), uses the historical comprehensive feature vector Fcom as an input item of the prediction model, trains multiple decision tree models, and each tree f i The relationship between the historical comprehensive feature vector Fcom and the target output is learned by minimizing the loss function;

[0036] The growth prediction modulus G(t+n) is obtained by the following prediction model formula;

[0037]

[0038] In the formula, G(t+n) represents the predicted cell growth value at the t+nth moment, f i represents the nonlinear function based on the comprehensive eigenvector Fcom, γ i represents the prediction model weight, and m represents the nonlinear function f used in the prediction model. i The total number of, i represents the nonlinear function f of the i-th comprehensive eigenvector Fcom i .

[0039] Preferably, the adaptive and real-time prediction module includes an adaptive adjustment unit, a real-time prediction unit and a prediction evaluation unit;

[0040] The adaptive adjustment unit is used to construct an adaptive adjustment formula to dynamically adjust the dynamic weight γ at different time points in the prediction model formula. i (t+1);

[0041] The dynamic weight γ i (t+1) is calculated and obtained by the following adaptive adjustment formula;

[0042]

[0043] In the formula, γ i (t+1) represents the dynamic weight after updating at time t+1, reflecting the nonlinear function f of the sub-model i The importance at the new time step, γ i (t) represents the weight of at time t, represents the partial derivative, η represents the learning rate, and L represents the loss function.

[0044] Preferably, the real-time prediction unit is used to input the growth feature set extracted in real time into the comprehensive feature vector algorithm formula after dynamically adjusting the prediction model, and calculate to obtain the actual comprehensive feature vector F com real-time, and then the actual comprehensive feature vector F com The real-time input is used in the adaptively adjusted prediction model to calculate the actual prediction value G. real-time ;

[0045] The actual predicted value G real-time The value is calculated by the following algorithm formula;

[0046]

[0047] In the formula, G real-time(t+n) represents the actual predicted value of stem cells when the predicted time point is n time steps after the current time;

[0048] The prediction evaluation unit presets the first growth threshold G1 and the second growth threshold G2 based on the experimental goal by the user, and then compares them with the actual prediction value G1 obtained. real-time Conduct a preliminary comparative evaluation to analyze the current stem cell growth and label the evaluation results. The specific evaluation contents are as follows;

[0049] When the actual predicted value G real-time >When the first growth threshold G1 is reached, abnormal stem cell growth is predicted, an "overgrowth" label is generated, and an optimization mechanism is executed to adjust the prediction model;

[0050] When the second growth threshold G2 ≤ ​​actual prediction value G real-time When ≤ the first growth threshold G1, the stem cell growth is predicted to be normal, and a “normal growth” label is generated at this time, and no optimization mechanism needs to be executed;

[0051] When the actual predicted value G real-time When the value is less than the first growth threshold G2, it is predicted that the stem cell growth is abnormal, and a "growth arrest" label is generated at this time, and an optimization mechanism is executed to adjust the prediction model.

[0052] Preferably, the iterative optimization module includes an optimization unit, an error analysis unit and an error evaluation unit;

[0053] The optimization unit is used to receive the labels generated by the preliminary evaluation in real time, and generate an optimization report based on the labels to adjust and optimize the stem cell growth environment. The specific generated content is as follows;

[0054] When receiving the "overgrowth" tag, a first optimization report is generated, prompting a reverse adjustment to optimize the nutrient concentration and stimulation factor in the stem cell culture dish, wherein the reverse adjustment is to reduce the nutrient concentration and stimulation factor in the culture dish;

[0055] When receiving a "normal growth" label, there is no need to generate an optimization report;

[0056] When the "growth stagnation" label is received, a second optimization report is generated, prompting positive adjustment to optimize the nutrient concentration and stimulation factors in the stem cell culture dish, where the positive adjustment is to increase the nutrient concentration and stimulation factors in the culture dish.

[0057] Preferably, the error analysis unit is used to perform secondary data collection after optimizing the stem cell growth environment and execute the prediction model to output the optimized prediction value G opt , based on the first growth threshold G1 and the second growth threshold G2 interval, the user defines the standard prediction value Gactual , perform mean error calculation to obtain the prediction error value MSE;

[0058] The prediction error value MSE is calculated by the following algorithm formula:

[0059]

[0060] In the formula, N represents the number of samples, which represents the optimized prediction value G collected within a certain period of time. opt (t+n) and the actual predicted value G real-time (t+n) corresponds to the number of data points.

[0061] Preferably, the error evaluation unit is used to perform secondary comparative evaluation on the preset optimization threshold E and the prediction error value MSE, and analyze the growth data of the optimized stem cells. The specific evaluation contents are as follows;

[0062] When the prediction error value MSE>optimization threshold E, it means that the optimized stem cell growth prediction value is normal. At this time, the optimization unit is iteratively executed to perform prompt optimization;

[0063] When the prediction error value MSE ≤ the optimization threshold E, it means that the optimized stem cell growth prediction value is normal.

