Control method for culture device, and control method for culture device in association with cell culture
Patent Information
- Application Number
- MYPI2025001845
- Authority / Receiving Office
- MY · MY
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2022-10-13
- Publication Date
- 2024-04-13
AI Technical Summary
Current methods for controlling cell culture processes lack effective real-time adjustments to changing culture conditions, such as cell residence time, which can impact the quality and productivity of biopharmaceuticals like antibodies.
A method involving a culture device control system that uses machine learning to derive new setting values for the culture environment based on actual and analysis values obtained at predetermined intervals, continuously updating to optimize the culture conditions for improved quality and productivity.
This approach allows for real-time adjustments to culture conditions, ensuring a more suitable environment for cell culture, leading to high-quality and high-productivity biopharmaceutical production, such as antibody drugs.
Smart Images

Figure PI2025001845_DRAWING_001 
Figure PI2025001845_DRAWING_002 
Figure PI2025001845_DRAWING_003
Abstract
Description
Method for controlling a culture device and method for controlling a culture device involving cell culture
[0001] The present invention relates to a method for controlling a culture device and a method for controlling a culture device involving cell culture.
[0002] Products produced by culture include cosmetics, medicines such as antibodies, etc. For example, when manufacturing antibody drugs, cells into which genes that produce antibodies have been incorporated are cultured, and the antibodies secreted from the cultured cells are purified to produce the antibody drugs.
[0003] Among the products produced through such cultivation, there are some that require high quality and need to be supplied to the market urgently, such as vaccines.
[0004] As a method for controlling the quality of a product produced by culture as described above, a method has been proposed in which purification is controlled while monitoring the quality characteristics of an intermediate product in a purification step (downstream process) such as the purification or concentration process of a biological product (see, for example, Patent Document 1).
[0005] By using such a control method, downstream processes can be controlled based on information from the purification and concentration processes monitored in real time.
[0006] Special table 2019-522802 publication
[0007] On the other hand, in the culture process (upstream process) where cells or microorganisms are grown and cultured, there remains the possibility of more appropriate control in response to changes in the culture state over time that occur due to factors such as cell retention time.
[0008] The present invention has been made based on the above circumstances, and its object is to provide a method for controlling a culture device that enables culture to be performed in a more suitable culture environment, and a method for controlling a culture device that involves cell culture.
[0009] One aspect of the present disclosure is a control method for a culture device that controls a culture device that performs culture by inputting setting values that set the culture environment, the control method comprising: a first step of acquiring, at a predetermined timing, actual measurement values of the culture environment at the setting values and analysis values of the culture solution obtained by culture; a second step of deriving, by machine learning, a function for determining new setting values for resetting the culture environment using the actual measurement values and analysis values acquired at the predetermined timing; and a third step of using the function to determine new setting values based on desired target values of the culture solution, and inputting this new setting value to perform culture, wherein the function used in the third step is a new function derived as needed by machine learning, including the latest actual measurement values and analysis values acquired at the predetermined timing.
[0010] The present disclosure also includes a control method for a culture apparatus involving cell culture, wherein the control method is used in the production of a biopharmaceutical involving cell culture.
[0011] In this specification, "culturing" means growing a single cell or microorganism (a unicellular organism or a multicellular organism) in a culture medium. "On-demand renewal" includes both continuous renewal (real-time renewal) and intermittent renewal at predetermined intervals.
[0012] The present invention can provide a method for controlling a culture device that allows culture to be performed in a more suitable culture environment, and a method for controlling a culture device that involves cell culture.
[0013] Fig. 1 is a schematic process diagram showing an example of a manufacturing process for producing an antibody pharmaceutical using a continuous cell culture apparatus. Fig. 2 is a schematic diagram showing an example of a continuous cell culture apparatus. Fig. 3 is a schematic diagram showing a control method of a culture apparatus in one embodiment. Fig. 4 is a schematic diagram showing an example of a continuous cell culture apparatus. Fig. 5 is a schematic process diagram showing an example of a manufacturing process for producing an antibody pharmaceutical using a batch culture apparatus.
[0014] <Control method of culture device> The control method of the culture device disclosed herein is a control method of a culture device that controls a culture device that performs culture by inputting setting values that set the culture environment, and includes: a first step of acquiring, at a predetermined timing, actual measured values of the culture environment at the setting values and analytical values of the culture solution obtained by culturing; a second step of deriving, by machine learning, a function for determining a new setting value for resetting the culture environment using the actual measured values and analytical values acquired at the predetermined timing; and a third step of using the function to determine a new setting value based on a desired target value of the culture solution, and inputting this new setting value to perform culture, wherein the function used in the third step is a new function derived as needed by machine learning, including the latest actual measured values and analytical values acquired at the predetermined timing.
