Electronic device, method performed by the same, medium, and computer program product

Abstract

This disclosure relates to electronic devices and methods, media, and computer program products thereof. An electronic device includes: a memory storing instructions thereon; and a processor configured to execute the instructions stored in the memory to cause the electronic device to perform the following operations: receiving capacitance data of a bioreactor; receiving temperature data of the bioreactor; estimating diameter data of cells in the bioreactor based on the capacitance data; selecting one of a plurality of trained models based on the temperature data and the diameter data; and estimating the density of living cells using the capacitance data through the selected trained model.

Description

Technical Field

[0001] This disclosure generally relates to the monitoring of live cell density, and more specifically, to estimating live cell density using machine learning models based on capacitance data. Background Technology

[0002] Process analysis techniques (PAT), especially in-situ and real-time monitoring techniques, are applied to assist in assessing process robustness, monitoring and improving process development, and pharmaceutical manufacturing. They show great potential in developing robust and novel processes for producing products with desired quality attributes. PAT enhances traditional offline analysis by providing convenient, high-volume, and continuous measurements to further understand the process.

[0003] Recently, the U.S. Food and Drug Administration (FDA) released key guidance on Process Assurance and Analysis (PAT) in the pharmaceutical industry. As a powerful tool for process monitoring and analysis systems, PAT enhances the ability to monitor cell culture processes in situ and in real time, improves the online monitoring and control of key process parameters, and is more in line with the concept that quality originates from design.

[0004] Viable cell density (VCD) is one of the most important monitoring parameters in cell culture process development. It is not only crucial for monitoring cell growth but also a key trigger for determining and adjusting strategies such as cooling and feeding. Traditional offline detection methods, such as trypan blue staining, typically require manual sampling and VCD calculation through cell counting. The low sampling frequency and long detection delay after sampling limit its ability to monitor changes in VCD, making it impossible to adjust the culture process "near real-time" based on cell growth, such as adjusting parameters like cooling, feeding, and cell extraction. Therefore, introducing automated PAT tools for VCD process monitoring and control is particularly important.

[0005] There remains a need to improve the monitoring of VCDs. Summary of the Invention

[0006] This invention generally relates to an electronic device comprising: a memory having instructions stored thereon; and a processor configured to execute the instructions stored in the memory to cause the electronic device to perform the following operations: receiving capacitance data of a bioreactor; receiving temperature data of the bioreactor; estimating diameter data of cells in the bioreactor based on the capacitance data; selecting one of a plurality of trained models based on the temperature data and the diameter data; and estimating the density of living cells using the capacitance data through the selected trained model.

[0007] In one implementation, when the temperature data is higher than a first threshold, a first trained model is selected, which is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is higher than the first threshold; and / or, when the temperature data is lower than the first threshold and the change in cell diameter data within a predetermined time is less than a second threshold, a second trained model is selected, which is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is lower than the first threshold and the change in cell diameter data within a predetermined time is less than the second threshold; and / or, when the temperature data is lower than the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold, a third trained model is selected, which is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is lower than the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold.

[0008] In one embodiment, when the temperature data is below a first threshold and the change in cell diameter data within a predetermined time period is greater than a second threshold, the processor is further configured to execute instructions stored in the memory to cause the electronic device to perform the following operations: estimating live cell volume data using capacitance data through a fourth trained model; estimating corrected cell diameter data using cell diameter data through a fifth trained model; and estimating live cell density using the estimated live cell volume data and the estimated corrected cell diameter data through a sixth trained model; wherein the fourth trained model is trained using measured capacitance data and measured live cell volume data when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time period is greater than the second threshold, wherein the fifth trained model is trained using cell diameter data estimated from capacitance data and measured cell diameter data when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time period is greater than the second threshold, and wherein the sixth trained model is trained using live cell volume data estimated by the fourth trained model, cell diameter data estimated by the fifth trained model, and live cell density data measured when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time period is greater than the second threshold.

[0009] In one implementation, the diameter data of cells in the bioreactor is estimated based on capacitance data using partial least squares regression.

[0010] In one embodiment, the processor is further configured to execute instructions stored in the memory to cause the electronic device to perform the following operations: periodically comparing the estimated live cell density with the offline detected live cell density; and retraining the plurality of trained models if the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold.

[0011] In one implementation, if the number of times the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold is greater than a set value, the frequency of comparing the estimated live cell density with the offline detected live cell density is increased.

[0012] In some aspects, the present invention also relates to a method performed by an electronic device, comprising: receiving capacitance data of a bioreactor; receiving temperature data of the bioreactor; estimating diameter data of cells in the bioreactor based on the capacitance data; selecting one of a plurality of trained models based on the temperature data and the diameter data; and estimating the density of living cells using the capacitance data through the selected trained model.

[0013] In one implementation, when the temperature data is higher than a first threshold, a first trained model is selected, which is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is higher than the first threshold; and / or when the temperature data is lower than the first threshold and the change in cell diameter data within a predetermined time is less than a second threshold, a second trained model is selected, which is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is lower than the first threshold and the change in cell diameter data within a predetermined time is less than the second threshold; and / or when the temperature data is lower than the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold, a third trained model is selected, which is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is lower than the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold.

[0014] In one embodiment, when the temperature data is below a first threshold and the change in cell diameter data within a predetermined time period is greater than a second threshold, the method further includes: estimating live cell volume data using capacitance data through a fourth trained model; estimating corrected cell diameter data using cell diameter data through a fifth trained model; and estimating live cell density using the estimated live cell volume data and the estimated corrected cell diameter data through a sixth trained model; wherein the fourth trained model is trained using measured capacitance data and measured live cell volume data when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time period is greater than the second threshold; wherein the fifth trained model is trained using cell diameter data estimated from capacitance data and measured cell diameter data when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time period is greater than the second threshold; and wherein the sixth trained model is trained using live cell volume data estimated by the fourth trained model, cell diameter data estimated by the fifth trained model, and live cell density data measured when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time period is greater than the second threshold.

