Learned model, estimation device, and estimation method

Abstract

To provide a learned model, with which index data relating to the kinetics of contents in a container can be evaluated with high accuracy; an estimation device; and an estimation method.SOLUTION: A learned model is trained using, as an explanatory variable, at least one of data relating to the shape of a stirrer provided in a container, data relating to the shape of a sparger provided in the container, data relating to the shape of the container, and data relating to physical properties, composition, chemical properties, biochemical properties, or concentration of contents accommodated in the container, and using, as an objective variable, a mass transfer capacity coefficient (kLa), a stirring power (P / V), a respiratory rate, a concentration of bacterial bodies, a concentration of a product from bacterial bodies and cells, or a titer.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a learned model, an estimation device, and an estimation method.

Background Art

[0002] Devices called stirred tanks and bubble columns are used as reactors that dissolve components in gas (such as oxygen) into a liquid by blowing gas (aerating) into the liquid in a container or by vibrating the container, and cause a chemical reaction with components contained in the liquid, or as culture tanks that supply oxygen or the like to microorganisms or animal cells in the liquid or degas carbon dioxide or the like for culturing, and are applied in fields such as petrochemicals, chemistry, pharmaceuticals, and life sciences.

[0003] In such stirred tanks and bubble columns with gas aeration, the gas aerated into the liquid is dispersed in the liquid as bubbles or gas masses, but the dissolution of gas components from the gas into the liquid or the transfer of dissolved components from the liquid to the gas (mass transfer) occurs at the interface between the gas and the liquid (gas-liquid interface) in these bubbles and gas masses. Therefore, the mass transfer rate depends on the total area of the gas-liquid interface and the dispersion state of the bubbles and gas masses in the liquid. In order to improve the mass transfer rate as much as possible, the device design is carried out so as to increase the gas-liquid interface area and uniformly disperse the bubbles and gas masses in the liquid.

[0004] As this index, the mass transfer capacity coefficient kL*a (hereinafter, kLa), which is obtained by multiplying the gas-liquid interface area concentration a (gas-liquid interface area per unit liquid volume) by kL, which is an index of the diffusion rate of gas components into the liquid, is well known (for example, see Patent Document 1). In particular, when scaling up from a small laboratory scale to a large commercial scale, the design is considered so that kLa in the commercial device is equivalent to that in the laboratory scale. Similarly, the stirring power per unit liquid volume (P / V), the respiration rate, the concentration of cells and bacteria, the concentration of products from cells and bacteria, the titer, etc. are also indices that are considered to be equivalent to those in the laboratory scale in the commercial scale.

Prior Art Documents

Patent Documents

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2001-75947 [Summary of the Invention] [Problems to be Solved by the Invention]

[0006] By the way, the kLa and P / V of a stirring tank or a bubble column are evaluated by a correlation formula established based on experimental results and physical considerations. However, for example, the kLa in a stirring tank is greatly affected by the operating conditions of the stirrer, the fluid properties, and the stirring shape. Therefore, there is a problem that it is difficult to accurately evaluate the kLa under all conditions by the correlation formula. In addition, the respiration rate, the concentration of bacteria and cells, the concentration of products from bacteria and cells, the titer, etc. are not based on the operation of an actual machine, and a correlation formula for evaluating them based on the shape of the stirrer or sparger, the shape of the container, and the physical properties of the contents of the tank, etc. has not been established, and there is a problem that it is difficult to accurately evaluate them.

[0007] An object of the present invention is to provide a learned model, an estimation device, and an estimation method capable of accurately evaluating index data related to the dynamics of the contents of a container, such as the above kLa, P / V, respiration rate, concentration of bacteria and cells, concentration of products from bacteria and cells, and titer. [Means for Solving the Problems]

[0008] According to the present invention, a learned model having the following configuration is provided. [1] A learned model, wherein at least one of data related to the shape of a stirrer provided in a container, data related to the shape of a sparger provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container is used as an explanatory variable. A trained model learned with the mass transfer coefficient (kLa), stirring power (P / V), respiration rate, concentration of cells and bacteria, concentration of products from cells and bacteria, or titer as the target variable.

[0009] According to the present invention, at least one of data related to the shape of the stirrer, data related to the shape of the sparger, data related to the shape of the container, physical properties, composition, chemical properties, or biochemical properties of the contents (explanatory variables) is input into the trained model, and the mass transfer coefficient (kLa), stirring power (P / V), respiration rate, concentration of cells and bacteria, concentration of products from cells and bacteria, or titer output as the target variable can be obtained. Therefore, compared with the prior art, index data related to the dynamics of the contents of the container can be evaluated with high accuracy.

[0010] Hereinafter, various embodiments of the present invention will be exemplified. The embodiments shown below can be combined with each other. [2][1] The trained model according to [1], wherein data related to the shape of the stirrer, data related to the shape of the sparger, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container are learned as explanatory variables. Trained model. [3][1] or the trained model according to [2], The data related to the shape of the stirrer includes data related to the shape of the stirring blade constituting the stirrer, The data related to the shape of the stirring blade and the data related to the shape of the sparger are Include data with numerical labels for each type of shape, or A first numerical value is assigned to the region occupied by the stirring blade and the sparger in the three-dimensional space, and a second numerical value is assigned to the region not occupied by the stirring blade and the sparger in the three-dimensional space, thereby including digitized data. Trained model. An estimation device comprising an input unit and an objective variable acquisition unit, wherein the input unit inputs the explanatory variable to the learned model according to any one of [1] to [3], and the objective variable acquisition unit acquires the objective variable output from the learned model when the explanatory variable is input to the learned model. The estimation device according to [5][4], further comprising an explanatory variable acquisition unit, wherein the explanatory variable acquisition unit acquires the value of the explanatory variable input to the learned model when the value of the objective variable output from the learned model becomes a predetermined value. The estimation device according to [6][5], wherein among the first data and the second data constituting the explanatory variable, the value of the first data is a fixed value, and the value of the second data is a variable value, and the explanatory variable acquisition unit acquires the value of the second data input to the learned model when the value of the objective variable output from the learned model becomes a predetermined value. An estimation method, comprising inputting the explanatory variable to the learned model according to any one of [1] to [3], and acquiring the objective variable output from the learned model when the explanatory variable is input to the learned model. [Advantages of the Invention]

[0011] According to the present invention, it is possible to accurately evaluate the index data related to the dynamics of the contents of the container. [Brief Description of the Drawings]

[0012]

Figure 1

Figure 2

Figure 3

[0013] Embodiments of the present invention will be described with reference to the accompanying drawings. In each figure, those denoted by the same reference numerals have the same or similar configurations.

[0014] (Configuration of the estimation device 10) FIG. 1 is a block diagram showing the functional configuration of the estimation device 10 in the present embodiment. By using a learned model, the estimation device 10 estimates (acquires) kLa (mass transfer coefficient), P / V (stirring power per unit liquid volume), respiration rate, concentration of cells and bacteria, concentration of products from cells and bacteria, or titer as index data related to the dynamics of the contents of the container. Here, mass transfer is the dissolution of components in the gas (e.g., oxygen, etc.) into the liquid in the container or the transfer of dissolved components from the liquid to the gas by blowing (ventilating) gas into the liquid in the container in a stirred tank or bubble column with gas ventilation, or by vibrating the container.

[0015] Note that the stirred tank and the bubble column are used as a reactor for dissolving components in the gas into the liquid and chemically reacting with the components contained in the liquid, or as a culture tank for supplying oxygen or degassing carbon dioxide, etc. to microorganisms and animal cells, etc. in the liquid for culturing, and are applied in the fields of petrochemicals, chemistry, pharmaceuticals, and life sciences.

