Cell culture optimization condition acquisition system, cell culture optimization condition acquisition method, and program

Abstract

To provide a cell culture optimization condition acquisition system that reduces the risk of contamination through collaboration between an operator and a machine, and that acquires an optimization condition under which cell culture is performed in response to physical stimuli and environmental stimuli in repeated cell culture.SOLUTION: The present invention is a system for acquiring a condition for optimizing an environment in a cell culture from culture-related data indicating an environment in which the cell culture is performed in each predetermined observation cycle in the cell culture. The system comprises: a related data acquisition part that acquires culture-related data; an activity degree data acquisition part that detects a cell activity degree for determining activity of cultured cells at each predetermined cycle; a factor parameter extraction part that extracts, as a factor parameter, culture-related data having a higher degree of influence than other culture-related data, which degree of influence is a factor of a change in the cell activity degree; and an optimum data extraction part that extracts, as optimum data, data of a factor parameter whose cell activity degree is equal to or greater than a predetermined value.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a cell culture optimization condition acquisition system, a cell culture optimization condition acquisition method, and a program.

Background Art

[0002] Conventionally, in cell culture during research and development, a culture process for culturing cells is evaluated based on evaluation indices such as the survival rate and growth efficiency of the cultured cells. Therefore, regarding the optimality of each operation in the culture process, when the above evaluation index that is optimal in the operation performed by the operator is obtained, it is determined as the processing condition. Then, each process of the cell process is carried out according to the determined processing condition, and cell culture of a predetermined cell is performed.

[0003] However, when cell culture is performed according to the culture process optimized for each operator, it is known that the results of the evaluation index vary depending on the qualifications of each operator. For this reason, in recent years, an automatic culture apparatus that performs stable processing using a work robot that mimics the processing in the culture process of an operator has been developed (see, for example, Patent Document 1, Patent Document 2, and Patent Document 3).

[0004] The apparatus described in Patent Document 1 is installed in an incubator, and the apparatus sequentially performs cell culture processes such as seeding, detachment, and recovery in a closed container according to a predetermined work protocol. The apparatus described in Patent Document 2 sequentially performs cell culture processes such as detachment and recovery according to a predetermined work protocol without the operator directly touching the culture dish for cell culture in order to reduce the risk of contamination. The apparatus described in Patent Document 3 captures an image of cells during cell culture in a culture dish, determines the culture state from the shape of the cells by image processing of the captured image, and evaluates whether to continue the culture.

Prior Art Documents

Patent Documents

[0005] [Patent Document 1] Japanese Patent Publication No. 2008-079544 [Patent Document 2] Japanese Patent Publication No. 2007-185165 [Patent Document 3] Japanese Patent Publication No. 2018-198605 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] The apparatus described in Patent Document 1 and Patent Document 2 above mechanizes cell culture by simulating the work of a skilled worker when manufacturing a work robot. However, when attempting to simulate the cell culture process performed by a robotic operator as if it were performed by a skilled worker, a problem arises due to the wide variation in the skill levels of the simulated operators.

[0007] In other words, the results of cell culture performed by the robot satisfy the optimization conditions of a simulated expert, but these expert optimization conditions do not necessarily coincide with the optimization for cell culture itself. Patent Document 1 and Patent Document 2 each describe a robotic machine that performs the same task according to a pre-set work protocol, and do not control the cell culture itself.

[0008] Furthermore, Patent Document 3 describes an image analysis of images of cells being cultured, and determines the culture status based on the shape of the cells. However, without information on the culture conditions that determine the cell shape, such as physical stimuli applied during the culture process (acceleration, vibration, etc.) and environmental stimuli in the culture environment (temperature, humidity, and light intensity, etc.), it is possible to determine the quality of the cells to be cultured, but it is not possible to determine the optimal conditions for cell culture processing.

[0009] This invention has been made in view of these circumstances, and aims to provide a cell culture optimization condition acquisition system, a cell culture optimization condition acquisition method, and a program that reduce the risk of contamination through cooperation between the operator and the machine, and acquire optimal conditions for cell culture in response to the physical and environmental stimuli mentioned above in repeatedly performed cell cultures. [Means for solving the problem]

[0010] This invention was made to solve the above-mentioned problems, and the cell culture optimization condition acquisition system of the present invention is a cell culture optimization condition acquisition system that acquires conditions for optimizing the environment in cell culture from culture-related data, which is environmental data of the environment in which the cell culture is performed, at each predetermined observation cycle in cell culture, and is characterized by comprising: a related data acquisition unit that acquires the culture-related data; an activity data acquisition unit that detects the cell activity level to determine the activity level of cultured cells at predetermined cycles; a factor parameter extraction unit that extracts culture-related data as factor parameters whose influence on the change in the cell activity level is higher than that of other culture-related data; and an optimal data extraction unit that extracts data of factor parameters when the cell activity level is above a predetermined threshold as optimal data.