[0015] In one implementation, the diameter data of cells in the bioreactor is estimated based on capacitance data using partial least squares regression.

[0016] In one implementation, the method further includes: periodically comparing the estimated live cell density with the offline detected live cell density; and retraining the plurality of trained models if the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold.

[0017] In one implementation, if the number of times the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold is greater than a set value, the frequency of comparing the estimated live cell density with the offline detected live cell density is increased.

[0018] In some aspects, the present invention relates to a non-transitory computer-readable medium having instructions stored thereon that, when executed by a processor of an electronic device, implement the aforementioned method executed by the electronic device.

[0019] In some aspects, the present invention relates to a computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the aforementioned method.

[0020] The foregoing is a summary and therefore inevitably contains simplifications, generalizations, and omissions of detail; thus, those skilled in the art will understand that the summary is merely illustrative and not intended to be limiting in any way. Other aspects, features, and advantages of the methods, compositions, and / or apparatuses and / or other subjects described herein will become apparent from the teachings set forth herein. The summary is provided to present a selection of concepts in a simplified form, which will be further described in the detailed descriptions below. This summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to assist in determining the scope of the claimed subject matter. Furthermore, all references, patents, and published patent applications cited throughout this application are incorporated herein by reference in their entirety. Attached Figure Description

[0021] To better understand this disclosure and to show how to implement it, it will now be described by way of example with reference to the accompanying drawings, in which:

[0022] Figure 1 This is a schematic diagram illustrating an exemplary electronic device according to an embodiment of the present disclosure;

[0023] Figure 2 This is a schematic diagram of a system used to implement online monitoring of VCDs;

[0024] Figure 3 An exemplary flowchart of a method for an electronic device according to an embodiment of the present disclosure is shown;

[0025] Figure 4 An exemplary block diagram of a method for an electronic device according to an embodiment of the present disclosure is shown;

[0026] Figure 5 An exemplary block diagram of a method for an electronic device according to another embodiment of the present disclosure is shown;

[0027] Figure 6 The diagram shows the predicted and offline detection values ​​of the training set VCD using different models, where (a) represents the results of estimating VCD using different models based on bioreactor temperature and cell diameter data; and (b) represents the results of estimating VCD using one model during cell culture.

[0028] Figure 7 The illustration shows a comparison between estimated values ​​and offline detection values ​​of training set data using different models according to an embodiment of the present disclosure, wherein (a) represents the comparison between estimated live cell density and offline detection values; (b) represents the comparison between estimated cell diameter and offline detection values; and (c) represents the comparison between estimated live cell volume and offline detection values; and

[0029] Figure 8Estimated data and offline detection values ​​of VCD during 14 days of cell culture are shown, wherein (a) represents estimated data and offline detection values ​​of VCD using a different model according to an embodiment of the present disclosure; and (b) represents estimated data and offline detection values ​​of VCD using a commercial software model. Detailed Implementation

[0030] While the invention may be embodied in many different forms, only specific illustrative embodiments thereof are disclosed herein, which illustrate the principles of the invention. It should be emphasized that the invention is not limited to the specific embodiments shown. Furthermore, any section headings used herein are for organizational purposes only and should not be construed as limiting the subject matter.

[0031] Unless otherwise defined herein, scientific and technical terms related to this invention shall have the meanings commonly understood by one of ordinary skill in the art. Furthermore, unless the context otherwise requires, singular terms shall include plural terms, and plural terms shall include singular terms. More specifically, as used in this specification and the appended claims, the singular form “a” may include the plural objects referred to, unless the context explicitly specifies otherwise. Thus, for example, reference to “a protein” may include multiple proteins; reference to “cell” includes mixtures of multiple cells, etc., cells including, but not limited to, animal cells, plant cells, bacterial cells, or fungal cells, animal cells such as mammalian cells. Suitable mammalian cells include, but are not limited to, Chinese hamster ovary (CHO) cells (e.g., CHO K1 cells), hybridomas, BHK (baby hamster kidney) cells, myeloma cells, human cells (e.g., HEK-293 cells), human lymphoblasts, E1 immortalized HER cells, and mouse cells (e.g., NSO cells and SP / 20 cells).

[0032] Generally, the nomenclature and techniques described herein relating to cell and tissue culture, molecular biology, immunology, microbiology, genetics, protein and nucleic acid chemistry, and hybridization are well-known and commonly used in the field. Unless otherwise stated, the methods and techniques of the present invention are generally performed according to conventional methods known in the art, as well as those described in the various general and more specific references cited and discussed throughout this specification. The nomenclature described herein relating to laboratory procedures and techniques in analytical chemistry, synthetic organic chemistry, and medicinal chemistry is among those well-known and commonly used in the field. Furthermore, any section headings used herein are for organizational purposes only and should not be construed as limiting the subject matter described.

[0033] To better understand this invention, the relevant terms are defined and explained as follows.

[0034] The term "perfusion cell culture" refers to a continuous process of culturing cells. Perfusion is an upstream treatment that retains cells within a bioreactor while continuously removing cells, cellular waste, and culture medium depleted of nutrients by cellular metabolism. The rate at which fresh culture medium is supplied to the cells is the same as the rate at which used culture medium is removed.