[0064] A data analysis method for stem cell culture, comprising the following steps:

[0065] S1, collecting growth data of stem cells, transmitting the growth data to a central processing center, and performing feature extraction to obtain a growth feature set;

[0066] S2, define a nonlinear mapping function Φ, and map the original feature space of the growth feature set to a high-dimensional feature space, and perform combined calculations to obtain a comprehensive feature vector Fcom;

[0067] S3. By using the comprehensive feature vector Fcom, a growth prediction model is constructed based on an integrated learning method, and a prediction model formula is set to calculate and obtain a growth prediction modulus G(t+n);

[0068] S4. Construct an adaptive adjustment formula to dynamically adjust the dynamic weight γ of the growth prediction model formula i (t+1), and then calculate the actual comprehensive feature vector F according to the real-time extracted growth feature set com real-time, substitute into the growth prediction model to calculate and obtain the actual prediction value G real-time The first growth threshold G1 and the second growth threshold G2 are preset and the actual prediction value G real-time Conduct preliminary comparative assessments and generate preliminary labeling suggestions;

[0069] S5. Receive the label prompts generated by the preliminary evaluation in real time, adjust and optimize the growth environment of stem cells according to the label prompts, and re-execute the system calculation after optimization to obtain the optimized prediction value G opt , and then compared with the standard prediction value G actual , perform error calculation to obtain the prediction error value MSE, and preset the optimization threshold E for secondary comparative evaluation.

[0070] The present invention provides a data analysis method and system for stem cell culture, which has the following beneficial effects:

[0071] (1) The system uses a high-precision data acquisition module to combine multiple cell growth feature data to perform real-time feature extraction and high-dimensional feature space mapping. Through the nonlinear mapping function Φ in the feature analysis module, these complex biological feature data are converted into a comprehensive feature vector Fcom, and then a growth prediction model is constructed. Based on the historical comprehensive feature vector Fcom, the model uses an integrated learning method to construct a multiple decision tree model, which effectively captures the nonlinear relationship in the stem cell growth process. Through this precise prediction mechanism, researchers can foresee the growth dynamics of stem cells in the future, provide a scientific basis to support decision-making and adjustments in stem cell culture, and improve the precision and accuracy of predictions.

[0072] (2) The adaptive and real-time prediction module of the system can dynamically adjust the model weights at different time points after the prediction model is built to ensure that the model can adapt to changes in the stem cell growth environment. Through the real-time collected growth data, the system can calculate the actual comprehensive feature vector F com real-time, and substitute it into the adaptively adjusted prediction model to obtain the real-time actual prediction value G real-time The system sets the first growth threshold G1 and the second growth threshold G2. real-time By making a preliminary comparison with the first growth threshold G1 and the second growth threshold G2, the system can instantly determine the growth status of stem cells and generate corresponding prompt labels. This real-time adaptive adjustment and instant response mechanism ensures that the system can make timely adjustments based on actual growth conditions to maintain the stability and health of stem cell growth.

[0073] (3) The iterative optimization module of the system can not only receive the label prompts of the preliminary evaluation, but also generate targeted optimization reports based on these prompts to guide the adjustment of the stem cell growth environment. For example, when the "overgrowth" label is detected, the system will recommend reverse adjustment of the nutrient concentration and stimulating factors in the culture dish to reduce the risk of overgrowth of stem cells; for the "growth stagnation" situation, the system will recommend increasing the concentration of nutrients and stimulating factors to promote cell growth. The optimized growth data is again calculated by the system to generate the optimized prediction value Gopt , and the standard predicted value G actual The system compares the results and calculates the predicted error value MSE, which is then compared and evaluated again with the preset optimization threshold E. Through this sophisticated iterative optimization and error evaluation mechanism, the system ensures that the growth conditions of stem cells continue to approach the optimal state, achieving precise regulation of the stem cell culture process. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is a schematic diagram of the flow of a data analysis system for stem cell culture according to the present invention;

[0075] Figure 2 This is a schematic diagram of the steps of a data analysis method for stem cell culture according to the present invention. DETAILED DESCRIPTION

[0076] 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.

[0077] Example 1

[0078] See also Figure 1 , the present invention provides a data analysis system for stem cell culture. To achieve the above purpose, the present invention is implemented through the following technical solutions: including a data acquisition module, a feature analysis module, a stem cell growth prediction module, an adaptive and real-time prediction module and an iterative optimization module;

[0079] The data acquisition module is used to collect the growth data of stem cells, transmit the growth data to the central processing center, and perform feature extraction to obtain a growth feature set;

[0080] The feature analysis module is used to define the nonlinear mapping function Φ, map the original feature space of the growth feature set to the high-dimensional feature space, and perform combined calculations to obtain the comprehensive feature vector Fcom;

[0081] The stem cell growth prediction module constructs a growth prediction model based on the integrated learning method by integrating the feature vector Fcom, and sets the prediction model formula to calculate and obtain the growth prediction modulus value G(t+n);