[0011] The cell culture optimization condition acquisition system of the present invention is characterized by further comprising: an image acquisition unit that acquires observation images of cells in the cell culture in each of the observation cycles; and an image learning unit that generates a machine learning model for estimating the cell activity level from the observation images.

[0012] The cell culture optimization condition acquisition system of the present invention is characterized in that the image acquisition learning unit generates a machine learning model MA that estimates whether all of the cell activity levels are above a predetermined threshold, and a machine learning model MB that estimates whether any of the cell activity levels are above a predetermined threshold.

[0013] The cell culture optimization condition acquisition system of the present invention is characterized in that the factor parameter extraction unit extracts the combination of culture-related data that maximizes the distance between the centroids of the numerical coordinate groups in each of the groups with different cell activity levels as the combination of factor parameters.

[0014] The cell culture optimization condition acquisition system of the present invention is characterized in that the factor parameter extraction unit determines the degree of difference in culture-related data within a group as the contribution to the activity difference, which is the difference in the cell activity levels of each group with different cell activity levels, and extracts culture-related data having a greater contribution than the contribution of other culture-related data as the factor parameter.

[0015] The present invention provides a method for obtaining cell culture optimization conditions, which, at each predetermined observation cycle in cell culture, obtains conditions for optimizing the environment in the cell culture from culture-related data, which are environmental data of the environment in which the cell culture is being performed. The method is characterized by including: a related data acquisition unit which performs a related data acquisition process to acquire the culture-related data; an activity data acquisition unit which performs an activity data acquisition process to detect cell activity for determining the activity level of cultured cells at predetermined cycles; a factor parameter extraction unit which extracts culture-related data that have a higher influence on the change in cell activity level compared to other culture-related data as factor parameters; and an optimal data extraction unit which extracts data of factor parameters when the cell activity level is above a predetermined threshold as optimal data.

[0016] The program of the present invention is a program for operating a computer as a cell culture optimization condition acquisition system that acquires conditions for optimizing the environment in the cell culture from culture-related data, which is environmental data of the environment in which the cell culture is being performed, in each of predetermined observation cycles in the cell culture. The program causes the computer to function as related data acquisition means for acquiring the culture-related data, activity data acquisition means for detecting cell activity for determining the activity of the cultured cells at predetermined intervals, factor parameter extraction means for extracting, as factor parameters, culture-related data having a higher influence degree that is a factor of the change in the cell activity compared to other culture-related data, and optimal data extraction means for extracting, as optimal data, data of the factor parameters when the cell activity is equal to or higher than a predetermined threshold value.

Advantages of the Invention

[0017] According to this invention, it is possible to provide a cell culture optimization condition acquisition system, a cell culture optimization condition acquisition method, and a program that reduce the risk of contamination through cooperation between an operator and a machine, and acquire optimization conditions for performing cell culture in response to the physical stimulation and environmental stimulation in repeatedly performed cell culture.

Brief Description of the Drawings

[0018] [Figure 1] It is a schematic diagram showing a configuration example of a room for performing data acquisition in which a cell culture optimization condition acquisition system according to an embodiment of the present invention is arranged. [Figure 2] It is a block diagram showing a configuration example of the cell culture optimization condition acquisition system 200 in the present embodiment. [Figure 3] It is a diagram showing a configuration example of a cell observation table in the data management storage unit 209. [Figure 4] It is a diagram showing a configuration example of a culture-related data table in the data management storage unit 209. [Figure 5] It is a diagram showing a configuration example of a robot hand operation data table in the data management storage unit 209. [Figure 6] It is a conceptual diagram showing an example of a cycle for obtaining a determination result of culturing of cultured cells and an observation imaging image by cell observation. [Figure 7] As an example of cell observation, it is a conceptual diagram explaining an example of acquisition of culture-related data in cell detachment, a determination result of cell viability, and an observation imaging image.