[0016] As shown in FIG. 1, the estimation device 10 includes a learned model storage unit 1, an operation reception unit 2, an explanatory variable input unit 3, an objective variable acquisition unit 4, a determination unit 5, an explanatory variable acquisition unit 6, and a display unit 7. Note that the explanatory variable input unit 3 functions as the "input unit" of the present invention.

[0017] Each component of the estimation device 10 may be implemented by software or by hardware. When implemented by software, various functions can be realized by a CPU executing a computer program. The program may be stored in a built-in storage unit or in a non-transitory computer-readable recording medium. Also, a program stored in an external storage unit may be read out and realized by so-called cloud computing. When implemented by hardware, it can be realized by various circuits such as an ASIC, an FPGA, or a DRP (Dynamically Reconfigurable Processor). In the present embodiment, various information and concepts including these are handled, and these are represented by the high and low of signal values as a set of binary bits composed of 0 or 1, and communication and calculation can be executed by the above software or hardware modes.

[0018] The learned model storage unit 1 is a database that stores a kLa model 1A, a P / V model 1B, a breathing rate model 1C, a cell body and cell concentration model 1D, and a titer model 1E as learned models machine-learned by a machine learning unit (not shown).

[0019] The machine learning unit performs machine learning using the learning data (teacher data) stored in a learning data storage unit (not shown). That is, the machine learning unit inputs a plurality of sets of learning data into the learning model, and causes the learning model to machine-learn (so-called supervised learning) the correlation between the input data and the output data constituting the learning data, thereby generating a learned model. For example, a CNN (Convolutional Neural Network) or the like is used for the learned model. And, for example, deep learning or the like is used for the learning algorithm of machine learning.

[0020] In the present embodiment, although the learned model is described as a CNN, the learned model is not limited to a CNN and may be learned by any learning algorithm such as a neural network other than a CNN, an SVM (Support Vector Machine), a Bayesian network, or a regression tree.

[0021] The kLa model 1A is a learned model learned using, as input data (explanatory variables) constituting the learning data, data regarding the shape of a stirrer provided in a container, data regarding the shape of a sparger provided in the container, data regarding the shape of the container, and data regarding the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container, and using, as output data (objective variables) constituting the learning data, the kLa (mass transfer coefficient).

[0022] Examples of data related to the shape of the stirrer include, for example, data related to the shape of the stirring blades that make up the stirrer. The types of stirring blades, the diameter of the stirring blades, and the number of stirring blades are set according to the shape of the stirring blades (paddle (with / without inclination), disk turbine (with / without inclination), propeller, anchor, etc.). The data related to the shape of the stirring blades includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (e.g., 1) to the area occupied by the stirring blades in the three-dimensional space and a second numerical value (e.g., 0) to the area not occupied by the stirring blades in the three-dimensional space. The shape data of the stirring blades digitized by the first numerical value and the second numerical value can be represented as one numerical value, for example, by taking the average value and standard deviation throughout the device, and can be used as an explanatory variable in the learning model. Also, a method of attaching numerical labels for each type of shape of the stirring blades and a method of digitizing by assigning a first numerical value (e.g., 1) to the area occupied by the stirring blades in the three-dimensional space and a second numerical value (e.g., 0) to the area not occupied by the stirring blades in the three-dimensional space may be combined. In this case, by attaching numerical labels for each type of shape of the stirring blades (paddle (with / without inclination), disk turbine (with / without inclination), propeller, anchor, etc.) and further expressing the presence or absence of occupation of the stirring blades in the three-dimensional space numerically, the shape of the stirring blades can be expressed more precisely as numerical data.

[0023] Examples of data related to the shape of the sparger provided in the container include, for example, the type of aeration sparger, the diameter of the aeration sparger, the pore diameter of the aeration sparger, and the number of pores of the aeration sparger, which are set according to the shape of the sparger (ring shape, cylindrical shape, circular shape). The data related to the shape of the sparger includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (e.g., 1) to the area occupied by the sparger in the three-dimensional space and a second numerical value (e.g., 0) to the area not occupied by the sparger in the three-dimensional space.

[0024] Data on the shape of the container is data on the shape of the container itself that holds the contents. Examples of data on the shape of the container include the type of the container's shape (such as cylindrical, rectangular, etc.), the dimensions of the container (inner diameter in the case of a cylinder, width in the case of a rectangle, height, etc.), the presence or absence of baffles (in the case of a stirring tank), and the like.

[0025] Examples of data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents include the physical properties, composition, chemical properties, and biochemical properties specific to the materials contained in the contents, the physical properties of the entire contents in the container, and the concentrations of the respective materials contained in the contents.

[0026] The physical properties specific to the materials contained in the contents include, for example, the density, viscosity, elasticity, non-Newtonian properties, etc. specific to the materials. The composition specific to the materials contained in the contents includes, for example, the elements constituting the materials and their contents, the ions dissolved in the materials and their contents, the presence or absence of surfactants, the contents of saccharides, the contents of amino acids, the contents of vitamins, the contents of inorganic salts, the contents of lipid components, the contents of trace metals, etc. The chemical properties specific to the materials contained in the contents include, for example, the reaction activity, activation energy, reaction rate, reaction rate constant, etc. in the chemical reactions of the materials. The biochemical properties specific to the materials contained in the contents include, for example, (when the material is a cell line) the presence or absence of hyphae formation, the doubling time, (when the material is a culture medium) the growth factors contained. The materials contained in the contents are, for example, the medium / culture solution and cells, and in that case, as the physical properties specific to the materials contained in the contents, the fluid physical properties of the medium / culture solution (density, viscosity, non-Newtonian properties, etc.), the medium composition (the presence or absence of surfactants, the types of growth factors, the concentration of saccharides contained in the medium, the concentration of amino acids contained in the medium, the concentration of vitamins contained in the medium, the concentration of inorganic salts contained in the medium, the concentration of lipid components contained in the medium, the concentration of trace metals contained in the medium, etc.), the types of cell lines / strains (mass, viscosity, elasticity, size, shape, the presence or absence of hyphae formation, doubling time, etc.) are used as explanatory variables of the learning model. The physical properties, composition, chemical properties, and biochemical properties specific to the materials contained in the contents are parameters that do not change after the operation of the device containing the contents or parameters whose values are known under operating conditions such as temperature and pressure.

[0027] The physical properties of the entire contents within the container include, for example, the temperature of the contents, osmotic pressure, oxidation-reduction potential (ORP), and pH. The concentration of each material contained in the contents includes, for example, the seeding amount, cell concentration, etc. (when the contents are a culture solution and cells). The data regarding the physical properties of the entire contents within the container used as explanatory variables for the learning model may be the initial values before the start of the container operation, or may be predicted values obtained by simulating after the start of the container operation (for example, numerical fluid dynamics simulation, for example, CFD (Computational Fluid Dynamics) simulation), or may be measured values continuously acquired after the start of the container operation. For example, the cell concentration and pH may be initial values set by values input by the user, or predicted values by CFD (Computational Fluid Dynamics) simulation (the average value + standard deviation value within the container), or values set as the average value + standard deviation value within the device obtained from a multi-point measurement method after the start of the operation.

[0028] As learning data for the learned model, among the data regarding the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container, it is preferably to include, in particular, the physical properties, composition, chemical properties, and biochemical properties specific to the materials contained in the contents, the physical properties of the entire contents within the container, and the initial values of the concentration of each material contained in the contents. Since these data can be grasped without actual machine operation, they can be easily obtained as explanatory variables for the learned model in order to predict the performance of the actual machine in device design. More preferably, in addition to the physical properties, composition, chemical properties, and biochemical properties specific to the materials contained in the contents, the physical properties of the entire contents within the container, and the initial values of the concentration of each material contained in the contents, it is also good to include values predicted by simulation for the physical properties of the entire contents within the container and the concentration of each material contained in the contents. By including not only the initial values of these values but also the values after device operation in the explanatory variables, the value of the target variable can be estimated more accurately.