[0035] As defined herein, a "bioreactor" is a device containing cell cultures (i.e., cells and a culture medium for culturing). It preferably maintains a favorable environment for the cells by providing suitable conditions, such as temperature, pH, dissolved oxygen concentration, ion concentration, continuous agitation, etc. It preferably provides a sterile environment.

[0036] Typically, the bioreactor may include, for example, a fermenter, a stirred tank reactor, an adhesive bioreactor, a wave bioreactor, a disposable bioreactor, etc.

[0037] Those skilled in the art will understand that cells are cultured to produce a desired product, such as a monoclonal antibody or recombinant protein.

[0038] In this application, unless otherwise stated, “or” is used to mean “and / or”. Furthermore, the use of the term “comprising” and other forms such as “including” and “containing” is not restrictive. Additionally, the scope provided in the specification and appended claims includes both endpoints and all points between them.

[0039] Currently, online monitoring tools such as near-infrared spectroscopy, Raman spectroscopy, and capacitance sensing have received widespread attention and continuous reporting. Among them, the capacitance method refers to predicting the volume per living cell (VCV) by measuring the capacitance value, and it can also predict the volume per living cell (VCD) when the diameter of the living cells remains constant. Due to its advantages such as high detection sensitivity, low measurement interference, fast scanning frequency, and ease of scaling up, this technology has shown great potential in early research and development and manufacturing.

[0040] However, the practical applications of commercial capacitance detection technology have been very limited to date, typically only applicable to N-generation steady-state perfusion cell culture or N-1 perfusion cell culture, and difficult to popularize in fed-batch culture. This is because when cells transition from the logarithmic growth phase to the plateau phase or when process parameters change, their diameter changes significantly due to aging or stress responses. While cell diameters are relatively stable in N-generation steady-state or N-1 perfusion cell cultures, they change significantly in fed-batch culture. This deviation from steady-state diameter leads to poorer prediction accuracy for VCD using this technology.

[0041] The development of multi-frequency scanning modes and the analysis of capacitance spectra have improved the detection performance of live cell diameter and VCD to some extent, but the improvement is limited. Related commercial tools still generally face many technical challenges. For example, under low VCD conditions, excessive background noise makes it impossible to detect live cell diameter; under high VCD conditions, the diameter predicted by multi-frequency scanning deviates significantly from the offline values, requiring further correction through methods such as linear regression; and the detection parameters are significantly affected by process parameters such as temperature.

[0042] Furthermore, existing technologies fail to address the technical issues related to VCD estimation. Therefore, it is essential to develop a PAT tool capable of accurately monitoring VCD, providing real-time online data for cell culture process control.

[0043] To address the technical problem of large errors in measuring live cell density and cell diameter using existing commercial tools, this disclosure, on the one hand, further mines capacitance spectrum information, including but not limited to selecting different models to estimate live cell density based on the temperature of the bioreactor and changes in cell diameter over a predetermined time. This solves the aforementioned technical problem of large errors in online detection of live cell density and cell diameter. On the other hand, this disclosure solves the technical problems of real-time high-frequency transmission, processing, display, and storage of signals such as live cell density, live cell volume, and cell diameter through the development of live cell density estimation and model correction functions and corresponding signal processing systems. This disclosure comprehensively improves the online monitoring capability of capacitive sensing technology for live cell density in bioreactors.

[0044] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0045] Figure 1 A block diagram illustrating an example of an electronic device 100 according to some embodiments is presented.

[0046] Electronic device 100 can be used to perform various embodiments of the methods according to this disclosure described below. Electronic device 100 may include a processing subsystem 110, a memory subsystem 112, and a networking subsystem 114. Processing subsystem 110 includes one or more devices configured to perform computational operations. For example, processing subsystem 110 may include one or more microprocessors, ASICs, microcontrollers, programmable logic devices, graphics processing units (GPUs), and / or one or more digital signal processors (DSPs).

[0047] The memory subsystem 112 includes one or more devices for storing data and / or for processing instructions for the subsystem 110 and the networking subsystem 114. For example, the memory subsystem 112 may include dynamic random access memory (DRAM), static random access memory (SRAM), and / or other types of memory (sometimes collectively or individually referred to as "computer-readable storage media").

[0048] In some embodiments, the memory subsystem 112 is coupled to one or more high-capacity mass storage devices (not shown). For example, the memory subsystem 112 may be coupled to a magnetic or optical drive, a solid-state drive, or another type of mass storage device. In these embodiments, the electronic device 100 may use the memory subsystem 112 as fast-access storage for frequently used data, while the mass storage device is used to store data that is not frequently used.

[0049] The networking subsystem 114 includes one or more devices configured to couple to and communicate over wired and / or wireless networks (e.g., to perform network operations), including: control logic 116, interface circuitry 118, and one or more antennas 120 (or antenna elements). (Although) Figure 1 The electronic device 100 may include one or more antennas 120, but in some embodiments, it may include one or more nodes, such as node 108, which may be coupled to one or more antennas 120. Therefore, the electronic device 100 may or may not include one or more antennas 120. For example, the networking subsystem 114 may include a Bluetooth networking system, a cellular networking system (e.g., 3G / 4G / 5G networks, such as UMTS, LTE, etc.), a USB networking system, a networking system based on standards described in IEEE 802.11 (e.g., a Wi-Fi networking system), an Ethernet networking system, and / or another networking system.

[0050] Within electronic device 100, a processing subsystem 110, a memory subsystem 112, and a networking subsystem 114 are coupled together using a bus 128. The bus 128 may include electrical, optical, and / or electro-optical connections that the subsystems can use to communicate commands and data, etc. Although only one bus 128 is shown for clarity, different embodiments may include different numbers or configurations of electrical, optical, and / or electro-optical connections within the subsystems.