[0082] The adaptive and real-time prediction module constructs an adaptive adjustment formula to dynamically adjust the dynamic weight γ of the growth prediction model formula i (t+1), and then calculate the actual comprehensive feature vector F according to the real-time extracted growth feature set comIn real-time, the growth prediction model is used to calculate the actual prediction value G. real-time The first growth threshold G1 and the second growth threshold G2 are preset and the actual prediction value G real-time Conduct preliminary comparative assessments and generate preliminary labeling suggestions;

[0083] The iterative optimization module is used to receive the label prompts generated by the preliminary evaluation in real time, adjust and optimize the growth environment of stem cells according to the label prompts, and re-execute the system calculation after optimization to obtain the optimized prediction value G opt , and then compared with the standard prediction value G actual , perform error calculation to obtain the prediction error value MSE, and preset the optimization threshold E for secondary comparative evaluation.

[0084] In this embodiment, the system accurately collects stem cell growth data through the data acquisition module, and transmits it to the central processing center for real-time processing and feature extraction to form a growth feature set. The feature analysis module further performs high-dimensional mapping on these feature data, and uses the nonlinear mapping function Φ to generate a comprehensive feature vector Fcom, which provides basic data support for subsequent growth predictions. Through these steps, the system can identify and analyze potential changing trends in the growth process of stem cells with higher accuracy. The stem cell growth prediction module uses an integrated learning method to construct a growth prediction model based on the comprehensive feature vector Fcom. By setting the prediction model formula, the system can foresee the growth of stem cells in the future. The advantage of this prediction mechanism is that it can not only combine historical data, but also dynamically adapt to the growth characteristics of stem cells under different environmental conditions. The adaptive and real-time prediction module dynamically adjusts the dynamic weight γ of the model. i (t+1), ensuring that the model maintains efficient prediction capabilities under variable experimental conditions. In addition, the system presets the first growth threshold G1 and the second growth threshold G2 to make the real-time evaluation of the growth status of stem cells more accurate, thereby achieving timely intervention in abnormal growth and ensuring the continuity and stability of the experiment. This system provides real-time optimization and error evaluation capabilities for the stem cell growth environment through an iterative optimization module. The module adjusts the culture conditions in real time based on the label prompts generated by the preliminary evaluation, and the error comparison after optimization further verifies the effectiveness of the adjustment measures.

[0085] Example 2

[0086] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data acquisition module includes a data acquisition unit, a data transmission unit and a feature extraction unit;

[0087] The data acquisition unit is used to install stem cell detection equipment around the stem cell culture dish to collect stem cell growth data in real time; the stem cell detection equipment includes microelectrodes, AFM probes, micro pH sensors, enzyme-linked immunosorbent assays, fluorescence resonance energy transfer and optical microscopes;

[0088] Growth data include stem cell internal potential value Vin, stem cell external potential value Vout, probe force response value Fpr, extracellular matrix contact area Aecm, pH value ph, cell volume Vtj and signal molecule concentration Csi;

[0089] The stem cell internal potential value Vin and the stem cell external potential value Vout are located by using a microscope and inserting a microelectrode into the stem cell. Meanwhile, another motor is placed in a culture dish to record the internal and external potential value data of the stem cell.

[0090] The probe force response value Fpr and the extracellular matrix contact area Aecm were obtained by gently touching the ECM surface with the AFM probe and recording the force response of the probe at different positions after applying force;

[0091] pH value: pH value of different regions of stem cells is measured by micro pH sensor;

[0092] The cell volume Vtj and signal molecule concentration Csi were measured by enzyme-linked immunosorbent assay to detect the concentration of cytokines or other secretions, while the physical interaction of intercellular signal molecules was detected by fluorescence resonance energy transfer, and the spatial distribution and intensity of intercellular signals were observed in real time using a spectral microscope;

[0093] The data transmission unit is used to send the collected growth data to the central processing center through a wireless transmission mechanism;

[0094] The feature extraction unit is used to extract the growth data in the central processing center in real time, and extract the features of the growth data to obtain the growth feature set, which includes the potential difference between the inside and outside of the cell △V cell , extracellular matrix tension T ECM , pH gradient and intercellular communication signal strength S comm ;

[0095] The potential difference between the inside and outside of the cell △V cell The specific extraction method △V is obtained by performing difference processing on the stem cell internal potential value Vin and the stem cell external potential value Vou in the growth data. cell =Vin-Vout;

[0096] Extracellular matrix tension T ECMThe feature extraction is performed by extracting the needle force response value Fpr and the extracellular matrix contact area Aecm in the growth data. The specific extraction method is

[0097] pH gradient The pH values ​​of different regions of stem cells obtained in the growth data are obtained by partial derivative calculation. The specific extraction method is In the formula, x represents the horizontal coordinate of the stem cell, and y represents the vertical coordinate of the stem cell. represents partial derivative;