Embodiments for Carrying Out the Invention

[0019] Hereinafter, a cell culture optimization condition acquisition system according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a schematic diagram showing a configuration example of a room for data acquisition in which a cell culture optimization condition acquisition system according to an embodiment of the present invention is arranged. In FIG. 1, inside a predetermined room 1 used as a sealed incubator, there are provided a robot hand 101, an installation table 102, a temperature sensor 103, a humidity sensor 104, an acceleration sensor 105, a CO2 concentration sensor 106, an illuminance sensor 107, an atmospheric pressure sensor 108, an imaging device 109, a syringe device 110, and a cell culture optimization condition acquisition system 200.

[0020] The installation table 102 is equipped with a vibration table 102S on which a culture vessel 300 is placed during the culture period. The vibration table 102S is vibrated (in the vertical direction or in the direction of an arc with a predetermined radius) at an arbitrary vibration frequency set by a vibration device provided on the installation table 102. The culture vessel 300 contains a cultured cell group together with a culture medium and is placed on the upper surface of the vibration table 102S.

[0021] When exchanging the culture medium of the culture vessel 300, the robot hand 101 grips and tilts the culture vessel 300 to discard the old culture medium at a discharge port or the like. The imaging device 109 images a cell region at a predetermined position in the cultured cell group in the culture vessel 300 to perform imaging of an observation imaging image. The syringe device 110 injects a new culture medium or a nutrient solution containing nutrients into the culture vessel 300 at a predetermined temperature and flow rate. Each of the above-mentioned imaging device 109 and syringe device 110 is attached to a support 111S which is attached to a fixed base 111.

[0022] The temperature sensor 103 is placed near the culture vessel 300 and measures the temperature of the area where the culture vessel 300 is located. The humidity sensor 104 is placed near the culture vessel 300 and measures the humidity in the area where the culture vessel 300 is located. The acceleration sensor 105 is attached to the vibration table 102S and measures the acceleration of the vibration table 102S, outputting the frequency and amplitude of vibrations in a predetermined period (for example, 1 second) as measurement results.

[0023] The CO2 concentration sensor 106 is placed near the culture vessel 300 and measures the CO2 concentration in the area where the culture vessel 300 is located. The illuminance sensor 107 is placed near the culture vessel 300 and measures the illuminance in the area where the culture vessel 300 is located. The pressure sensor 108 is placed near the culture vessel 300 and measures the atmospheric pressure in the room where the culture vessel 300 is located.

[0024] The cell culture optimization condition acquisition system 200 acquires culture-related data, such as physical and environmental stimuli applied to cultured cells, during repeated cell cultures, and records the acquired culture-related data in association with the observed images each time the cells are observed. Furthermore, the cell culture optimization condition acquisition system 200 combines culture-related data with the evaluation of observed and captured images to perform machine learning and generate a machine learning model. This model extracts the optimal conditions for culture-related data and uses this information to inform the protocol for mechanized cell culture (details will be provided later).

[0025] Figure 2 is a block diagram showing an example configuration of the cell culture optimization condition acquisition system 200 in this embodiment. The cell culture optimization condition acquisition system 200 comprises a sensor data acquisition unit 201, an operation data acquisition unit 202, an image acquisition unit 203, an activity data acquisition unit 204, a data management unit 205, a factor parameter extraction unit 206, an optimal data extraction unit 207, an image learning unit 208, a data management storage unit 209, and an optimal condition database 210.

[0026] The sensor data acquisition unit 201 acquires temperature, humidity, frequency / amplitude (hereinafter referred to as vibration data), CO2 concentration, and illuminance from the temperature sensor 103, humidity sensor 104, acceleration sensor 105, CO2 concentration sensor 106, and illuminance sensor 107 as culture-related data at predetermined intervals (for example, 1 second), and outputs them to the data management unit 205. The frequency / amplitude is a physical stimulus, while temperature, humidity, CO2 concentration, and illuminance are environmental stimuli.

[0027] The operation data acquisition unit 202 acquires drive information (movement amount, movement speed, movement time, etc.) of the robot hand 101 (movement amount, movement speed, movement time, etc.) when the robot hand 101 grasps the culture vessel 300 and moves the culture vessel 300, and outputs it to the data management unit 205. The image acquisition unit 203 captures images of the cultured cell population in the culture vessel 300 at each observation cycle of the cultured cells (for example, every hour), and outputs the captured observation images to the data management unit 205.