[0051] In some embodiments, electronic device 100 includes a display subsystem 126 for displaying information on a display, which may include a display driver and a display, such as a liquid crystal display, a multi-touch screen, etc.

[0052] Although specific components are used to describe electronic device 100, different components and / or subsystems may be present in alternative embodiments. For example, electronic device 100 may include one or more additional processing subsystems, memory subsystems, networking subsystems, and / or display subsystems. Additionally, one or more of these subsystems may not be present in electronic device 100. Furthermore, in some embodiments, electronic device 100 may include... Figure 1 One or more additional subsystems not shown. Additionally, although in Figure 1 Individual subsystems are shown, but in some embodiments, some or all of a given subsystem or component may be integrated into one or more other subsystems or components in electronic device 100. For example, in some embodiments, program instructions 122 are included in operating system 124 and / or control logic 116 is included in interface circuitry 118.

[0053] Figure 2 This is a schematic diagram of a system used to implement online monitoring of VCDs.

[0054] According to embodiments of this disclosure, a system 200 for implementing online monitoring of a VCD may include: a bioreactor 201, a monitoring system 202, a system 203 for estimating live cell density, and a process control system 204. The bioreactor 201 may be a device containing cell cultures (i.e., cells and culture medium) as described above. The bioreactor 201 maintains a favorable environment for cells by providing suitable conditions, such as temperature, pH, dissolved oxygen concentration, ion concentration, continuous stirring, etc. The monitoring system 202 may include a temperature sensor 202-1 and a capacitance electrode 202-2. The temperature sensor 202-1 is used to measure the temperature of the bioreactor 201, and the capacitance electrode 202-2 is immersed in the bioreactor 201 to collect capacitance spectrum data during the cell culture process at specific acquisition intervals (e.g., but not limited to 30 s intervals). Those skilled in the art will understand that a capacitance spectrum refers to a data set of capacitance values ​​collected by a capacitance electrode immersed in the bioreactor at multiple measurement frequencies (e.g., 50-20000 kHz). The capacitance data may originate from the entire cell culture process within the bioreactor.

[0055] Those skilled in the art will understand that the monitoring system 202 may also include other types of sensors to measure other parameters of the bioreactor.

[0056] The measured temperature and capacitance data are then transmitted to system 203 for estimating live cell density via communication protocols supported by the live cell electrodes (such as Modbus, OPC, serial port, etc.). In system 203, the diameter of cells in the bioreactor is estimated based on the capacitance data, and different models are selected based on the temperature and diameter data to estimate the live cell density. The estimated live cell density can be used in process control system 204 to achieve automatic control of the cell culture process. Alternatively, the corresponding functions of process control system 204 can be used to achieve online visualization of the predicted values.

[0057] The system 203 for estimating live cell density can convert capacitance spectrum data into target online predicted values ​​through different models, including but not limited to cell diameter, live cell density, live cell volume, viability, etc.

[0058] The system 203 for estimating live cell density may also include a unit that implements a correction function. This correction function periodically validates the online predictions. For example, the correction function may periodically compare the estimated live cell density with the offline detected live cell density, and if the difference between the estimated and offline detected live cell densities exceeds a predetermined threshold, retrain multiple used models.

[0059] During the training of multiple models, the collected capacitance spectra can be correlated with various measurements detected offline, including but not limited to cell diameter, viable cell density, viable cell volume, and viability. For example, after preprocessing the capacitance spectrum data (such as standard normal transformation, first derivative, etc.), partial least squares (PLS) regression or linear regression can be used to fit and correlate the capacitance spectrum data with the above measurements.

[0060] In the technical solution of this application, the changes in cell diameter within the bioreactor over a predetermined time period are considered to have a significant impact on the estimation of viable cell density using capacitance data. Therefore, different models are used for different temperatures and cell diameter changes within the bioreactor over a predetermined time period. Those skilled in the art will understand that different models can be models with different structures or models using different parameters.

[0061] Those skilled in the art will understand that the model can be a classifier to be trained. Classifiers can include random decision forests, linear classifiers, support vector machines (SVMs), or neural networks such as recurrent neural networks (RNNs).

[0062] Those skilled in the art will understand that suitable artificial neural networks include, but are not limited to, feedforward neural networks, radial basis function networks, self-organizing maps, learned vector quantization, recurrent neural networks, Hopfield networks, Boltzmann machines, echo-state networks, long short-term memory, bidirectional recurrent neural networks, hierarchical recurrent neural networks, stochastic neural networks, modular neural networks, associative neural networks, deep neural networks, deep belief networks, convolutional neural networks, convolutional deep belief networks, large memory storage and retrieval neural networks, deep Boltzmann machines, deep stacked networks, tensor deep stacked networks, spike and plate-constrained Boltzmann machines, composite hierarchical deep models, deep coding networks, multi-layer kernel machines, or deep Q-networks.

[0063] Those skilled in the art can select a suitable classifier to implement the technical solution of this application as needed.

[0064] Figure 3 An exemplary flowchart 300 of a method for an electronic device according to an embodiment of the present disclosure is shown. The method according to an embodiment of the present disclosure can be used in, for example... Figure 1 Executed on the aforementioned electronic device.

[0065] In step 301, capacitance data from the bioreactor is received. As described above, capacitance data of the culture medium in the bioreactor can be acquired at predetermined time intervals using capacitance electrodes in the bioreactor.

[0066] In step 302, temperature data from the bioreactor is received. As described above, temperature data can be acquired using a temperature sensor connected to the bioreactor.