[0015] The control method for a cell culture device according to the present invention will be described below by taking a control method for a cell culture device that cultures cells as an example. Such a control method for a cell culture device can be used, for example, to control a cell culture device that cultures cells to produce pharmaceuticals such as antibodies. In this specification, a continuous cell culture device is exemplified as the cell culture device. In a continuous cell culture device, a medium is continuously supplied to a culture system while an equal amount of culture solution is simultaneously withdrawn from the culture system.
[0016] The control method of the culture device described below is not limited to the control of the cell culture device. Furthermore, although one embodiment of the present invention will be described with reference to the drawings, the present invention is not limited to the embodiment shown in the drawings.
[0017] Figure 1 is a schematic diagram showing an example of a manufacturing process for producing an antibody drug using a continuous cell culture apparatus. As shown in Figure 1, an antibody drug substance can be produced by, for example, performing a culture step P1 and a purification step P2 in this order.
[0018] The culturing step P1 is a step of culturing (growing) cells. The culturing step P1 of this embodiment is performed by controlling the continuous cell culture apparatus. As shown in FIG. 2 , the continuous cell culture apparatus 1 can be configured, for example, with a continuous cell culture vessel 100, a culture medium addition device 200, a cell separation device 300, a culture environment adjustment device 400, a control device 500, and a culture solution analyzer 600.
[0019] The continuous cell culture vessel 100 cultures cells in a culture solution c under a specific culture environment. Sensors 101, 102 and a culture solution analyzer 600 provided in the continuous cell culture vessel 100 measure actual measured values PV (Process Variables) and analytical values AV (Analytical Variables), which will be described later. The culture medium addition device 200 stores fresh culture medium and supplies this medium to the continuous cell culture vessel 100. The cell separation device 300 separates the culture solution and cells with a membrane and extracts only the culture solution. The culture environment adjustment device 400 adjusts the culture environment of the culture solution c, such as the temperature, pH, and dissolved oxygen concentration. The control device 500 controls the continuous cell culture vessel 100, the culture medium addition device 200, the cell separation device 300, the culture environment adjustment device 400, etc. Details of the control device 500 will be described later.
[0020] The set value SV (Set Variable) is a variable that sets the culture environment of the continuous cell culture vessel 100. The set value SV is composed of one or more parameters. Examples of the parameters include the medium supply rate, the temperature of the culture solution in the continuous cell culture vessel 100, pH, dissolved oxygen concentration, the stirring rate of the culture solution, and the perfusion rate. In order to realize a desired culture environment using the set value SV, the continuous cell culture apparatus 1 is provided with a temperature regulator for adjusting the temperature of the culture solution c, a pH regulator for adjusting the pH of the culture solution c, an oxygen concentration regulator for adjusting the dissolved oxygen concentration in the culture solution c, an agitator for uniformly stirring the culture solution c, a perfusion device for perfusing the culture solution c, and the like, all of which are not shown.
[0021] The purification process P2 is a process for purifying specific products secreted from cells (such as physiologically active substances, useful proteins such as antibodies, etc.) by separating them from other unnecessary substances (such as cell shells, viruses, and products secreted from cells such as lactic acid and ammonia, etc.) so that the product has a predetermined quality.
[0022] In the purification step P2, the purification device (method) used is not particularly limited as long as it can separate a specific product. Examples of purification devices include a continuous capture device, a virus inactivation device, and a filtration sterilization device, as shown in FIG. 1, as well as a gravity settling device, an ultrasonic aggregation device, a filtration separation device, and the like, which are not shown. In the purification step P2, these purification devices can be used alone or in combination.
[0023] In the above-mentioned purification step P2, known techniques and devices can be used.
[0024] Here, one embodiment of a method for controlling the continuous cell culture apparatus 1 used in the culture step P1 among the above-mentioned steps will be described in detail with reference to FIGS.
[0025] The control method for the continuous cell culture apparatus 1 inputs a set value SV that sets the culture environment to control the continuous cell culture apparatus 1. As shown in Fig. 3, the control method for the continuous cell culture apparatus 1 of this embodiment includes a first step S1, a second step S2, and a third step S3.
[0026] [First Step] The first step S1 is a step of acquiring, at a predetermined timing, an actual measurement value PV of the culture environment at the set value SV and an analysis value AV of the culture solution c obtained by culturing. The first step S1 is executed by measuring the actual measurement value PV and the analysis value AV using the sensors 101, 102, the culture solution analyzer 600, etc.