[0029] In another example, in addition to the physical properties, composition, chemical properties, and biochemical properties specific to the materials contained in the contents, the physical properties of the entire contents in the container, and the initial values of the concentrations of the respective materials contained in the contents, and the physical properties of the entire contents in the container predicted by simulation and the values of the concentrations of the respective materials contained in the contents, it is preferable to include the measured values of the physical properties of the entire contents in the container and the concentrations of the respective materials contained in the contents obtained during the operation of the actual machine. Since the physical properties of the entire contents in the container and the concentrations of the respective materials contained in the contents are values that change moment by moment during the operation of the device, by inputting these data as explanatory variables during the operation, the learned model can be used for the evaluation of the actual machine operation.

[0030] Note that the kLa model 1A may be learned with the superficial gas velocity, aeration rate, turbulent energy dissipation rate, gas hold-up, bubble diameter, control method of dissolved oxygen (DO) (upper and lower limit values, etc.), rotation speed of the stirrer, culture solution input / discharge conditions, shear stress, feed rate / perfusion rate of the culture solution (in the case of fed-batch culture / perfusion culture) as the input data (explanatory variables) constituting the learning data, and kLa (mass transfer coefficient) as the output data (objective variable) constituting the learning data. By configuring the kLa model 1A to include one or more of these parameters in the input data, the prediction accuracy of kLa can be further improved.

[0031] The superficial gas velocity is a set value calculated from the aeration rate and the tank diameter input by the user. The aeration rate, rotation speed, control method of dissolved oxygen, and feeding rate / perfusion rate (culture solution input / discharge conditions) are set with the values input by the user. The turbulent kinetic energy dissipation rate is a value theoretically calculated from P / V (stirring power), or the average value + standard deviation value inside the device obtained from the multi-point measurement method in CFD simulation or experiment. The gas hold-up is a value calculated from the liquid level difference with / without aeration, or the average value + standard deviation value inside the device obtained from the multi-point measurement method in CFD simulation or experiment. The bubble diameter is a value measured by a photographed image or the like, or the average value + standard deviation value inside the device obtained from the multi-point measurement method in CFD simulation or experiment. The shear stress is set with the value input by the user or the average value + standard deviation value inside the device obtained from the multi-point measurement method in CFD simulation or experiment. Further, the explanatory variable of the kLa model 1A may include kLa approximately estimated by a conventional approximation formula for approximately deriving kLa. By including this as an explanatory variable, in the kLa model 1A, it is possible to evaluate taking into account the correlation between the kLa estimated by the approximation formula and the kLa output by the model 1A.

[0032] The learning data of kLa, which is the target variable of the kLa model 1A, is obtained by a known method. kLa may be determined experimentally, for example, in an actual machine (for example, measuring the change in the dissolved oxygen concentration in the liquid with an oxygen electrode provided in the container and analyzing the increase curve of the dissolved oxygen concentration, a static method), or may be determined by CFD simulation. Preferably, the learning data is preferably composed of data experimentally determined in an actual machine. However, when the data in the actual machine is insufficient, the learning data may include, in addition to the data experimentally determined in the actual machine, data determined by CFD simulation. Note that since the learned model according to the present invention is a model that replaces the conventional approximation formula for obtaining kLa, it is preferable that the kLa obtained by the conventional correlation formula without an experiment using an actual machine is not included in the learning data of the target variable of the kLa model 1A.

[0033] The P / V model 1B is a learned model that uses, as input data (explanatory variables) constituting learning data, data related to the shape of a stirrer provided in a container and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container, and uses kLa (stirring power per unit liquid volume) as output data (objective variable) constituting the learning data.

[0034] Examples of data related to the shape of the stirrer include data related to the shape of the stirring blades that make up the stirrer, such as the power number and the stirring blade diameter set according to the shape of the stirring blades. The data related to the shape of the stirring blades includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (e.g., 1) to the area occupied by the stirring blades in three-dimensional space and a second numerical value (e.g., 0) to the area not occupied by the stirring blades in three-dimensional space. Examples of data related to the physical properties, composition, chemical properties, or biochemical properties of the contents include the density of the culture solution set according to the culture solution.

[0035] Note that the P / V model 1B may be learned using, as input data (explanatory variables) constituting learning data, data related to the shape of a sparger provided in the container, data related to the shape of the container, the operating parameters of the stirrer, or the aeration conditions, and using P / V as output data (objective variable) constituting the learning data. Here, examples of the operating parameters of the stirrer include the rotation speed set by a value input by the user. Examples of the aeration conditions include the aeration flow rate set by a value input by the user. Also, the explanatory variables of the P / V model 1B may include P / V approximately estimated by a conventional approximation formula (such as the experimental formula of Michel-Miller) for approximately deriving P / V. By including this in the explanatory variables, in the P / V model 1B, it is possible to evaluate taking into account the correlation between the P / V estimated by the approximation formula and the P / V output by the model 1B.

[0036] The learning data of P / V, which is the target variable of the P / V model 1B, is obtained by a known method. For example, in a physical machine, the stirring power required for the operation of the stirrer is measured, and this is obtained by dividing it by the volume of the container. Similar to the kLa model 1A, it is preferable that the P / V obtained by the conventional correlation formula using a physical machine is not included in the learning data of the target variable.

[0037] The respiration rate model 1C is a learned model learned, for example, using data related to the shape of a stirrer provided in a container, data related to the shape of a sparger provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container as input data (explanatory variables) constituting the learning data, and the respiration rate as output data (target variable) constituting the learning data.

[0038] Examples of data related to the shape of the stirrer include data related to the shape of the stirring blades that make up the stirrer, and the type of stirring blades, the diameter of the stirring blades, and the number of stirring blades set according to the shape of the stirring blades. The data related to the shape of the stirring blades includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (for example, 1) to the area occupied by the stirring blades in three-dimensional space and a second numerical value (for example, 0) to the area not occupied by the stirring blades in three-dimensional space.

[0039] Examples of data related to the shape of the sparger provided in the container include the type of aeration sparger, the diameter of the aeration sparger, the pore diameter of the aeration sparger, and the number of pores of the aeration sparger set according to the shape of the sparger. The data related to the shape of the sparger includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (for example, 1) to the area occupied by the sparger in three-dimensional space and a second numerical value (for example, 0) to the area not occupied by the sparger in three-dimensional space.

[0040] Data on the shape of the container is data on the shape of the container itself that houses the contents. Examples of data on the shape of the container include the type of the shape of the container (such as cylindrical type, rectangular type), the dimensions of the container (inner diameter (in the case of a cylinder), width (in the case of a rectangle), height, etc.), the presence or absence of baffles (in the case of a stirred tank), and the like.

[0041] Examples of data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents include the physical properties, composition, chemical properties, and biochemical properties inherent to the materials contained in the contents, the physical properties of the entire contents in the container, and the concentration of each material contained in the contents.

[0042] Note that the respiration rate model 1C may be learned with the gas superficial velocity, the aeration rate, the turbulent energy dissipation rate, the gas hold-up, the bubble diameter, the control method of dissolved oxygen (DO) (such as upper and lower limit values), the rotation speed of the stirrer, the shear stress, the feeding rate / perfusion rate of the culture solution (in the case of fed-batch culture / perfusion culture) as the input data (explanatory variables) constituting the learning data, and the respiration rate as the output data (objective variable) constituting the learning data. By configuring the respiration rate model 1C to include one or more of these parameters in the input data, the prediction accuracy of the respiration rate can be further improved.