[0098] Intercellular communication signal strength S comm By extracting the cell volume Vtj and signal molecule concentration Csi obtained from the growth data, the specific extraction method is

[0099] In the present embodiment, the system realizes the real-time collection of growth data of stem cells by integrating advanced detection equipment, such as microelectrodes, AFM probes, micro pH sensors, etc. These data cover key biophysical parameters such as the internal potential value Vin of stem cells, the external potential value Vout of stem cells, the probe force response value Fpr, the extracellular matrix contact area Aecm, the pH value ph, the cell volume Vtj and the signal molecule concentration Csi. And feature extraction is performed on the growth data, and the system can generate a comprehensive growth feature set. This data acquisition and processing method not only ensures the multi-dimensional monitoring of the growth state of stem cells, but also significantly improves the real-time and precision of the data. Traditional technology often relies on the monitoring of a single or a small amount of parameters, and there is a lag in data processing, while this system can transmit data to the central processing center in real time through a wireless transmission mechanism, and quickly complete feature extraction. This efficient data flow and processing makes the monitoring of stem cell growth more dynamic and comprehensive, so as to better capture the subtle changes in the cell growth process.

[0100] Example 3

[0101] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the feature analysis module includes a mapping function building unit and a ,comprehensive feature analysis unit;

[0102] The mapping function construction unit is used to construct a nonlinear mapping function Φ using a radial basis function RBF kernel to map the acquired growth feature set to a high-dimensional feature space;

[0103] The nonlinear mapping function Φ is mapped by the following algorithm formula;

[0104]

[0105] The comprehensive feature analysis unit is used to construct a comprehensive feature vector algorithm formula based on the mapped growth characteristics, and perform calculations to obtain a comprehensive feature vector Fcom;

[0106] The comprehensive feature vector Fcom is calculated by the following algorithm formula;

[0107]

[0108] In the formula, a1 represents the potential difference between the inside and outside of the cell △V cell and extracellular matrix tension T ECM The weight value of the interaction term, a2 represents the PH gradient and intercellular communication signal strength S comm The weight value of the interaction term, β represents the weight value of the sum of the square terms of all related features in the growth feature set; the specific value is set by the user;

[0109] a1·(ΔV cell ×T ECM ) represents the potential difference between the inside and outside of the binding cell △V cell and extracellular matrix tension T ECM , reflecting the effect of the interaction between the potential difference between the inside and outside of the cell membrane and the mechanical force on cell behavior;

[0110] Indicates the combination of pH gradient and intercellular communication signal strength S comm , reflecting the effects of microenvironmental chemical and signaling regulation on cell metabolism and group behavior;

[0111] It represents the comprehensive analysis of the nonlinear effects of the parameters in the growth feature set, reflects the contribution of each parameter to the system through the square term, and captures its potential nonlinear relationship.

[0112] The stem cell growth prediction module adopts an integrated learning method and uses the comprehensive feature vector Fcom to build a growth prediction model, and sets the prediction model formula to obtain the growth prediction model value G(t+n). The historical comprehensive feature vector Fcom is used as the input item of the prediction model to train multiple decision tree models. Each tree f i The relationship between the historical comprehensive feature vector Fcom and the target output is learned by minimizing the loss function;

[0113] The growth prediction modulus G(t+n) is obtained by the following prediction model formula;

[0114]

[0115] In the formula, G(t+n) represents the predicted cell growth value at the t+nth moment, f irepresents the nonlinear function based on the comprehensive eigenvector Fcom, γ i represents the prediction model weight, and m represents the nonlinear function f used in the prediction model. i The total number of, i represents the nonlinear function f of the i-th comprehensive eigenvector Fcom i .

[0116] In this embodiment, the system significantly improves the analysis and prediction capabilities of stem cell culture data through the feature analysis module and the stem cell growth prediction module. The feature analysis module uses a mapping function construction unit, uses a radial basis function RBF to define a nonlinear mapping function Φ, and maps the original feature space of the growth feature set to a high-dimensional feature space. This process enables the system to more accurately capture the complex nonlinear relationship in stem cell growth. Through the mapped data, the comprehensive feature analysis unit calculates the comprehensive feature vector, taking into account the interaction between each growth feature. The stem cell growth prediction module uses the comprehensive feature vector Fcom to construct a growth prediction model, uses an integrated learning method and combines multiple decision tree models for training, constructs a growth prediction model, and sets a prediction model formula to calculate and obtain the growth prediction module value G(t+n). This method generates accurate growth prediction values ​​by minimizing the loss function and optimizing the model's fit to historical data. This prediction mechanism continuously updates and optimizes the prediction results through the learning of historical data and the input of real-time data, and can provide more accurate growth predictions than traditional methods.