[0067] In step 303, the diameter data of cells in the bioreactor is estimated based on the capacitance data. For example, the capacitance data can be preprocessed using multivariate data analysis (MVDA) techniques, and then partial least squares (PLS) regression can be used to correlate the capacitance values ​​and scan frequencies in the capacitance data with the cell diameters detected offline using a cell counting analyzer for model training. The trained model then uses the measured capacitance data to obtain the cell diameter data. Those skilled in the art will understand that other methods can be used to estimate cell diameter data from capacitance data.

[0068] In step 304, one of several trained models is selected based on the temperature data and diameter data.

[0069] In step 305, the live cell density is estimated using the capacitance data through a selected trained model.

[0070] Typically, the temperature of the bioreactor is high in the initial stage of the culture process, while the temperature decreases in the later stages due to changes in process conditions. Furthermore, the diameter of cells in the bioreactor remains constant in the initial stage of the culture process, but when cells transition from the logarithmic growth phase to the plateau phase or when process parameters change, the cell diameter changes significantly due to aging or stress responses, deviating from steady state. Figure 3 In the method shown according to one embodiment of this disclosure, the effects of changes in the temperature of the bioreactor and the cell diameter within the bioreactor over a predetermined time period on the estimation of viable cell density using capacitance data are considered. Different models are selected to estimate viable cell density based on different temperatures and changes in cell diameter over a predetermined time period. By training the model on capacitance data and viable cell density data under different conditions, the influence of temperature and cell diameter on model parameters is incorporated into the model training process, making the trained model more accurate in estimating viable cell density.

[0071] In one embodiment according to this disclosure, the estimated live cell density is periodically compared with the offline detected live cell density. If the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold, multiple trained models are retrained. That is, to ensure the accuracy and robustness of the online output results, the method according to embodiments of this disclosure also implements an automatic correction function for the model itself. For example, once the deviation between the online and offline VCD values ​​exceeds a set value (e.g., 3.0E6 cells / mL), the system incorporates the offline detection data into the existing model, retrains it, and uses new model association parameters.

[0072] In one embodiment of the present disclosure, when the number of times the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold is greater than a set value, for example, when the difference between the estimated live cell density and the offline detected live cell density is greater than the third threshold twice consecutively, the frequency of comparing the estimated live cell density with the offline detected live cell density is increased.

[0073] The following will be referenced Figure 4 This describes a method 400 for estimating live cell density according to an embodiment of the present disclosure.

[0074] exist Figure 4 In section 401, the input signal is received. The input signal can come from, for example... Figure 2 The 202 monitoring system shown. Input signals may include signals from, for example... Figure 2 The temperature data from the temperature sensor 202-1 and the capacitance data from the capacitive electrode 202-2 are shown. For example, the capacitance value at a frequency of 580kHz can be selected as the capacitance data used for subsequent linear regression.

[0075] In 402, the input signal is preprocessed, for example by standard normal transformation, first derivative, etc., and then the diameter of the cells in the bioreactor is obtained from the capacitance data by fitting and correlating the data through partial least squares (PLS) regression or linear regression.

[0076] In step 403, it is determined whether the temperature data is higher than a first threshold. The first threshold can be determined based on the culture process. In step 405, when the temperature data is higher than the first threshold, a first trained model is selected, and the viable cell density is estimated using the capacitance data. In step 408, the estimated viable cell density is output. In one embodiment of this disclosure, the output estimated viable cell density can be further used to achieve automatic control of the VCD.

[0077] In one embodiment of this disclosure, the first trained model is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is above a first threshold.

[0078] As the process continues, the temperature of the bioreactor will decrease. When the temperature falls below a first threshold, the method proceeds to step 404. In step 404, it is determined whether the change in cell diameter data within a predetermined time period exceeds a second threshold. The second threshold can be determined based on the culture process. For example, the second threshold can be determined by experimentally measuring the difference between diameter signals of cells under different states or conditions. For example, the predetermined time period can be 24 hours, and the difference between the diameter signal at the current moment and the diameter signal 24 hours ago can be used to determine the change in cell diameter data. Those skilled in the art will understand that the predetermined time period for determining the change in cell diameter data can be set according to actual needs.

[0079] When the temperature data is below a first threshold and the change in cell diameter data within a predetermined time is less than a second threshold, a second trained model is selected at step 406 to estimate the viable cell density using capacitance data. At step 408, the estimated viable cell density is output. In one embodiment of this disclosure, the second trained model is trained using measured capacitance data and measured viable cell density data obtained when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time is less than the second threshold.

[0080] As previously mentioned, when cells transition from the logarithmic growth phase to the plateau phase or when process parameters change, cell diameters deviate significantly from steady state due to aging or stress responses. When the temperature data is below a first threshold and the change in cell diameter data within a predetermined time exceeds a second threshold, in step 407, a third trained model is selected, and the viable cell density is estimated using capacitance data. In step 408, the estimated viable cell density is output. In one embodiment of this disclosure, the third trained model is trained using measured capacitance data and measured viable cell density data when the bioreactor temperature is below the first threshold and the change in cell diameter data within a predetermined time exceeds the second threshold.

[0081] The following will be referenced Figure 5 This describes a method 500 for estimating live cell density according to an embodiment of the present disclosure.

[0082] Figure 5 The block diagrams 501, 502, 503, 504, 505, 506, and 510 in the diagram are... Figure 4 Similar to block diagrams 401, 402, 403, 404, 405, 406 and 408, these block diagrams will not be described again for the sake of a clearer illustration of the technical solution; instead, only the different parts will be described.