[0043] The superficial gas velocity is a set value calculated from the ventilation volume and the tank diameter input by the user. The ventilation volume, rotation speed, control method of dissolved oxygen, and feeding rate / perfusion rate (culture solution input / discharge conditions) are set with the values input by the user. The turbulent kinetic energy dissipation rate is a value theoretically calculated from P / V (stirring power), or a value of the average value + standard deviation within the device obtained from the multi-point measurement method in CFD simulation or experiment. The gas hold-up is a value calculated from the liquid level difference with / without ventilation, or a value of the average value + standard deviation within the device obtained from the multi-point measurement method in CFD simulation or experiment. The bubble diameter is a value measured by a photographed image or the like, or a value of the average value + standard deviation within the device obtained from the multi-point measurement method in CFD simulation or experiment. The shear stress is set with the value input by the user or set with a value of the average value + standard deviation within the device obtained from the multi-point measurement method in CFD simulation or experiment.

[0044] The learning data of the respiration rate, which is the target variable of the respiration rate model 1C, is obtained by a known method. For example, the learning data of the respiration rate is calculated from the oxygen concentration and exhaust gas concentration measured in the actual machine during operation.

[0045] The cell mass and cell concentration model 1D is a learned model learned with data on the shape of the stirrer provided in the container, data on the shape of the sparger provided in the container, data on the shape of the container, and data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container as input data (explanatory variables) constituting the learning data, and the cell mass and cell concentration as output data (target variables) constituting the learning data.

[0046] Examples of data related to the shape of the stirrer include, for example, data related to the shape of the stirring blades that make up the stirrer, such as the type of stirring blade, the diameter of the stirring blade, and the number of stirring blades, which are set according to the shape of the stirring blade. The data related to the shape of the stirring blade includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (for example, 1) to the area occupied by the stirring blade in three-dimensional space and a second numerical value (for example, 0) to the area not occupied by the stirring blade in three-dimensional space.

[0047] Examples of data related to the shape of the sparger provided in the container include, for example, the type of aeration sparger, the diameter of the aeration sparger, the pore diameter of the aeration sparger, and the number of pores of the aeration sparger, which are set according to the shape of the sparger. The data related to the shape of the sparger includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (for example, 1) to the area occupied by the sparger in three-dimensional space and a second numerical value (for example, 0) to the area not occupied by the sparger in three-dimensional space.

[0048] The data related to the shape of the container is data related to the shape of the container itself that houses the contents. Examples of the data related to the shape of the container include, for example, the type of container shape (such as cylindrical, rectangular, etc.), the dimensions of the container (inner diameter in the case of a cylinder, width in the case of a rectangle, height, etc.), and the presence or absence of baffles (in the case of a stirring tank).

[0049] Examples of data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents include, for example, the material-specific physical properties, composition, chemical properties, and biochemical properties contained in the contents, the physical properties of the entire contents in the container, and the concentration of each material contained in the contents.

[0050] Note that the cell and cell concentration model 1D may be trained using the superficial gas velocity, aeration rate, turbulent energy dissipation rate, gas hold-up, bubble diameter, control method of dissolved oxygen (DO) (upper and lower limit values, etc.), rotation speed of the stirrer, shear stress, feeding rate / perfusion rate of the culture solution (in the case of fed-batch culture / perfusion culture), addition timing of the fed-batch medium (in the case of fed-batch culture / perfusion culture), fed-batch culture method, number of culture days, cell (cell mass) survival rate, and viable cell (cell mass) density as input data (explanatory variables) that make up the learning data, and the cell and cell concentration as output data (objective variables) that make up the learning data.

[0051] The superficial gas velocity is a set value calculated from the aeration rate and tank diameter input by the user. The aeration rate, rotation speed, control method of dissolved oxygen, and feeding rate / perfusion rate are set with values input by the user. The turbulent energy dissipation rate is a value theoretically calculated from P / V (stirring power), or a value of the average value + standard deviation in the device obtained from a multi-point measurement method in CFD simulation or experiment. The gas hold-up is a value calculated from the liquid level difference with / without aeration, or a value of the average value + standard deviation in the device obtained from a multi-point measurement method in CFD simulation or experiment. The bubble diameter is a value measured by a photographed image or the like, or a value of the average value + standard deviation in the device obtained from a multi-point measurement method in CFD simulation or experiment. The shear stress may be set with a value input by the user, or set with a value of the average value + standard deviation in the device obtained from a multi-point measurement method in CFD simulation or experiment.

[0052] Furthermore, the cell and cell concentration model 1D may use, in addition to or instead of the cell and cell concentration, the concentration of the product from the cells and cell mass as the objective variable. The product from the cells and cell mass includes proteins, genes, etc. that are produced by the cells and cell mass and discharged outside the cells and cell mass. Proteins include target production substances (such as antibodies) and inhibitory substances that inhibit cell growth and the like.

[0053] The learning data of the objective variables of the cell and cell concentration model 1D, namely the cell and cell concentration, and / or the concentration of the product from the cells, is obtained by known methods. For example, these learning data include the cell and cell concentration, and / or the concentration of the product from the cells, which are measured by sampling some of the contents in an actual machine during operation or by an optical sensor.

[0054] The titer model 1E is a learned model learned with, for example, data on the shape of the stirrer provided in the container, data on the shape of the sparger provided in the container, data on the shape of the container, and data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container as input data (explanatory variables) constituting the learning data, and the titer as output data (objective variable) constituting the learning data.

[0055] Examples of the data on the shape of the stirrer include data on the shape of the stirring blades constituting the stirrer, and the type, diameter, and number of stirring blades set according to the shape of the stirring blades. The data on the shape of the stirring blades includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (for example, 1) to the area occupied by the stirring blades in the three-dimensional space and a second numerical value (for example, 0) to the area not occupied by the stirring blades in the three-dimensional space.

[0056] Examples of the data on the shape of the sparger provided in the container include the type, diameter, pore diameter, and number of pores of the aeration sparger set according to the shape of the sparger. The data on the shape of the sparger includes data with numerical labels for each type of shape, or data digitized by assigning a first numerical value (for example, 1) to the area occupied by the sparger in the three-dimensional space and a second numerical value (for example, 0) to the area not occupied by the sparger in the three-dimensional space.

[0057] Data on the shape of the container is data on the shape of the container itself that holds the contents. Examples of data on the shape of the container include the type of the container shape (such as cylindrical, rectangular, etc.), the dimensions of the container (inner diameter in the case of a cylinder, width in the case of a rectangle, height, etc.), the presence or absence of baffles (in the case of a stirred tank), and the like.

[0058] Examples of data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents include the physical properties, composition, chemical properties, and biochemical properties specific to the materials contained in the contents, the physical properties of the entire contents in the container, and the concentration of each material contained in the contents.

[0059] Note that the titer model 1E may be learned with the superficial gas velocity, aeration rate, turbulent energy dissipation rate, gas hold-up, bubble diameter, control method of dissolved oxygen (DO) (such as upper and lower limit values), rotation speed of the stirrer, feeding rate / perfusion rate of the culture solution (in the case of fed-batch culture / perfusion culture), shear stress, addition timing of the fed medium (in the case of fed-batch culture / perfusion culture), fed-batch culture method, number of culture days, cell (cell mass) survival rate, and viable cell (cell mass) density as input data (explanatory variables) constituting the learning data, and the titer as output data (objective variable) constituting the learning data.