[0117] Example 4

[0118] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the adaptive and real-time prediction module includes an adaptive ,adjustment unit, a real-time prediction unit and a prediction evaluation unit;

[0119] The adaptive adjustment unit is used to construct an adaptive adjustment formula to dynamically adjust the dynamic weight γ in the prediction model formula at different time points. i (t+1);

[0120] Dynamic Weights i (t+1) is calculated and obtained by the following adaptive adjustment formula;

[0121]

[0122] In the formula, γ i (t+1) represents the dynamic weight after updating at time t+1, reflecting the nonlinear function f of the sub-model i The importance at the new time step, γ i (t) represents the weight of at time t, represents the partial derivative, η represents the learning rate, and L represents the loss function.

[0123] The real-time prediction unit is used to dynamically adjust the prediction model, input the real-time extracted growth feature set into the comprehensive feature vector algorithm formula, and calculate to obtain the actual comprehensive feature vector F com real-time, and then the actual comprehensive feature vector F com The real-time input is used in the adaptively adjusted prediction model to calculate the actual prediction value G. real-time ;

[0124] Actual predicted value G real-time The value is calculated by the following algorithm formula;

[0125]

[0126] In the formula, G real-time (t+n) represents the actual predicted value of stem cells when the predicted time point is n time steps after the current time;

[0127] The prediction evaluation unit presets the first growth threshold G1 and the second growth threshold G2 based on the experimental goal by the user, and then compares them with the actual prediction value G1 obtained. real-time Conduct a preliminary comparative evaluation to analyze the current stem cell growth and label the evaluation results. The specific evaluation contents are as follows;

[0128] When the actual predicted value G real-time >When the first growth threshold G1 is reached, abnormal stem cell growth is predicted, an "overgrowth" label is generated, and an optimization mechanism is executed to adjust the prediction model;

[0129] When the second growth threshold G2 ≤ ​​actual prediction value G real-time When ≤ the first growth threshold G1, the stem cell growth is predicted to be normal, and a “normal growth” label is generated at this time, and no optimization mechanism needs to be executed;

[0130] When the actual predicted value G real-time When the value is less than the first growth threshold G2, it is predicted that the stem cell growth is abnormal, and a "growth arrest" label is generated at this time, and an optimization mechanism is executed to adjust the prediction model.

[0131] In this embodiment, the adaptive and real-time prediction module of the system integrates adaptive adjustment, real-time prediction and prediction evaluation functions, which brings significant dynamic adjustment and real-time response capabilities to the stem cell growth prediction system. The adaptive adjustment unit constructs a dynamic weight adjustment formula to obtain the dynamic weight γ i(t+1), and the weights in the prediction model can be updated in real time at each time point. This dynamic adjustment mechanism ensures that the prediction model can respond to changes in the stem cell growth process in a timely manner and improves the model's adaptability to new data. The real-time prediction unit uses the growth feature set extracted in real time for prediction based on the adaptive adjustment. By updating the actual comprehensive feature vector F com The real-time input is then fed into the prediction model to generate the actual prediction value G at the current moment. real-time . This method enables the system to track the growth status of stem cells in real time and obtain the latest prediction results in a timely manner, so as to quickly respond to changes in actual growth conditions. The prediction and evaluation unit conducts preliminary growth analysis and labeling by comparing with the preset first growth threshold G1 and second growth threshold G2. According to the relationship between the actual predicted value and the first growth threshold G1 and the second growth threshold G2, the system can accurately identify abnormal growth of stem cells and generate corresponding "overgrowth" or "growth stagnation" labels. At this time, the system will execute the optimization mechanism to adjust the prediction model to ensure the continuous improvement of the prediction model. Compared with traditional methods, this process improves the real-time response capability and prediction accuracy of the system, making the detection and response to abnormal conditions during stem cell culture more timely and accurate.

[0132] Example 5

[0133] This embodiment is explained in Example 4. Please refer to Figure 1 ,Specifically: the iterative optimization module includes an optimization unit, an ,error analysis unit and an error evaluation unit;

[0134] The optimization unit is used to receive the labels generated by the preliminary evaluation in real time, and generate an optimization report based on the labels to adjust and optimize the growth environment of stem cells. The specific generated content is as follows;

[0135] When receiving the "overgrowth" tag, a first optimization report is generated, prompting a reverse adjustment to optimize the nutrient concentration and stimulation factor in the stem cell culture dish, wherein the reverse adjustment is to reduce the nutrient concentration and stimulation factor in the culture dish;

[0136] When receiving a "normal growth" label, there is no need to generate an optimization report;

[0137] When the "growth stagnation" label is received, a second optimization report is generated, prompting positive adjustment to optimize the nutrient concentration and stimulation factors in the stem cell culture dish, where the positive adjustment is to increase the nutrient concentration and stimulation factors in the culture dish.

[0138] In this embodiment, the optimization unit of the system generates a targeted optimization report by analyzing the "overgrowth" or "growth stagnation" labels generated by the preliminary evaluation. Specifically, when the system detects "overgrowth", the optimization report will recommend a reverse adjustment of the concentrations of nutrients and stimulating factors in the culture environment to reduce their levels, thereby slowing down the excessive growth of cells. When the system detects "growth stagnation", it recommends a positive adjustment to increase the concentrations of nutrients and stimulating factors to promote cell growth. This adjustment method based on actual growth conditions enables the culture environment to dynamically adapt to the needs of cells, improving the response speed and adjustment accuracy of the system.