[0027] Specifically, in the first step S1, actual measured values PV of the culture environment (such as the temperature, pH, pressure, dissolved oxygen concentration, dissolved carbon dioxide concentration, and culture solution volume) in the continuous cell culture apparatus 1 operated under a culture environment with a preset set value SV (hereinafter also referred to as the "existing set value SV") are acquired using the sensor 101, etc. Also, in the first step S1, an analysis value AV, which is an index of quality and productivity (such as the number of viable cells, antibody concentration, and impurity concentration) for the culture solution c cultured using the existing set value SV, is acquired using the sensor 102, the culture solution analyzer 600, etc. The timing (hereinafter also referred to as the "predetermined timing") at which the actual measured values PV and the analysis value AV are acquired (actually measured) may be continuous ("constant") or intermittent ("timely"). Furthermore, the actual measured values PV and the analysis value AV may be acquired at the same time or at different times.
[0028] [Second Step] The second step S2 is a step of deriving a function F by machine learning to obtain a new set value SV for resetting the culture environment using the actual measured value PV and the analysis value AV obtained at a predetermined timing. The second step S2 is executed by the database 501, the data extraction unit 502, and the machine learning unit 503 of the control device 500, which will be described later.
[0029] Specifically, in the second step S2, a function F for calculating a new set value SV is derived through machine learning using the actual measured value PV and the analysis value AV obtained in the first step S1, which serve as training data.
[0030] Examples of learning methods used in machine learning include Bayesian estimation, support vector regression, random forest regression, and neural networks.
[0031] Among these, machine learning is preferably performed using Bayesian inference or support vector regression. By applying Bayesian inference or support vector regression to machine learning, a more accurate function F can be derived from a relatively small amount of data on actual measured values PV and analyzed values AV, which serve as training data. Such learning methods using Bayesian inference or support vector regression are particularly effective when using limited samples during process development (when there is little data available, such as actual measured values PV and analyzed values AV), and when there is an urgent need to supply pharmaceuticals to the market.
[0032] [Third Step] The third step S3 is a step in which a new set value SV is calculated based on a target variable (TV) of the desired culture solution c using the function F, and the new set value SV is input to the continuous cell culture apparatus 1 to perform culture. The third step S3 is executed by the set value calculation unit 504 (described later) of the control device 500, the continuous cell culture vessel 100, the culture medium addition device 200, the cell separation device 300, and the culture environment adjustment device 400.
[0033] In the third step S3, first, a new set value SV is calculated using the desired target value TV of the culture solution c and the function F derived in the second step S2.
[0034] The desired target value TV of the culture solution c is an index related to quality and productivity (number of viable cells, antibody concentration, impurity concentration, etc.).
[0035] The new set value SV may be determined to be a value within a predetermined range. Specifically, for example, conditions (setting ranges of each parameter to be observed) such as the culture environment in which the continuous cell culture apparatus 1 is operated may be predetermined. In such a case, if the new set value SV calculated using the function F is outside the above-mentioned setting range, the new set value SV calculated using the function F may be corrected to fall within the above-mentioned setting range, and the corrected value may be used as the new set value SV. This allows cells to be cultured under, for example, approved pharmaceutical manufacturing conditions.
[0036] Next, the calculated new set value SV is input to the continuous cell culture apparatus 1, and the culture solution c is cultured in the culture environment of the continuous cell culture apparatus 1 operated using this new set value SV. That is, by replacing the existing set value SV with the new set value SV, the culture solution c is cultured in the new culture environment. Note that when the values of the parameters constituting the existing set value SV and the new set value SV are the same, the parameters are maintained at the same values.
[0037] Here, in the control method for the continuous cell culture apparatus 1, a new function F derived as needed by machine learning is used as the function F used in the third step S3, including the latest measured values PV and the latest analyzed values AV obtained at a predetermined timing. That is, in the third step S3, the latest function F updated as needed based on the latest information on the culture solution c is used, and a new set value SV is calculated using the latest function F.
[0038] The function F may be updated to a new function at the same timing (continuous "always" or intermittent "timely") as the predetermined timing at which the actual measured values PV and analysis values AV are obtained, or at a timing different from the predetermined timing. When updating at the same timing as the predetermined timing, the culture solution c can be cultured in a culture environment with a new set value SV obtained using the latest actual measured values PV and the latest analysis values AV.
[0039] Next, a control device capable of executing the above-described control method for the continuous cell culture device will be described. As shown in Fig. 4, the control device 500 of the continuous cell culture device 1 can be configured, for example, with a database 501, a data extraction unit 502, a machine learning unit 503, and a set value calculation unit 504.
[0040] The database 501 stores data such as set values SV, actual measured values PV and analysis values AV obtained at predetermined times, and other lot and process information.
[0041] The data extraction unit 502 extracts data from the data stored in the database 501 at appropriate times to be used by the machine learning unit 503 and the setting value calculation unit 504, which will be described later.
[0042] The machine learning unit 503 is a component that executes step S2. The machine learning unit 503 uses the actual measured values PV and analysis values AV acquired from the database 501 via the data extraction unit 502 to perform machine learning to derive a function F for determining a new set value SV for resetting the culture environment. The actual measured values PV and analysis values AV that serve as training data when performing machine learning may include the latest actual measured values PV and latest analysis values AV. By including the latest actual measured values PV and latest analysis values AV, a more accurate function F can be derived through machine learning using more training data, including the latest information.