[0060] The superficial gas velocity is a set value calculated from the ventilation rate and the tank diameter input by the user. The ventilation rate, rotation speed, control method of dissolved oxygen, and feeding rate / perfusion rate are set with the values input by the user. The turbulent kinetic energy dissipation rate is a value theoretically calculated from P / V (stirring power), or a value of the average value + standard deviation inside the device obtained from the multi-point measurement method in CFD simulation or experiment. The gas hold-up is a value calculated from the liquid level difference with / without ventilation, or a value of the average value + standard deviation inside the device obtained from the multi-point measurement method in CFD simulation or experiment. The bubble diameter is a value measured by a photographed image or the like, or a value of the average value + standard deviation inside the device obtained from the multi-point measurement method in CFD simulation or experiment. The shear stress is set with the value input by the user or set with a value of the average value + standard deviation inside the device obtained from the multi-point measurement method in CFD simulation or experiment.

[0061] The learning data of the titer, which is the objective variable of the titer model 1E, is obtained by a known method. For example, the learning data of the titer includes the titer values measured by sampling some of the contents in the actual machine during operation or measured by an optical sensor.

[0062] The operation reception unit 2 is composed of, for example, a keyboard and a mouse, and receives inputs of various operations by the user. In the present embodiment, the operation reception unit 2 selects the kLa model 1A, P / V model 1B, respiration rate model 1C, cell and cell concentration model 1D, or titer model 1E stored in the learned model storage unit 1, and receives from the user a setting operation for setting the input data (explanatory variable) to be input to the selected model. Then, the operation reception unit 2 outputs setting operation information indicating the content (model, input data) of the received setting operation to the explanatory variable input unit 3. Here, the operation reception unit 2 may receive the input data one set at a time, or may receive a plurality of data sets simultaneously. When receiving a plurality of data sets simultaneously, for example, by receiving the setting of the range of each input data included in the data set, it is configured to receive the numerical combinations of each input data included in the range.

[0063] When the model indicated by the setting operation information output from the operation reception unit 2 is the kLa model 1A, the explanatory variable input unit 3 inputs the input data indicated by the setting operation information into the kLa model 1A. Here, the input data input into the kLa model 1A is the input data that constitutes the learning data used for the learning of the kLa model 1A, that is, data related to the shape of the stirrer provided in the container, data related to the shape of the sparger provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container.

[0064] When the model indicated by the setting operation information output from the operation reception unit 2 is the P / V model 1B, the explanatory variable input unit 3 inputs the input data indicated by the setting operation information into the P / V model 1B. Here, the input data input into the P / V model 1B is the input data that constitutes the learning data used for the learning of the P / V model 1B, that is, data related to the shape of the stirrer provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container.

[0065] When the model indicated by the setting operation information output from the operation reception unit 2 is the respiration rate model 1C, the explanatory variable input unit 3 inputs the input data indicated by the setting operation information into the respiration rate model 1C. Here, the input data input into the respiration rate model 1C is the input data that constitutes the learning data used for the learning of the respiration rate model 1C, that is, data related to the shape of the stirrer provided in the container, data related to the shape of the container, data related to the shape of the sparger provided in the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container.

[0066] When the model indicated by the setting operation information output from the operation reception unit 2 is the cell mass and cell concentration model 1D, the explanatory variable input unit 3 inputs the input data indicated by the setting operation information into the cell mass and cell concentration model 1D. Here, the input data input into the cell mass and cell concentration model 1D is the input data constituting the learning data used for the learning of the cell mass and cell concentration model 1D, that is, data related to the shape of the stirrer provided in the container, data related to the shape of the sparger provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container.

[0067] When the model indicated by the setting operation information output from the operation reception unit 2 is the titer model 1E, the explanatory variable input unit 3 inputs the input data indicated by the setting operation information into the titer model 1E. Here, the input data input into the titer model 1E is the input data constituting the learning data used for the learning of the titer model 1E, that is, data related to the shape of the stirrer provided in the container, data related to the shape of the sparger provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container.

[0068] When input data is input into the kLa model 1A by the explanatory variable input unit 3, the objective variable acquisition unit 4 acquires the kLa output from the kLa model 1A. Then, the objective variable acquisition unit 4 outputs the information indicating the acquired kLa to the display unit 7 to display the kLa. As a result, the user referring to the display unit 7 can accurately evaluate the kLa, which is one of the index data related to the dynamics of the contents of the container, as compared with the prior art evaluated using the correlation formula. Further, the objective variable acquisition unit 4 outputs the information indicating the acquired kLa to the determination unit 5.

[0069] When input data is input into the P / V model 1B by the explanatory variable input unit 3, the objective variable acquisition unit 4 acquires the P / V output from the P / V model 1B. Then, the objective variable acquisition unit 4 outputs information indicating the acquired P / V to the display unit 7 to display the P / V. Thereby, the user referring to the display unit 7 can accurately evaluate the P / V, which is one of the index data related to the dynamics of the contents of the container, as compared with the prior art evaluated using the correlation formula. Further, the objective variable acquisition unit 4 outputs information indicating the acquired P / V to the determination unit 5.

[0070] When input data is input into the respiration rate model 1C by the explanatory variable input unit 3, the objective variable acquisition unit 4 acquires the respiration rate output from the respiration rate model 1C. Then, the objective variable acquisition unit 4 outputs information indicating the acquired respiration rate to the display unit 7 to display the respiration rate. Thereby, the user referring to the display unit 7 can accurately evaluate the respiration rate, which is one of the index data related to the dynamics of the contents of the container and was conventionally difficult to evaluate without relying on the operation of the actual machine. Further, the objective variable acquisition unit 4 outputs information indicating the acquired respiration rate to the determination unit 5.

[0071] When input data is input into the cell and cell concentration model 1D by the explanatory variable input unit 3, the objective variable acquisition unit 4 acquires the cell and cell concentration output from the cell and cell concentration model 1D. Then, the objective variable acquisition unit 4 outputs information indicating the acquired cell and cell concentration to the display unit 7 to display the cell and cell concentration. Thereby, the user referring to the display unit 7 can accurately evaluate the cell and cell concentration, which is one of the index data related to the dynamics of the contents of the container and was conventionally difficult to evaluate without relying on the operation of the actual machine. Further, the objective variable acquisition unit 4 outputs information indicating the acquired cell and cell concentration to the determination unit 5.

[0072] When input data is input to the titer model 1E by the explanatory variable input unit 3, the objective variable acquisition unit 4 acquires the titer output from the titer model 1E. Then, the objective variable acquisition unit 4 outputs information indicating the acquired titer to the display unit 7 to display the titer. As a result, a user referring to the display unit 7 can accurately evaluate the titer, which is one of the index data related to the dynamics of the contents of the container that was conventionally difficult to evaluate without relying on the operation of the actual machine. Also, the objective variable acquisition unit 4 outputs information indicating the acquired titer to the determination unit 5.

[0073] The determination unit 5 determines whether or not the value of kLa output from the objective variable acquisition unit 4 becomes a predetermined value, and outputs information indicating the determination result to the explanatory variable acquisition unit 6. Here, the predetermined value is, for example, a desired value of kLa designed so that the kLa in the commercial device becomes equivalent to that in the laboratory scale when scaling up from a small laboratory scale to a large commercial scale, and it may be one value or a plurality of values.

[0074] The determination unit 5 determines whether or not the value of P / V output from the objective variable acquisition unit 4 becomes a predetermined value, and outputs information indicating the determination result to the explanatory variable acquisition unit 6. Here, the predetermined value is, for example, a desired value of P / V designed so that the P / V in the commercial device becomes equivalent to that in the laboratory scale when scaling up from a small laboratory scale to a large commercial scale, and it may be one value or a plurality of values.