[0083] When the temperature data is below the first threshold and the change in cell diameter data within a predetermined time exceeds the second threshold. Figure 5 The method shown proceeds to block diagram 507. In 507, live cell volume data is estimated using capacitance data through a fourth trained model; in 508, corrected cell diameter data is estimated using cell diameter data through a fifth trained model; and in 509, live cell density is estimated using the estimated live cell volume data and the estimated corrected cell diameter data through a sixth trained model. In one embodiment of this disclosure, the fourth trained model is trained using measured capacitance data and measured live cell volume data when the bioreactor temperature is below a first threshold and the change in cell diameter data within a predetermined time is greater than a second threshold. The fifth trained model is trained using cell diameter data estimated from capacitance data and measured cell diameter data when the bioreactor temperature is below the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold. The sixth trained model is trained using live cell volume data estimated by the fourth trained model, cell diameter data estimated by the fifth trained model, and live cell density data measured when the bioreactor temperature is below the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold.

[0084] When the temperature falls below a first threshold and the change in cell diameter within a predetermined time exceeds a second threshold, the process undergoes significant changes. On one hand, the bioreactor temperature affects the parameters of the model used to estimate live cell density; on the other hand, the cell diameter has also changed significantly, deviating from steady state. The accuracy of directly using capacitance data to estimate live cell density will be affected. Figure 5 In the method shown according to an embodiment of this disclosure, in step 507, capacitance data is used to estimate live cell volume data, and in step 508, a corrected cell diameter is estimated using the cell diameter estimated by capacitance spectrum in preprocessing step 502. That is, the fifth trained model in step 508 uses the offline measured cell diameter for model training, thereby correcting the cell diameter used to estimate live cell density. This reduces the impact of cell diameter deviation from steady state on the prediction, making the prediction of live cell density more accurate. In step 509, the sixth trained model uses the estimated live cell volume data and the corrected cell diameter data to estimate the live cell volume. In step 511, the estimated live cell density, estimated live cell volume, and estimated live cell diameter data are output. The above signal processing logic enables real-time prediction and output of VCD under different conditions.

[0085] The technical solution of this disclosure will be illustrated by experiments below. Those skilled in the art will understand that the experimental content is merely exemplary and is not intended to limit the technical solution of this disclosure.

[0086] experiment

[0087] Material:

[0088] Bioprocess controller system: Finesse G3Lab universal controller, TruBio Delta V 5.0 control software, 3L Applikon glass bioreactor.

[0089] Capacitance spectrum monitoring system: ring capacitance electrode probe (Aber Instruments Ltd), standard integrated Futura host, Futura Connect single-channel hub, Futura SCADA software, FUTURA Tool software.

[0090] Offline detection system: Vi-CELL XR cell counting analyzer (Beckman Coulter Life Sciences).

[0091] Cell lines and culture media: Chinese hamster ovary (CHO) cell lines expressing human IgG1 monoclonal antibodies, basal culture media with clearly defined chemical components, and supplemental culture media, all of which are commercially available.

[0092] Capacitance spectrum acquisition:

[0093] Online capacitance spectrum data acquisition was performed using Aber's ring capacitance electrode probe and its integrated main unit and hub. During measurement, the probe was immersed in the bulk culture medium of the bioreactor, and the measurement frequency range was 50 kHz to 20000 kHz. The electrode probe acquired capacitance spectra every 30 seconds, continuously throughout the entire culture cycle.

[0094] Model training:

[0095] Based on multivariate data analysis (MVDA) technology, the capacitance spectrum data is first preprocessed, such as by standard normal transformation and first derivative. Then, partial least squares (PLS) regression is used to correlate the capacitance values ​​and scanning frequency in the capacitance spectrum with the cell diameter detected offline using a cell counting analyzer for model training. Linear regression is used to correlate single-point capacitance values ​​at specific frequencies with the offline detected viable cell density (VCD) and viable cell volume (VCV) for model training.

[0096] In the early stages of cell culture, the model can accurately predict VCD directly based on capacitance values. However, when the process involves cooling, the model used at high temperatures becomes inapplicable. Using different models, i.e., segmented models, can improve prediction accuracy. Similarly, with significant changes in cell diameter, the prediction accuracy of a single model for VCD deteriorates. Experimental results show that training different models on data from different temperatures and diameter changes can improve the prediction accuracy of VCD.

[0097] In the experiment, the model was adjusted to predict cell diameter and live cell volume, and then VCD was indirectly predicted based on these. Figure 6 The figures show the predicted and offline detection values ​​of VCD for the training set using different models. (a) represents the results of estimating VCD using different models based on bioreactor temperature and cell diameter data; (b) represents the results of estimating VCD using a single model during cell culture. As can be seen from the figures, the predicted viable cell density using different models is closer to the offline monitoring values ​​than the predicted viable cell density using a single model, especially after day 5 of the culture process. The segmented model (left figure) has relatively higher accuracy in predicting VCD, with a maximum deviation of only 2.9E6 cells / mL, significantly lower than the maximum deviation (4.5E6 cells / mL) of the non-segmented model (right figure). The maximum deviation of the segmented model occurs on day 5 of culture, precisely when the model was segmented; therefore, this deviation can be attributed to the timing of the segmentation and the differences between the models before and after segmentation.

[0098] To demonstrate the feasibility of the segmented model proposed in this invention for monitoring VCD, VCV, and cell diameter during cell culture, Figure 7 Table 1 summarizes the model's prediction results and root mean square error, respectively.