[0139] Example 6

[0140] This embodiment is explained in Example 5. Please refer to Figure 1 Specifically: the error analysis unit is used to perform secondary data collection after optimizing the stem cell growth environment and execute the prediction model to output the optimized prediction value G opt , based on the first growth threshold G1 and the second growth threshold G2 interval, the user defines the standard prediction value G actual , perform mean error calculation to obtain the prediction error value MSE;

[0141] The prediction error value MSE is calculated by the following algorithm formula;

[0142]

[0143] In the formula, N represents the number of samples, which represents the optimized prediction value G collected within a certain period of time. opt (t+n) and the actual predicted value G real-time (t+n) corresponds to the number of data points.

[0144] The error evaluation unit is used to perform a secondary comparative evaluation between the preset optimization threshold E and the predicted error value MSE, and analyze the growth data of the optimized stem cells. The specific evaluation contents are as follows;

[0145] When the prediction error value MSE>optimization threshold E, it means that the optimized stem cell growth prediction value is normal. At this time, the optimization unit is iteratively executed to perform prompt optimization;

[0146] When the prediction error value MSE ≤ the optimization threshold E, it means that the optimized stem cell growth prediction value is normal.

[0147] In this embodiment, the error analysis unit of the system first calculates the prediction error value MSE by collecting the optimized growth data and the actual prediction value. opt Compared with the standard predicted value G actualThe prediction error value MSE is compared and calculated through the formula. This precise error calculation method can reveal the actual effect of the optimization measures on the prediction model and ensure the prediction accuracy of the model in different time periods. The error evaluation unit performs a secondary comparative analysis on the prediction error value MSE based on the preset optimization threshold E. When the prediction error value MSE is greater than the optimization threshold E, it means that the model still needs further optimization, and the system will trigger the optimization mechanism to make adjustments; when the prediction error value MSE is less than or equal to the optimization threshold E, it means that the optimized prediction model has achieved the expected accuracy. This error-based secondary evaluation mechanism makes the optimization process more scientific and targeted, and can adjust the prediction model in time to obtain more accurate prediction results.

[0148] Example 7

[0149] See also Figure 1 and Figure 2 , a data analysis method for stem cell culture, comprising the following steps:

[0150] S1, collecting growth data of stem cells, transmitting the growth data to a central processing center, and performing feature extraction to obtain a growth feature set;

[0151] S2, define a nonlinear mapping function Φ, and map the original feature space of the growth feature set to a high-dimensional feature space, and perform combined calculations to obtain a comprehensive feature vector Fcom;

[0152] S3. By using the comprehensive feature vector Fcom, a growth prediction model is constructed based on an integrated learning method, and a prediction model formula is set to calculate and obtain a growth prediction modulus G(t+n);

[0153] S4. Construct an adaptive adjustment formula to dynamically adjust the dynamic weight γ of the growth prediction model formula i (t+1), and then calculate the actual comprehensive feature vector F according to the real-time extracted growth feature set com real-time, substitute into the growth prediction model to calculate and obtain the actual prediction value G real-time The first growth threshold G1 and the second growth threshold G2 are preset and the actual prediction value G real-time Conduct preliminary comparative assessments and generate preliminary labeling suggestions;

[0154] S5. Receive the label prompts generated by the preliminary evaluation in real time, adjust and optimize the growth environment of stem cells according to the label prompts, and re-execute the system calculation after optimization to obtain the optimized prediction value G opt , and then compared with the standard prediction value G actual , perform error calculation to obtain the prediction error value MSE, and preset the optimization threshold E for secondary comparative evaluation.

[0155] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A data analysis system for stem cell culture, characterized in that: It includes data acquisition module, feature analysis module, stem cell growth prediction module, adaptive and real-time prediction module and iterative optimization module; The data acquisition module is used to collect growth data of stem cells, transmit the growth data to the central processing center, and perform feature extraction to obtain a growth feature set; The feature analysis module is used to define a nonlinear mapping function Φ, and map the original feature space of the growth feature set to a high-dimensional feature space, and perform combined calculations to obtain a comprehensive feature vector Fcom; The stem cell growth prediction module constructs a growth prediction model based on an integrated learning method by integrating the feature vector Fcom, and sets a prediction model formula to calculate and obtain a growth prediction modulus value G(t+n); The adaptive and real-time prediction module constructs an adaptive adjustment formula to dynamically adjust the dynamic weight γ of the growth prediction model formula. i (t+1), and then calculate the actual comprehensive feature vector F according to the real-time extracted growth feature set com In real-time, the growth prediction model is used to calculate the actual prediction value G. real-time The first growth threshold G1 and the second growth threshold G2 are preset and the actual prediction value G real-time Conduct preliminary comparative assessments and generate preliminary labeling suggestions; The iterative optimization module is used to receive the label prompts generated by the preliminary evaluation in real time, adjust and optimize the growth environment of the stem cells according to the label prompts, and re-execute the system calculation after optimization to obtain the optimized prediction value G opt , and then compared with the standard prediction value G actual , perform error calculation to obtain the prediction error value MSE, and perform a secondary comparative evaluation with the preset optimization threshold E.