[0028] The activity data acquisition unit 204 collects some cells from the cultured cell group during the observation cycle of the cultured cells and outputs the activity data of some of the cultured cells acquired from the culture vessel 300, which has been measured by an external device, to the data management unit 205. Here, the activity data is information indicating the cellular activity of cultured cells, and consists of cell examination data for pluripotency (diversity), proliferation, and extracellular matrix production. Pluripotency is the ratio of the measured expression level to the reference expression level, obtained from gene expression analysis by PCR (polymerase chain reaction) testing of a specific gene or series of genes, which are expressed when the cell maintains an undifferentiated state. Proliferation is the rate at which cells proliferate and increase. For example, by first obtaining a graph of the relationship between cell number and absorbance, measuring the absorbance in the culture medium at each unit of time, and reading the number of cells corresponding to the measured absorbance from the relationship graph, the increase in the number of cells per unit time can be determined as proliferation. Extracellular matrix production is the amount of extracellular matrix produced, which is a secreted substance that constitutes the microenvironment, determined from the analysis of culture medium components, and is expressed in μg. The extracellular matrix is ​​the amount of extracellular matrix produced by cultured cells that supports the physical structure of the cell and also supports its biological function, and is expressed in μg.

[0029] The data management unit 205 adds a timestamp (for example, date and time information) for each observation cycle of cultured cells, and writes the activity data and observation image data to the cell observation table in the data management storage unit 209 for storage. Figure 3 shows an example of the configuration of the cell observation table in the data management storage unit 209. In Figure 3, the cell observation table has columns for timestamp, image index, pluripotency, proliferative capacity, extracellular matrix production, and judgment result for each record.

[0030] The timestamp indicates the date and time the observation image was captured, i.e., the time the cell observation was performed during the observation cycle. The image index is information that indicates the area where the data of the observed image is written, for example, it is an address that indicates a memory area. Pluripotency is a numerical value that indicates the diversity of cultured cells. Proliferative capacity is a numerical value that indicates the growth rate of cultured cells during the observation cycle. Extracellular matrix production refers to substances produced and secreted outside the cell by cells, such as glycoproteins like proteoglycans that indicate the state of cell motility.

[0031] The judgment result indicates whether the numerical values ​​of each judgment parameter, pluripotency, proliferation, and extracellular matrix production, fall within a predetermined range. In this embodiment, for example, pluripotency, proliferation, and extracellular matrix production are each evaluated on a scale of level #1 (appropriate), level #2 (intermediate), and level #3 (inappropriate). The optimal culture is determined when all of pluripotency (data A), proliferation (data B), and extracellular matrix production (data B) are at level #1. Levels #1, #2, and #3 are each set as predetermined numerical ranges. In the judgment result column, the data is written and stored as a data set (data A, data B, data C).

[0032] Furthermore, the data management unit 205 writes and stores each of the culture-related data, such as temperature, humidity, atmospheric pressure, illuminance, CO2 concentration, and vibration data (vibration frequency and amplitude), in the culture-related data table in the data management storage unit 209, along with a timestamp (for example, date and time information (information of the day and time)) at predetermined measurement cycles (similar to or shorter than the observation cycle). Figure 4 shows an example of the configuration of the culture-related data table in the data management storage unit 209. In Figure 4, the culture-related data table has columns for timestamp, temperature, humidity, atmospheric pressure, illuminance, and CO2 concentration for each record.

[0033] The timestamp is the date and time on which the temperature, humidity, vibration data, illuminance, and atmospheric pressure data were acquired by each of the temperature sensor 103, humidity sensor 104, acceleration sensor 105, CO2 concentration sensor 106, illuminance sensor 107, and atmospheric pressure sensor 108 during a predetermined measurement cycle, i.e., the time when the measurement was performed during the measurement cycle. Temperature is the ambient temperature surrounding the culture vessel 300, and humidity is the humidity of the ambient environment surrounding the culture vessel 300. Vibration data indicates the vibration state of the vibration table 102S on which the culture vessel 300 is mounted. Illuminance is the illuminance of the light shining on the area where the culture vessel 300 is located. Atmospheric pressure is the atmospheric pressure in the room where the culture vessel 300 is located.