[0043] The set value calculation unit 504 is a component that calculates a new set value SV in step S3. The set value calculation unit 504 calculates the new set value SV based on the target value TV of the desired culture solution c, using the function F derived by the machine learning unit 503. Information (values) of each parameter included in the new set value SV calculated by the set value calculation unit 504 is sent to the continuous cell culture vessel 100, the culture medium addition device 200, the cell separation device 300, and the culture environment adjustment device 400.
[0044] The above-described control device 500 can be configured, for example, by the following hardware. That is, the database 501 is configured by a storage device that stores various data. The storage device can be configured by any type of storage medium, and may include, for example, a semiconductor memory, a hard disk drive, etc. The data extraction unit 502, the machine learning unit 503, and the setting value calculation unit 504 can be configured by a central processing unit (CPU) or the like that can execute the above-described processes.
[0045] As described above, according to the control method for the continuous cell culture apparatus 1, the function F used in the third step S3 is a new function F derived as needed through machine learning, including the latest measured values PV and analyzed values AV obtained at a predetermined timing. Therefore, a more accurate new set value SV can be determined through machine learning using a larger amount of training data, including the latest information, and culture can be performed under a more suitable culture environment. As a result, products with the desired quality and productivity can be reliably manufactured.
[0046] <Control method of a culture device involving cell culture> The control method of a culture device involving cell culture of the present disclosure is characterized in that the control method of a culture device described above in the section <Control method of a culture device> is used in the production of biopharmaceuticals involving cell culture.
[0047] Like the control method for the continuous cell culture apparatus 1 exemplified in the above-described embodiment, the control method for the culture apparatus of the present disclosure allows cells to be cultured in a more suitable culture environment, thereby enabling the production of biopharmaceuticals such as antibody drugs with high quality and high productivity. Therefore, the control method for the culture apparatus of the present disclosure can be suitably applied to a control method for a culture apparatus that involves cell culture and is used in the production of biopharmaceuticals that involve cell culture.
[0048] The present disclosure is not limited to the configurations of the above-described embodiments, but is intended to include all modifications within the scope and meaning equivalent to the claims, as defined by the claims. Part of the configurations of the above-described embodiments may be deleted or replaced with other configurations, or other configurations may be added to the configurations of the above-described embodiments.
[0049] For example, in the above-described embodiment, a control method for a continuous cell culture apparatus 1 for producing antibody pharmaceuticals (biopharmaceuticals) was described as an example of a control method for a culture apparatus. However, the control method for a culture apparatus of the present disclosure can also be applied to the culture of other organisms as long as they can be grown by culture. The control method for a culture apparatus of the present disclosure can also be applied to a control method for a culture apparatus for producing cosmetics and the like by culturing microorganisms such as unicellular organisms or multicellular organisms.
[0050] In the above-described embodiment, the continuous culture apparatus 1 was described as an example of a culture apparatus to which the culture apparatus control method is applied. However, the culture apparatus control method of the present disclosure may also be applied to a culture apparatus that performs batch processing (batch-type culture apparatus) as shown in Figure 5 (Figure 5 illustrates a manufacturing process for a biopharmaceutical drug substance using a batch-type cell culture apparatus). The control method for a batch-type culture apparatus can also handle set values, measured values, analytical values, target values, and functions, just like a continuous culture apparatus.
[0051] 1 Continuous cell culture device (culture device) SV Set value PV Actual measured value AV Analyzed value TV Target value S1 First step S2 Second step S3 Third step P1 Culture process P2 Purification process
Claims
1. A control method for a culture device that controls a culture device that performs culture by inputting setting values that set the culture environment, the control method comprising: a first step of acquiring, at a predetermined timing, actual measurement values of the culture environment at the setting values and analytical values of the culture solution obtained by culture; a second step of deriving, by machine learning, a function for determining new setting values for resetting the culture environment using the actual measurement values and analytical values acquired at the predetermined timing; and a third step of using the function to determine new setting values based on the desired target values of the culture solution, and inputting this new setting value to perform culture, wherein the function used in the third step is a new function derived as needed by machine learning, including the latest actual measurement values and analytical values acquired at the predetermined timing.
2. The control method according to claim 1, wherein the function is updated to the new function at the same timing as the predetermined timing.
3. The control method according to claim 1, wherein the machine learning is performed using Bayesian estimation or support vector regression.
4. The control method according to claim 1, wherein the new set value is determined to be within a predetermined range.
5. A control method for a culture device involving cell culture, wherein the control method according to any one of claims 1 to 4 is used in the production of biopharmaceuticals involving cell culture.