[0075] The determination unit 5 determines whether or not the value of the respiration rate output from the objective variable acquisition unit 4 becomes a predetermined value, and outputs information indicating the determination result to the explanatory variable acquisition unit 6. Here, the predetermined value is, for example, a desired value of the respiration rate designed so that the respiration rate in the commercial device becomes equivalent to that in the laboratory scale when scaling up from a small laboratory scale to a large commercial scale, and it may be one value or a plurality of values.

[0076] The determination unit 5 determines whether the values of the cell mass and cell concentration output from the objective variable acquisition unit 4 reach a predetermined value, and outputs information indicating the determination result to the explanatory variable acquisition unit 6. Here, the predetermined value is, for example, a desired value of the cell mass and cell concentration designed such that the cell mass and cell concentration in a commercial device become equivalent to those in a laboratory scale when scaling up from a small laboratory scale to a large commercial scale, and it may be one value or a plurality of values.

[0077] The determination unit 5 determines whether the value of the titer output from the objective variable acquisition unit 4 reaches a predetermined value, and outputs information indicating the determination result to the explanatory variable acquisition unit 6. Here, the predetermined value is, for example, a desired value of the titer designed such that the kLa in a commercial device becomes equivalent to that in a laboratory scale when scaling up from a small laboratory scale to a large commercial scale, and it may be one value or a plurality of values.

[0078] The explanatory variable acquisition unit 6 refers to the determination result output from the determination unit 5. When the value of kLa output from the objective variable acquisition unit 4 reaches a predetermined value, it acquires the input data (explanatory variables) input to the kLa model 1A when the value of kLa was output from the kLa model 1A, and outputs information indicating the acquired input data to the display unit 7 to display the input data. Thereby, the user referring to the display unit 7 can easily grasp the input data without groping for the input data for which the value of kLa becomes the desired value.

[0079] The explanatory variable acquisition unit 6 refers to the determination result output from the determination unit 5. When the value of P / V output from the objective variable acquisition unit 4 reaches a predetermined value, it acquires the input data (explanatory variables) input to the P / V model 1B when the value of P / V was output from the P / V model 1B, and outputs information indicating the acquired input data to the display unit 7 to display the input data. Thereby, the user referring to the display unit 7 can easily grasp the input data without groping for the input data for which the value of P / V becomes the desired value.

[0080] The explanatory variable acquisition unit 6 refers to the determination result output from the determination unit 5, and when the value of the respiratory rate output from the objective variable acquisition unit 4 becomes a predetermined value, it acquires the input data (explanatory variable) input to the respiratory rate model 1C when the value of the respiratory rate is output from the respiratory rate model 1C, and outputs information indicating the acquired input data to the display unit 7 to display the input data. As a result, the user referring to the display unit 7 can easily grasp the input data without groping for the input data for which the value of the respiratory rate becomes a desired value.

[0081] The explanatory variable acquisition unit 6 refers to the determination result output from the determination unit 5, and when the values of the cell mass and cell concentration output from the objective variable acquisition unit 4 become a predetermined value, it acquires the input data (explanatory variable) input to the cell mass and cell concentration model 1D when the values of the cell mass and cell concentration are output from the cell mass and cell concentration model 1D, and outputs information indicating the acquired input data to the display unit 7 to display the input data. As a result, the user referring to the display unit 7 can easily grasp the input data without groping for the input data for which the values of the cell mass and cell concentration become a desired value.

[0082] The explanatory variable acquisition unit 6 refers to the determination result output from the determination unit 5, and when the value of the titer output from the objective variable acquisition unit 4 becomes a predetermined value, it acquires the input data (explanatory variable) input to the titer model 1E when the value of the titer is output from the titer model 1E, and outputs information indicating the acquired input data to the display unit 7 to display the input data. As a result, the user referring to the display unit 7 can easily grasp the input data without groping for the input data for which the value of the titer becomes a desired value.

[0083] The display unit 7 is configured by, for example, a touch panel display provided in the estimation device 10, and displays various information output from the objective variable acquisition unit 4 and the explanatory variable acquisition unit 6.

[0084] (Estimation process performed by the estimation device 10) Figure 2 is a flowchart showing an example of the estimation process performed by the estimation device 10 in the present embodiment. The process of step S100 in Figure 2 starts when the operation reception unit 2 receives from the user a setting operation for selecting the kLa model 1A, P / V model 1B, respiration rate model 1C, cell mass and cell concentration model 1D, or titer model 1E stored in the learned model storage unit 1 and setting the input data (explanatory variables) to be input to the selected model.

[0085] First, the explanatory variable input unit 3 inputs the input data set via the operation reception unit 2 to the learned model (kLa model 1A, P / V model 1B, respiration rate model 1C, cell mass and cell concentration model 1D, or titer model 1E) selected via the operation reception unit 2 (step S100).

[0086] Next, the objective variable acquisition unit 4 acquires the output data (objective variable: kLa, P / V, respiration rate, cell mass and cell concentration, concentration of the product from the cells and cells, or titer) output from the learned model when the input data is input to the learned model by the explanatory variable input unit 3, and causes the display unit 7 to display the acquired output data (step S110).

[0087] Next, the determination unit 5 determines whether the value of the output data acquired by the objective variable acquisition unit 4 becomes a predetermined value (step S120). As a result of the determination, if the value of the output data does not become the predetermined value (step S120, NO), the process transitions to step S140.

[0088] On the other hand, when the value of the output data becomes the predetermined value (step S120, YES), the explanatory variable acquisition unit 6 acquires the input data (explanatory variables) input to the learned model when the value of the output data was output from the learned model, and causes the display unit 7 to display the acquired input data (step S130). Thereafter, the process transitions to step S140.

[0089] In step S140, the explanatory variable input unit 3 determines whether the input of the input data set via the operation reception unit 2 to the learned model (kLa model 1A, P / V model 1B, respiration rate model 1C, cell and cell concentration model 1D, or titer model 1E) selected via the operation reception unit 2 has been completed. As a result of the determination, if the input of the input data has not been completed (step S140, NO), the process returns to before step S100. On the other hand, if the input of the input data has been completed (step S140, YES), the estimation device 10 ends the estimation process shown in FIG. 2.

[0090] (Effect of this Embodiment) As described in detail above, in this embodiment, the estimation device 10 includes an explanatory variable input unit 3 and an objective variable acquisition unit 4. The explanatory variable input unit inputs explanatory variables (input data) to the learned model (kLa model 1A, P / V model 1B, respiration rate model 1C, cell and cell concentration model 1D, or titer model 1E). When the explanatory variable is input to the learned model, the objective variable acquisition unit 4 acquires the objective variable (output data) output from the learned model. The learned model uses, as explanatory variables, at least one of data on the shape of a stirrer provided in the container when gas is blown into the liquid in the container, data on the shape of a sparger provided in the container, data on the shape of the container, physical properties, composition, chemical properties, biochemical properties, or concentration of the contents contained in the container, and is learned with the mass transfer coefficient (kLa), stirring power (P / V), respiration rate, cell and cell concentration, concentration of a product from the cells and cells, or titer as the objective variable.