[0034] Furthermore, the data management unit 205 writes and stores the process performed by operating the robot hand, the name of the process, and the processing time (the time the robot hand held the culture vessel) in the robot hand operation data table in the data management storage unit 209. Figure 5 shows an example of the configuration of the robot hand operation data table in the data management storage unit 209. The cell observation table has columns for timestamp, processing name, and processing time for each record.

[0035] The timestamp is the date and time when the robot hand 101 was operated, that is, the time when the culture vessel 300 was grasped by the robot hand 101. The process names describe the names of processes performed on the culture container 300, such as seeding, culture medium replacement, and subculturing, which involve grasping the culture container 300 with the robot hand 101 and moving it. The processing time is the time it took for the robot hand 101 to grasp and move the culture vessel 300.

[0036] The factor parameter extraction unit 206 forms a group of coordinate values ​​in the feature space, using the coordinate values ​​of each culture-related data (temperature, humidity, atmospheric pressure, illuminance, CO2 concentration, vibration frequency, vibration amplitude value) corresponding to levels #1, #2, and #3 of the pluripotency determination result. The factor parameter extraction unit 206 then changes the types of culture-related data to be combined and extracts combinations of culture-related data, such as temperature, illuminance, and CO2 concentration, that result in the Pluripotency determination being the furthest distance between the centroids of each coordinate group for level #1 and level #3.

[0037] Furthermore, in this embodiment, the degree of influence was determined by the distance between the centroids of the coordinate groups between the groups of good and bad cell culture results, which were determined by pluripotency, proliferative capacity, and extracellular matrix production. However, it is also possible to calculate and use the contribution to the change in the numerical value of the judgment result by dividing the difference in the numerical value of the culture-related data between the above-mentioned good and bad groups by the difference in the change in the numerical value of the judgment result.

[0038] As a result, the factor parameter extraction unit 206 extracts combinations of culture-related data that have an influence on the judgment result of the quality determination of pluripotency in cell culture (i.e., have a high intensity compared to other culture-related data) as factor parameters. Furthermore, the factor parameter extraction unit 206 performs the same processing as the extraction of combinations of culture-related data that have a high influence on the Pluripotency determination result described above, and extracts each combination of culture-related data that has an influence on the determination results of proliferative capacity and extracellular matrix production as factor parameters.

[0039] The optimal data extraction unit 207 clusters all types of culture-related data for each combination of factor parameters extracted for pluripotency, proliferative capacity, and extracellular matrix production. The optimal data extraction unit 207 then extracts clusters from the clusters formed in the feature space in which all determination results for pluripotency, proliferative capacity, and extracellular matrix production amount are level #1.

[0040] In this case, if multiple clusters are extracted, the optimal data extraction unit 207 extracts the cluster with the highest combination of values ​​for pluripotency, proliferative capacity, and extracellular matrix production. The optimal data extraction unit 207 then obtains numerical values ​​for combinations of culture-related data from cells that have high pluripotency, proliferative capacity, and extracellular matrix production, i.e., cells with higher activity than other groups. This allows for the acquisition of optimal values ​​for factor parameters as optimization conditions for cell culture, corresponding to culture-related data such as physical and environmental stimuli in repeatedly performed cell cultures. By using these values ​​as setting values ​​for environmental data when performing cell culture operations, it becomes possible to culture cells with higher cell activity compared to conventional methods.

[0041] The image acquisition learning unit 208 uses the observed image and the corresponding judgment result (judgment result shown in Figure 3) as training data to generate a machine learning model that estimates the judgment result from the observed image, for example, by deep learning. In other words, the image acquisition learning unit 208 generates a machine learning model MA that outputs the numerical value "1 (good)" as an estimated result when an observed image is input in which the combination of pluripotency, proliferative capacity, and extracellular matrix production is all at level #1, and outputs the numerical value "0 (bad)" as an estimated result when an observed image with any other judgment result is input. When performing cell culture operations, observation images are acquired at each observation cycle and input into the machine learning model MA mentioned above. This eliminates the need for operators to take some cells or a portion of the culture medium from the cell container to perform gene expression analysis or medium component analysis, thereby reducing the risk of contamination through collaboration between the operator and the machine.