[0099] Table 1: Root Mean Square Error (RMSE) of the Training Set

[0100]

[0101] like Figure 7 As shown in Table 1, the overall prediction results for VCD are acceptable, R 2 The accuracy was 0.98, and the RMSE was 1.1E6 cells / mL. The error mainly came from high VCD data (>18.5E6 cells / mL), in which case the model's predicted value was slightly higher than the offline detection value. The prediction results for cell diameter were also acceptable, with R... 2 The value was 0.93, and the RMSE was 0.4 μm. It is important to emphasize that the prediction of cell diameter has certain requirements for VCD (>5.0E6 cells / mL). When the VCD is too low, its predictive performance is poor. Figure 7 The reason why there are fewer data points related to cell diameter than for VCD or VCV is that the model performs best in predicting VCV, R. 2 The value was 0.99, the RMSE was 2024.3E6μm3 / mL, and the normalized RMSE (NRMSE) was only 3.4%.

[0102] Figure 8 The figure shows estimated data and offline detection values ​​of VCD during a 14-day cell culture process, where (a) represents estimated data and offline detection values ​​of VCD using different models according to an embodiment of the present disclosure; and (b) represents estimated data and offline detection values ​​of VCD using a commercial software model. Circles in the figure represent offline detection values ​​of VCD, while solid lines represent online VCD values ​​predicted by the system based on capacitance spectra. As shown in (a), a first trained model is used before cooling on day 5; a second trained model is used before a significant change in cell diameter after cooling; and fourth, fifth, and sixth trained models are used on day 7, after a significant change in cell diameter after cooling. As shown in (a), the circles remain near the solid lines throughout the 14-day culture period, indicating that the online and offline prediction values ​​of VCD in the embodiment of the present disclosure are essentially consistent, demonstrating that the system of the present disclosure can effectively achieve continuous, stable, and accurate online monitoring of VCD. In contrast, the prediction values ​​using a single model shown in (b) deviate from the offline monitoring values ​​after temperature reduction and significant diameter changes, indicating a decrease in prediction accuracy.

[0103] While some operations in the foregoing embodiments are implemented in software, in general, the operations in the foregoing embodiments can be implemented in a variety of configurations and architectures. Therefore, some or all of the operations in the foregoing embodiments can be performed in hardware, software, or both. For example, at least some operations in the communication technology can be implemented using program instructions 122 of electronic device 100, operating system 124 (such as a driver for interface circuit 118), or in firmware within interface circuit 118. Alternatively or additionally, at least some operations in the communication technology can be implemented at the physical layer, such as in hardware within interface circuit 118 of electronic device 100.

[0104] This disclosure can be implemented as any combination of apparatus, system, integrated circuit, and computer program on a non-transitory computer-readable medium. One or more processors can be implemented as integrated circuits (ICs), application-specific integrated circuits (ASICs), or large-scale integrated circuits (LSIs), system LSIs, super LSIs, or ultra LSI components that perform some or all of the functions described in this disclosure.

[0105] The steps of the method according to this disclosure can also be performed separately by multiple components included in the apparatus. According to one embodiment, these components can be implemented as computer program modules created to implement the steps of the method, and the apparatus including these components can be a program module architecture that implements the method by computer program.

[0106] This disclosure includes the use of software, application programs, computer programs, or algorithms. Software, application programs, computer programs, or algorithms may be stored on a non-transitory computer-readable medium to cause a computer, such as one or more processors, to perform the steps described above and in the accompanying drawings. For example, one or more memories may store software or algorithms with executable instructions, and one or more processors may be associated with executing a set of instructions of the software or algorithm to enhance security in any number of wireless networks according to embodiments described in this disclosure.

[0107] Software and computer programs (also referred to as programs, software applications, applications, components, or code) include machine instructions for programmable processors and can be implemented in high-level procedural languages, object-oriented programming languages, functional programming languages, logic programming languages, assembly languages, or machine languages. The term "computer-readable medium" means any computer program product, apparatus, or device used to provide machine instructions or data to a programmable data processor, such as magnetic disks, optical disks, solid-state storage devices, memories, and programmable logic devices (PLDs), including computer-readable media that receive machine instructions as computer-readable signals.

[0108] For example, computer-readable media may include dynamic random access memory (DRAM), random access memory (RAM), read-only memory (ROM), electrically erasable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage devices, magnetic disk storage devices or other magnetic storage devices, or any other medium that can be used to carry or store required computer-readable program code in the form of instructions or data structures, and that can be accessed by a general-purpose or special-purpose computer or a general-purpose or special-purpose processor. As used herein, a disk or disc includes compact discs (CD), laser discs, optical discs, digital versatile discs (DVD), floppy disks, and Blu-ray discs, wherein a disk typically copies data magnetically, while a disc copies data optically using a laser. Combinations of the above are also included within the scope of computer-readable media.

[0109] In one or more embodiments, the use of the terms "can," "able to," "operable as," or "configurable as" refers to means of means, logic, hardware, and / or elements designed to be used in a specified manner. The subject matter of this disclosure is provided as examples of means, systems, methods, and programs for performing the features described in this disclosure. However, other features or variations are contemplated in addition to the features described above. It is contemplated that components and functions of this disclosure can be implemented using any emerging techniques that may replace any of the above-described implementations.

[0110] Furthermore, the above description provides examples and does not limit the scope, applicability, or configuration set forth in the claims. Changes may be made to the function and arrangement of the elements discussed without departing from the spirit and scope of this disclosure. Various processes or components may be appropriately omitted, substituted, or added in various embodiments. For example, features described with respect to certain embodiments may be combined in other embodiments.

[0111] Similarly, although the operations are depicted in a specific order in the accompanying drawings, this should not be construed as requiring the operations to be performed in the specific order shown or in sequential order, or requiring the execution of all illustrated operations to achieve the desired result. In some cases, multitasking and parallel processing can be advantageous.