[0091] According to the present embodiment configured as described above, at least one of the data related to the shape of the stirrer, the data related to the shape of the sparger, the data related to the shape of the container, the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents (explanatory variable) is input into the learned model, and the mass transfer coefficient (kLa), the stirring power per unit volume (P / V), the respiration rate, the cell and cell concentration, the concentration of the product from the cells and cells, or the titer output as the target variable can be obtained. Therefore, compared with the prior art, the index data related to the dynamics of the contents of the container can be accurately evaluated. For example, the conventional correlation equations for the mass transfer coefficient (kLa) and the stirring power per unit volume (P / V) only consider some parameters such as the aeration rate and the rotation speed of the stirrer, and do not take into account the effects related to the shapes of the stirrer and the sparger, and the physical properties, composition, chemical properties, biochemical properties, and concentration of the contents of the container. Therefore, in the scale-up design of the device, even if the calculation by the correlation equation results in the mass transfer coefficient (kLa) and the stirring power per unit volume (P / V) satisfying the predetermined values, in the operation of the actual machine designed accordingly, the mass transfer coefficient (kLa) and the stirring power per unit volume (P / V) may not satisfy the predetermined values. In contrast, the present invention calculates these parameters by a machine learning model that takes into account the shapes of the stirrer, the sparger, and the container, which have complex effects as the device is scaled up, and the physical properties, composition, chemical properties, biochemical properties, and concentration of the contents of the container, so that the accurate mass transfer coefficient (kLa) and the stirring power per unit volume (P / V) can be calculated.

[0092] Further, according to the present embodiment, at least one of data related to the shape of the stirrer, data related to the shape of the sparger, data related to the shape of the container, physical properties, composition, chemical properties, biochemical properties, or concentration of the content (explanatory variable) is input into the learned model, and the respiration rate, the concentration of cells and bacteria, the concentration of the product from the cells and bacteria, or the titer can be obtained as the target variable. These parameters were conventionally difficult to estimate their values at the stage of designing the actual machine without relying on the operation of the actual machine. In the invention of the present application, since these parameters can be estimated at the time of design, it becomes possible to design a device that more surely exhibits desired performance in the operation of the actual machine.

[0093] Furthermore, the learned model according to the present invention can be used not only for design but also for the purpose of estimating the dynamics of the content during the operation of the actual machine. Regarding kLa, P / V, respiration rate, cell and bacteria concentration, concentration of the product from the cells and bacteria, and titer indicating the dynamics of the content, their values fluctuate after the start of operation. In order to estimate this in real time, the learned model of the present invention can be used. In this case, it may include the same parameters as those described above. When the learned model according to the present invention is used for the purpose of estimating the dynamics of the content during the operation of the actual machine, it is preferable to include in the explanatory variable of the learned model the physical properties of the entire content in the container obtained during the operation of the actual machine and the measured values of the concentration of each material contained in the content. By continuously acquiring these over the course of operation and inputting them into the learned model, it becomes possible to more accurately estimate the current dynamics. Furthermore, as the explanatory variable, it may include a captured image from a side window or an upper window of the container, a measured value by an odor sensor related to the content, a measured value related to the vibration of the container and the accompanying sound, the survival rate of cells and bacteria, or the density of living cells and bacteria. When the captured image is included as the explanatory variable, for example, information related to the color of the content in the captured image (e.g., RGB value) is extracted and used as the explanatory variable, so that the correlation between the color of the content and the index related to the dynamics of the content can be reflected in the learned model.

[0094] (Modification example) In the above-described embodiment, among the first data (for example, data related to the shape of the stirrer) and the second data (for example, data related to the shape of the sparger) that constitute the input data (explanatory variable) input to the learned model, the value of the first data may be a fixed value, and the value of the second data may be a variable value. In this case, when the value of the output data (objective variable) output from the learned model becomes a predetermined value, the explanatory variable acquisition unit 6 acquires the value of the second data input to the learned model. As a result, the user can easily acquire the second data without groping for the second data for which the value of the output data output from the learned model becomes a desired value while keeping the value of the first data fixed. This embodiment is useful, for example, when there are restrictive conditions such as when designing a device that cannot change the stirrer to be used and only the shape of the sparger can be varied to satisfy a desired kLa value during design.

[0095] Also, in the above-described embodiment, the operation reception unit 2 selects a plurality of models from among the kLa model 1A, P / V model 1B, respiration rate model 1C, cell mass and cell concentration model 1D, and titer model 1E stored in the learned model storage unit 1, and receives from the user a setting operation for setting input data (explanatory variables) to be input to the selected plurality of models. Here, the input data may receive the input data for each of the plurality of models separately, or may receive the input data common to the plurality of models collectively. Then, the operation reception unit 2 outputs setting operation information indicating the content (model, input data) of the received setting operation to the explanatory variable input unit 3, the objective variable acquisition unit 4 acquires the output of each of the plurality of models, and the determination unit 5 determines whether or not the output of each of the plurality of models becomes a respective predetermined value. And the explanatory variable acquisition unit 6 may acquire the explanatory variables when the outputs of all of the plurality of models become predetermined values, or may acquire the explanatory variables when a part of the outputs of the plurality of models becomes a predetermined value. However, in the scale-up design of the apparatus, since it is preferable to design such that all the outputs become predetermined values, the explanatory variable acquisition unit 6 preferably acquires the explanatory variables when the outputs of all of the plurality of models become predetermined values. For example, when the plurality of models are the kLa model 1A and the P / V model 1B, the explanatory variable acquisition unit 6 may be configured to acquire the explanatory variables when the output of the kLa model 1A becomes a predetermined value and the output of the P / V model 1B becomes a predetermined value. In the scale-up design of the apparatus, since it is particularly important whether or not the values of kLa and P / V become predetermined values, the plurality of models preferably include at least the kLa model 1A and the P / V model 1B. More preferably, in order to ensure that the designed apparatus achieves a desired operation, the plurality of models preferably include all of the kLa model 1A, P / V model 1B, respiration rate model 1C, cell mass and cell concentration model 1D, and titer model 1E.

[0096] Also, in the above embodiment, the objective variable of the first model selected from the kLa model 1A, P / V model 1B, respiration rate model 1C, cell mass and cell concentration model 1D, and titer model 1E may be included in the explanatory variables of the second model selected from the kLa model 1A, P / V model 1B, respiration rate model 1C, cell mass and cell concentration model 1D, and titer model 1E. kLa, P / V, respiration rate, cell mass and cell concentration, the concentration of the product from the cell mass and cells, and the titer, which are index data representing the dynamics of the contents, are all correlated parameters. Therefore, by adopting the index estimated by one model as the explanatory variable of the model with other indexes as the objective variable, the objective variable can be estimated more accurately. For example, as described above, kLa is an index indicating the mass transfer rate of gas into the liquid. However, when the mass transfer rate increases, the intake of air by the cells and cells in the liquid is promoted, the respiration rate increases, and as a result, the cell mass and cell concentration, the concentration of the product from the cell mass and cells, and the titer may also increase. Therefore, kLa (the objective variable of the first model) estimated by the kLa model 1A may be included as the explanatory variable (the explanatory variable of the second model) of the respiration rate model 1C, the cell mass and cell concentration model 1D, or the titer model 1E. Similarly, the value of the objective variable of the P / V model 1B may be included in the explanatory variables of the kLa model 1A, the respiration rate model 1C, the cell mass and cell concentration model 1D, or the titer model 1E, or the value of the objective variable of the respiration rate model 1C may be included in the explanatory variables of the cell mass and cell concentration model 1D, or the titer model 1E.

[0097] In addition, in the above-described embodiments, a culture tank with aeration and agitation has been described, but the object of the learned model according to the present invention is not limited thereto. For example, a culture tank without aeration or agitation (e.g., a shaking culture tank) may be targeted. In this case, as explanatory variables of Models 1A to 1E, data related to the shape of the stirrer and data related to the shape of the sparger provided in the container are not included, and data related to the shape of the container and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container may be included. Further, a culture tank without aeration but with agitation may be targeted. In this case, as explanatory variables of Models 1A to 1E, data related to the shape of the sparger provided in the container is not included, and data related to the shape of the stirrer, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container may be included. Further, a culture tank with aeration but without agitation may be targeted. In this case, as explanatory variables of Models 1A to 1E, data related to the shape of the stirrer is not included, and data related to the shape of the sparger provided in the container, data related to the shape of the container, and data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container may be included. Further, a reactor (such as a bubble column) that performs a chemical reaction (such as a chemical synthesis reaction or a chemical decomposition reaction) inside may be the object of the learned model. In this case as well, it is advisable to set the explanatory variables according to the presence or absence of a stirrer and the presence or absence of a sparger. In this case, as data related to the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container, in particular, chemical properties specific to the materials contained in the contents (e.g., reaction activity, activation energy, reaction rate, reaction rate constant in the chemical reaction of the materials) may be included.