[0042] Furthermore, the image acquisition learning unit 208 generates other machine learning models that estimate which judgment result is worse from the observed image, using combinations of judgment results other than the combination in which all of Pluripotency, proliferative capacity, and extracellular matrix production amount are at level #1. In other words, the image acquisition learning unit 208 uses observation images with combinations of judgment results other than the combination where Pluripotency, proliferative capacity, and extracellular matrix production amount are all at level #1 to generate a machine learning model MB_1 that outputs the number "0 (bad)" when an observation image with a Pluripotency judgment result of level #2 or #3 is input, and outputs the number "1 (good)" when an observation image with a Pluripotency judgment result of level #1 is input.

[0043] Similarly, the image acquisition learning unit 208 generates machine learning models MB_2 and MB_3, respectively, which estimate the quality of proliferation and extracellular matrix production, using observational images other than combinations where pluripotency, proliferative capacity, and extracellular matrix production are all at level #1. As a result, according to this embodiment, it is possible to generate a machine learning model MA that can easily determine whether or not cultured cells are active, that is, whether or not cell culture is being carried out normally, without performing gene expression analysis by PCR or culture medium component analysis.

[0044] Furthermore, according to this embodiment, machine learning models MB_1, MB_2, and MB_3 can be generated to easily determine which of the cell activity evaluations—pluripotency, proliferative capacity, or extracellular matrix production—is low. By combining this with the degree of influence of culture-related data on each cell activity, it becomes possible to easily estimate the factors that led to insufficient cell activity from observed and captured images.

[0045] Figure 6 is a conceptual diagram showing an example of the cycle for obtaining the culture determination results and observed images of cultured cells through cell observation. After seeding (subculturing), cell observation is performed at predetermined intervals (times), and observation images are captured using the imaging device 109. At the same time, gene expression analysis and culture medium component analysis are performed to determine cell activity.

[0046] Here, in cell observation during observation cycle T1, the evaluation results of cell culture from seeding ( / subculturing) to cell observation at the end of observation cycle T1 are shown to be influenced by the conditions under which seeding ( / subculturing) was performed and the culture-related data within the observation cycle T1. Furthermore, when performing a culture medium change (at the end of observation cycle T4), the culture evaluation results and observed images of the cultured cells immediately before the change are acquired to obtain information on the influence of each culture-related data on cell culture within the observation cycle T4.

[0047] In this embodiment, for each observation cycle described above, the results of determining pluripotency, proliferative capacity, and extracellular matrix production in cell culture are accumulated, and machine learning models MA, MB_1, MB_2, and MB_3 are generated. As a result, according to this embodiment, by performing cell culture using the optimal conditions (optimal values ​​of culture-related data) extracted for each observation cycle as described above, it becomes possible to obtain cultured cells with a predetermined cell activity. Furthermore, when performing cell culture operations, by capturing observation images of cells at the end of the observation cycle and inputting these images into a machine learning model MA, it becomes possible to easily estimate from the output of the machine learning model MA whether or not the cells cultured during the final observation period possess a predetermined level of activity.

[0048] Furthermore, according to this embodiment, by inputting observation images that have been determined by the machine learning model MA to lack a predetermined activity into the respective machine learning models MB_1, MB_2, and MB_3, it is possible to estimate which of pluripotency, proliferation, or extracellular matrix production has not reached level #1. This makes it possible to easily extract culture-related data that has a high impact on pluripotency, proliferation, and extracellular matrix production, and to support the detection of factors that hinder normal culture.

[0049] Figure 7 is a conceptual diagram illustrating an example of cell observation, showing the acquisition of culture-related data, cell activity determination results, and observed images during cell detachment. When subculturing cells, an enzyme such as trypsin, which degrades the extracellular matrix as a substrate, is used to detach the cultured cells from the culture vessel in order to transfer them to another culture vessel (a similar process is performed when retrieving cultured cells). Here, the trypsin solution contains EDTA (ethylenediaminetetraacetic acid) as a chelating agent to fix calcium and magnesium ions. Furthermore, during the exfoliation process described above, although not shown in Figure 4 as culture-related data, the processing time and trypsin concentration required for trypsin treatment are included.

[0050] Cell activity is determined at the end of an observation cycle performed at the start of subculturing, between groups treated with trypsin for a predetermined first hour and groups treated for a shorter second hour (e.g., half the first hour). In this case, if the effect of the change in trypsin treatment time on the change in cell activity is greater than that of other culture-related data, it is determined that the trypsin treatment time has an influence on cell activity. Then, at the end of the observation cycle, which began with the subculturing, an image of the subculturized and cultured cells is captured.