[0112] Those skilled in the art will further understand that the invention can be implemented in other specific forms without departing from the spirit or central attributes of the invention. Since the foregoing description of the invention discloses only exemplary embodiments thereof, it should be understood that other variations are contemplated within the scope of the invention. Therefore, the invention is not limited to the specific embodiments already described in detail herein. Rather, reference should be made to the appended claims to define the scope and content of the invention.

Claims

1. An electronic device, comprising: A memory that stores instructions; as well as The processor is configured to execute instructions stored in the memory to cause the electronic device to perform the following operations: Receive capacitance data from the bioreactor; Receive temperature data from the bioreactor; The diameter of cells in the bioreactor is estimated based on the capacitance data. Choose one of several trained models based on temperature and diameter data; as well as The density of live cells is estimated using capacitance data through a selected trained model.

2. The electronic device of claim 1, wherein when the temperature data is higher than a first threshold, a first trained model is selected, the first trained model being trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is higher than the first threshold; and / or, When the temperature data is below a first threshold and the change in cell diameter data within a predetermined time is less than a second threshold, a second trained model is selected. This second trained model is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time is less than the second threshold; and / or, When the temperature data is below the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold, a third trained model is selected. The third trained model is trained using measured capacitance data and measured live cell density data when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold.

3. The electronic device of claim 1, wherein when the temperature data is below a first threshold and the change in cell diameter data within a predetermined time is greater than a second threshold, the processor is further configured to execute instructions stored in the memory to cause the electronic device to perform the following operations: Live cell volume data are estimated using capacitance data through a fourth trained model; The corrected cell diameter data is estimated using the cell diameter data through a fifth trained model; as well as Live cell density is estimated using estimated live cell volume data and estimated corrected cell diameter data through a sixth-order trained model. The fourth trained model was trained using measured capacitance data and measured live cell volume data when the temperature of the bioreactor was below a first threshold and the change in cell diameter data was greater than a second threshold within a predetermined time. The fifth trained model is trained using cell diameter data estimated from capacitance data and cell diameter data measured when the temperature of the bioreactor is below a first threshold and the change in cell diameter data within a predetermined time is greater than a second threshold. The sixth trained model is trained using live cell volume data estimated by the fourth trained model, corrected cell diameter data estimated by the fifth trained model, and live cell density data measured when the temperature of the bioreactor is below a first threshold and the change in cell diameter data is greater than a second threshold within a predetermined time.

4. The electronic device of claim 1, wherein the diameter data of cells in the bioreactor is estimated from the capacitance data by partial least squares regression.

5. The electronic device of claim 1, wherein the processor is further configured to execute instructions stored in the memory to cause the electronic device to perform the following operations: The estimated live cell density is periodically compared with the live cell density detected offline; and If the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold, the plurality of trained models are retrained.

6. The electronic device of claim 5, wherein when the number of times the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold is greater than a set value, the frequency of comparing the estimated live cell density with the offline detected live cell density is increased.

7. A method performed by an electronic device, comprising: Receive capacitance data from the bioreactor; Receive temperature data from the bioreactor; The diameter of cells in the bioreactor is estimated based on the capacitance data. Choose one of several trained models based on temperature and diameter data; as well as The density of live cells is estimated using capacitance data through a selected trained model.

8. The method of claim 7, wherein when the temperature data is higher than a first threshold, a first trained model is selected, the first trained model being trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is higher than the first threshold; and / or When the temperature data is below a first threshold and the change in cell diameter data within a predetermined time is less than a second threshold, a second trained model is selected. This second trained model is trained using measured capacitance data and measured live cell density data obtained when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time is less than the second threshold; and / or When the temperature data is below the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold, a third trained model is selected. The third trained model is trained using measured capacitance data and measured live cell density data when the temperature of the bioreactor is below the first threshold and the change in cell diameter data within a predetermined time is greater than the second threshold.

9. The method of claim 7, wherein when the temperature data is below a first threshold and the change in cell diameter data within a predetermined time is greater than a second threshold, the method further comprises: Live cell volume data are estimated using capacitance data through a fourth trained model; The corrected cell diameter data is estimated using the cell diameter data through a fifth trained model; as well as Live cell density is estimated using estimated live cell volume data and estimated corrected cell diameter data through a sixth-order trained model. The fourth trained model was trained using measured capacitance data and measured live cell volume data when the temperature of the bioreactor was below a first threshold and the change in cell diameter data was greater than a second threshold within a predetermined time. The fifth trained model is trained using cell diameter data estimated from capacitance data and cell diameter data measured when the temperature of the bioreactor is below a first threshold and the change in cell diameter data within a predetermined time is greater than a second threshold. The sixth trained model is trained using live cell volume data estimated by the fourth trained model, corrected cell diameter data estimated by the fifth trained model, and live cell density data measured when the temperature of the bioreactor is below a first threshold and the change in cell diameter data is greater than a second threshold within a predetermined time.

10. The method of claim 7, wherein the diameter data of cells in the bioreactor are estimated from the capacitance data by partial least squares regression.

11. The method of claim 7, further comprising: The estimated live cell density is periodically compared with the live cell density detected offline. as well as If the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold, the plurality of trained models are retrained.

12. The method of claim 11, wherein when the number of times the difference between the estimated live cell density and the offline detected live cell density is greater than a third threshold is greater than a set value, the frequency of comparing the estimated live cell density with the offline detected live cell density is increased.

13. A non-transitory computer-readable medium having instructions stored thereon that, when executed by a processor of an electronic device, implement the method performed by the electronic device as described in any one of claims 7-12.

14. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program performs the steps of the method according to any one of claims 7 to 12.

Images