2. A data analysis system for stem cell culture according to claim 1, characterized in that: The data acquisition module includes a data acquisition unit, a data transmission unit and a feature extraction unit; The data acquisition unit is used to install stem cell detection equipment around the stem cell culture dish to collect stem cell growth data in real time; the stem cell detection equipment includes a microelectrode, an AFM probe, a micro pH sensor, an enzyme-linked immunosorbent assay, fluorescence resonance energy transfer and an optical microscope; The growth data include the stem cell internal potential value Vin, the stem cell external potential value Vout, the probe force response value Fpr, the extracellular matrix contact area Aecm, the pH value ph, the cell volume Vtj and the signal molecule concentration Csi; The data transmission unit is used to send the collected growth data to the central processing center through a wireless transmission mechanism; The feature extraction unit is used to extract the growth data in the central processing center in real time, and extract the features of the growth data to obtain a growth feature set, which includes the potential difference ΔV between the inside and outside of the cell. cell , extracellular matrix tension T ECM , pH gradient ▽ ph and intercellular communication signal strength S comm ; The potential difference between the inside and outside of the cell is △V cell The specific extraction method △V is obtained by performing difference processing on the stem cell internal potential value Vin and the stem cell external potential value Vou in the growth data. cell =Vin-Vout; The extracellular matrix tension T ECM The feature extraction is performed by extracting the needle force response value Fpr and the extracellular matrix contact area Aecm in the growth data. The specific extraction method is The pH gradient The pH values ​​of different regions of stem cells obtained in the growth data are obtained by partial derivative calculation. The specific extraction method is In the formula, x represents the horizontal coordinate of the stem cell, and y represents the vertical coordinate of the stem cell. represents partial derivative; The intercellular communication signal strength S comm By extracting the cell volume Vtj and signal molecule concentration Csi obtained from the growth data, the specific extraction method is 3. A data analysis system for stem cell culture according to claim 2, characterized in that: The feature analysis module includes a mapping function construction unit and a comprehensive feature analysis unit; The mapping function construction unit is used to construct a nonlinear mapping function Φ using a radial basis function RBF kernel to map the acquired growth feature set to a high-dimensional feature space; The nonlinear mapping function Φ is mapped by the following algorithm formula; The comprehensive feature analysis unit is used to construct a comprehensive feature vector algorithm formula according to the mapped growth characteristics, and perform calculation to obtain the comprehensive feature vector Fcom; The comprehensive feature vector Fcom is calculated and obtained by the following algorithm formula: In the formula, a1 represents the potential difference between the inside and outside of the cell △V cell and extracellular matrix tension T ECM The weight value of the interaction term, a2 represents the PH gradient and intercellular communication signal strength S comm The weight value of the interaction term, β represents the weight value of the sum of the square terms of all related features in the growth feature set; the specific value is set by the user; a1·(ΔV cell ×T ECM ) represents the potential difference between the inside and outside of the binding cell △V cell and extracellular matrix tension T ECM , reflecting the effect of the interaction between the potential difference between the inside and outside of the cell membrane and the mechanical force on cell behavior; Indicates the combination of pH gradient and intercellular communication signal strength S comm , reflecting the effects of microenvironmental chemical and signal regulation on cell metabolism and group behavior; It represents the comprehensive analysis of the nonlinear effects of the parameters in the growth feature set, reflects the contribution of each parameter to the system through the square term, and captures its potential nonlinear relationship.

4. A data analysis system for stem cell culture according to claim 3, characterized in that: The stem cell growth prediction module adopts an integrated learning method and uses a comprehensive feature vector Fcom to construct a growth prediction model, and sets a prediction model formula to obtain a growth prediction modulus value G(t+n); The growth prediction modulus G(t+n) is obtained by the following prediction model formula; In the formula, G(t+n) represents the predicted cell growth value at the t+nth moment, f i represents the nonlinear function based on the comprehensive eigenvector Fcom, γ i represents the prediction model weight, and m represents the nonlinear function f used in the prediction model. i The total number of, i represents the nonlinear function f of the i-th comprehensive eigenvector Fcom i .

5. A data analysis system for stem cell culture according to claim 4, characterized in that: The adaptive and real-time prediction module includes an adaptive adjustment unit, a real-time prediction unit and a prediction evaluation unit; The adaptive adjustment unit is used to construct an adaptive adjustment formula to dynamically adjust the dynamic weight γ at different time points in the prediction model formula. i (t+1); The dynamic weight γ i (t+1) is calculated and obtained by the following adaptive adjustment formula; In the formula, γ i (t+1) represents the dynamic weight after updating at time t+1, reflecting the nonlinear function f of the sub-model i The importance at the new time step, γ i (t) represents the weight of at time t, represents the partial derivative, η represents the learning rate, and L represents the loss function.