[0098] In addition, in the above embodiment, the output data (objective variable: kLa, P / V, respiration rate, cell and cell concentration, concentration of the product from the cells and cells, or titer) output from the learned model (kLa model 1A, P / V model 1B, respiration rate model 1C, cell and cell concentration model 1D, or titer model 1E) may change over time.

[0099] Figure 3 is an image diagram showing the change in the value of the output data according to the number of culture days. In Figure 3, curve L1 shows the change in the value of the output data (kLa) output from the kLa model 1A according to the number of culture days (in the illustrated example, 1 day to 7 days). Curve L2 shows the change in the value of the output data (P / V) output from the P / V model 1B according to the number of culture days. Curve L3 shows the change in the value of the output data (respiration rate) output from the respiration rate model 1C according to the number of culture days. Curve L4 shows the change in the value of the output data (cell concentration) output from the cell and cell concentration model 1D according to the number of culture days. Curve L5 shows the change in the value of the output data (titer) output from the titer model 1E according to the number of culture days. As shown in Figure 3, the values of the output data (P / V, particularly cell concentration, titer) change significantly according to the number of culture days.

[0100] Considering the change in the value of the output data shown in Figure 3, the value of the output data (kLa, P / V, respiration rate, cell and cell concentration, concentration of the product from the cell and cell, or titer) output from the learned model (kLa model 1A, P / V model 1B, respiration rate model 1C, cell and cell concentration model 1D, or titer model 1E) may be the value when the number of culture days is the final day (7 days), or may be the value when the number of culture days is the intermediate day (4 days), or may be both the value when the number of culture days is the final day (7 days) and the value when the number of culture days is the intermediate day (4 days), or may be the average value for the entire period (7 days).

[0101] In addition, the kLa model 1A may be learned using the number of culture days, in addition to data related to the shape of the stirrer, data related to the shape of the sparger, and data related to the physical properties, composition, chemical properties, or biochemical properties of the contents, as input data (explanatory variables) constituting the learning data, and kLa (mass transfer volume coefficient) as output data (objective variable) constituting the learning data. In this case, the explanatory variable input unit 3 inputs the input data including the number of culture days into the kLa model 1A.

[0102] In addition, the P / V model 1B may be trained using the number of days of culture, in addition to data on the shape of the stirrer and data on the physical properties, composition, chemical properties, or biochemical properties of the contents, as input data (explanatory variables) that make up the training data, and using P / V as output data (objective variables) that make up the training data. In this case, the explanatory variable input unit 3 inputs the input data including the number of days of culture into the P / V model 1B.

[0103] In addition, the respiration rate model 1C may be trained using the number of days of culture, in addition to data on the shape of the stirrer, data on the shape of the sparger, data on the shape of the container, and data on the physical properties, composition, chemical properties, or biochemical properties of the contents, as input data (explanatory variables) that make up the training data, and using the respiration rate as output data (objective variables) that make up the training data. In this case, the explanatory variable input unit 3 inputs the input data including the number of days of culture into the respiration rate model 1C.

[0104] In addition, the cell mass and cell concentration model 1D may be trained using the number of days of culture, in addition to data on the shape of the stirrer, data on the shape of the sparger, data on the shape of the container, and data on the physical properties, composition, chemical properties, or biochemical properties of the contents, as input data (explanatory variables) that make up the training data, and using the cell mass and cell concentration and the product concentration from the cell mass and cells as output data (objective variables) that make up the training data. In this case, the explanatory variable input unit 3 inputs the input data including the number of days of culture into the cell mass and cell concentration model 1D.

[0105] In addition, the titer model 1E may be trained using the number of days of culture, in addition to data on the shape of the stirrer, data on the shape of the sparger, data on the shape of the container, and data on the physical properties, composition, chemical properties, or biochemical properties of the contents, as input data (explanatory variables) that make up the training data, and using the titer as output data (objective variables) that make up the training data. In this case, the explanatory variable input unit 3 inputs the input data including the number of days of culture into the titer model 1E.

[0106] In addition, taking into consideration that the values ​​of the output data (kLa, P / V, respiration rate) change slightly depending on the number of culture days, mainly due to the influence of data regarding the physical properties, composition, chemical properties, or biochemical properties of the contents, data regarding the physical properties, composition, chemical properties, or biochemical properties of the contents expected during the expected number of culture days can be input into the trained models (kLa model 1A, P / V model 1B, respiration rate model 1C), and it can be confirmed whether the values ​​of the output data (kLa, P / V, respiration rate) output from the trained models satisfy the specified values.

[0107] In addition, the above-mentioned embodiments are merely examples of the embodiment of the present invention, and the technical scope of the present invention should not be interpreted as being limited by these. In other words, the present invention can be embodied in various forms without departing from the gist or main characteristics of the present invention. [Explanation of symbols]

[0108] 1: trained model storage device, 1A: kLa model, 1B: P / V model, 1C: respiration rate model, 1D: bacterial cell concentration model, 1E: titer model, 2: operation reception unit, 3: explanatory variable input unit, 4: objective variable acquisition unit, 5: judgment unit, 6: explanatory variable acquisition unit, 7: display unit, 10: estimation device

Claims

1. A learned model, using, as explanatory variables, at least one of data on the shape of a stirrer provided in a container, data on the shape of a sparger provided in the container, data on the shape of the container, and data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container, and being a learned model learned with a mass transfer coefficient (kLa), stirring power per unit volume (P / V), respiration rate, cell concentration, product concentration from cells, or titer as an objective variable.

2. The learned model according to Claim 1, using, as explanatory variables, data on the shape of the stirrer, data on the shape of the sparger, data on the shape of the container, and data on the physical properties, composition, chemical properties, biochemical properties, or concentration of the contents accommodated in the container.

3. The learned model according to Claim 1, wherein the data on the shape of the stirrer includes data on the shape of a stirring blade constituting the stirrer, and the data on the shape of the stirring blade and the data on the shape of the sparger include data with numerical labels assigned to each type of shape, or include digitized data by assigning a first numerical value to a region occupied by the stirring blade and the sparger in a three-dimensional space and a second numerical value to a region not occupied by the stirring blade and the sparger in the three-dimensional space. A learned model.

4. An estimation device including an input unit and an objective variable acquisition unit, wherein the input unit inputs the explanatory variables to the learned model according to any one of Claims 1 to 3, and the objective variable acquisition unit acquires the objective variable output from the learned model when the explanatory variables are input to the learned model. An estimation device.

5. The estimation device according to Claim 4, further including an explanatory variable acquisition unit, wherein the explanatory variable acquisition unit acquires the value of the explanatory variable input to the learned model when the value of the objective variable output from the learned model becomes a predetermined value. An estimation device.

6. The estimation device according to Claim 5, wherein, among first data and second data constituting the explanatory variables, the value of the first data is a fixed value and the value of the second data is a variable value. The explanatory variable acquisition unit acquires the value of the second data input to the learned model when the value of the target variable output from the learned model becomes a predetermined value. Estimation device. **Claim 7** An estimation method, comprising: inputting the explanatory variable into the learned model according to any one of claims 1 to 3; when the explanatory variable is input to the learned model, acquiring the target variable output from the learned model. Estimation method.

Images