6. A data analysis system for stem cell culture according to claim 5, characterized in that: The real-time prediction unit is used to input the growth feature set extracted in real time into the comprehensive feature vector algorithm formula after dynamically adjusting the prediction model, and calculate to obtain the actual comprehensive feature vector F com real-time, and then the actual comprehensive feature vector F com The real-time input is used in the adaptively adjusted prediction model to calculate the actual prediction value G. real-time ; The actual predicted value G real-time The value is calculated by the following algorithm formula; In the formula, G real-time (t+n) represents the actual predicted value of stem cells when the predicted time point is n time steps after the current time; The prediction evaluation unit presets the first growth threshold G1 and the second growth threshold G2 based on the experimental goal by the user, and then compares them with the actual prediction value G1 obtained. real-time Conduct a preliminary comparative evaluation to analyze the current stem cell growth and label the evaluation results. The specific evaluation contents are as follows; When the actual predicted value G real-time >When the first growth threshold G1 is reached, abnormal stem cell growth is predicted, an "overgrowth" label is generated, and an optimization mechanism is executed to adjust the prediction model; When the second growth threshold G2 ≤ ​​actual prediction value G real-time When ≤ the first growth threshold G1, it is predicted that the stem cell growth is normal, and a "normal growth" label is generated at this time, and there is no need to execute the optimization mechanism; When the actual predicted value G real-time When the value is less than the first growth threshold G2, it is predicted that the stem cell growth is abnormal, and a "growth arrest" label is generated. The optimization mechanism is executed to adjust the prediction model.

7. A data analysis system for stem cell culture according to claim 6, characterized in that: The iterative optimization module includes an optimization unit, an error analysis unit and an error evaluation unit; The optimization unit is used to receive the labels generated by the preliminary evaluation in real time, and generate an optimization report based on the labels to adjust and optimize the stem cell growth environment. The specific generated content is as follows; When receiving the "overgrowth" label, a first optimization report is generated, prompting reverse adjustment and optimization of the nutrient concentration and stimulation factors in the stem cell culture dish; When receiving the "Normal Growth" tag, there is no need to generate an optimization report; When the "growth stagnation" label is received, a second optimization report is generated, prompting positive adjustments to optimize the nutrient concentrations and stimulating factors in the stem cell culture dish.

8. A data analysis system for stem cell culture according to claim 7, characterized in that: The error analysis unit is used to perform secondary data collection after optimizing the growth environment of stem cells and execute the prediction model to output the optimized prediction value G opt , based on the first growth threshold G1 and the second growth threshold G2 interval, the user defines the standard prediction value G actual , perform mean error calculation to obtain the prediction error value MSE; The prediction error value MSE is calculated by the following algorithm formula: In the formula, N represents the number of samples.

9. A data analysis system for stem cell culture according to claim 8, characterized in that: The error evaluation unit is used to perform secondary comparative evaluation on the preset optimization threshold E and the prediction error value MSE, and analyze the growth data of the optimized stem cells. The specific evaluation contents are as follows; When the prediction error value MSE>optimization threshold E, it means that the optimized stem cell growth prediction value is normal. At this time, the optimization unit is iteratively executed to perform prompt optimization; When the prediction error value MSE ≤ the optimization threshold E, it means that the optimized stem cell growth prediction value is normal.

10. A data analysis method for stem cell culture, applied to a data analysis system for stem cell culture according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1, collecting growth data of stem cells, transmitting the growth data to a central processing center, and performing feature extraction to obtain a growth feature set; S2, define a nonlinear mapping function Φ, and map the original feature space of the growth feature set to a high-dimensional feature space, and perform combined calculations to obtain a comprehensive feature vector Fcom; S3. By using the comprehensive feature vector Fcom, a growth prediction model is constructed based on an integrated learning method, and a prediction model formula is set to calculate and obtain a growth prediction modulus G(t+n); S4. Construct an adaptive adjustment formula to dynamically adjust the dynamic weight γ of the growth prediction model formula i (t+1), and then calculate the actual comprehensive feature vector F according to the real-time extracted growth feature set com In real-time, substitute the growth prediction model to calculate and obtain the actual prediction value G real-time The first growth threshold G1 and the second growth threshold G2 are preset and the actual prediction value G real-time Conduct preliminary comparative assessments and generate preliminary labeling suggestions; S5. Receive the label prompts generated by the preliminary evaluation in real time, adjust and optimize the growth environment of stem cells according to the label prompts, and re-execute the system calculation after optimization to obtain the optimized prediction value G opt , and then compared with the standard prediction value G actual , perform error calculation to obtain the prediction error value MSE, and preset the optimization threshold E for secondary comparative evaluation.

Images