[0051] Furthermore, if the influence of the trypsin treatment time is higher than that of other culture-related data, the observation images acquired at the end of the observation cycle, which was performed at the start of the subculturing, will have characteristics that reflect the influence of the trypsin treatment time more than that of the other culture-related data. Therefore, by generating a machine learning model MB (Deep learner) using observational images in which the results of pluripotency, proliferative capacity, and extracellular matrix production are not all at level #1, it is possible to estimate from the observational images that the reason for the unsuccessful cell culture was the processing time of trypsin treatment, without performing gene expression processing or culture medium component analysis.

[0052] Alternatively, a program for realizing the functions of the cell culture optimization condition acquisition system 200 shown in Figure 2 may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be loaded into a computer system and executed to acquire cell culture optimization conditions and generate a machine learning model for estimating the state of cell culture. The term "computer system" here includes hardware such as the operating system and peripheral devices. Furthermore, "computer system" shall also include the homepage provisioning environment (or display environment) if a WWW system is being used. Furthermore, "computer-readable recording media" refers to portable media such as flexible disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into computer systems. Moreover, "computer-readable recording media" also includes those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs over networks such as the Internet or communication lines such as telephone lines, and those that hold programs for a certain period of time, such as volatile memory inside computer systems that act as servers or clients in such cases. In addition, the above-mentioned programs may be for the purpose of realizing some of the functions described above, and may also be able to realize the above-mentioned functions in combination with programs already recorded in the computer system.

[0053] While embodiments of this invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and includes designs and the like that do not depart from the spirit of this invention. [Explanation of Symbols]

[0054] 101…Robot hand 102…Mounting stand 102S…Vibration table 103…Temperature sensor 104…Humidity sensor 105…Accelerometer 106…CO2 concentration sensor 107…Illuminance sensor 108…Barometric pressure sensor 111…Fixing stand 111S…Support 200…Cell culture optimization condition acquisition system 201…Sensor data acquisition unit 202…Operation data acquisition unit 203…Image acquisition unit 204…Activation data acquisition unit 205…Data management unit 206…Factor parameter extraction unit 207…Optimal data extraction unit 208…Image learning unit 209…Data management storage unit 210…Optimal condition database 300…Culture vessel

Claims

1. a cell culture optimization condition acquisition system that acquires, in each of predetermined observation periods for cell culture, conditions for optimizing the environment for cell culture from culture-related data that is environmental data on an environment in which the cell culture is performed, a related data acquisition unit that acquires the culture-related data; an activity data acquisition unit that detects the activity of cells and determines the activity of the cultured cells at predetermined intervals; a factor parameter extraction unit that extracts, as a factor parameter, culture-related data that has a higher influence as a factor of a change in the cell activity level than other culture-related data; an optimum data extraction unit that extracts data of factor parameters when the cell activity is equal to or greater than a predetermined threshold as optimum data; A cell culture optimization condition acquisition system comprising:

2. an image acquisition unit that acquires an observation image of the cells in the cell culture in each of the observation periods; an image learning unit that uses the observed / captured image to generate a machine learning model that estimates the cell activity level from the observed / captured image; The cell culture optimization condition acquisition system according to claim 1, further comprising:

3. The captured image learning unit generates a machine learning model MA that estimates whether all of the cell activity levels are equal to or greater than a predetermined threshold, and a machine learning model MB that estimates whether any of the cell activity levels are equal to or greater than a predetermined threshold. The cell culture optimization condition acquisition system according to claim 2 .

4. The factor parameter extraction unit extracts, as the factor parameter combination, the combination of the culture-related data that maximizes the distance to the center of gravity of the coordinate group of the numerical values ​​in each of the groups with different cell activity levels. The cell culture optimization condition acquisition system according to any one of claims 1 to 3.

5. The factor parameter extraction unit calculates, as a contribution, a degree of difference in culture-related data in the group with respect to an activity difference, which is a difference in the cell activity between each of the groups with different cell activity levels, and extracts, as the factor parameter, culture-related data having a contribution greater than the contribution of other culture-related data. The cell culture optimization condition acquisition system according to any one of claims 1 to 3.

6. A method for acquiring optimized cell culture conditions, comprising the steps of: acquiring, in each of predetermined observation periods for cell culture, conditions for optimizing the environment for cell culture from culture-related data, which is environmental data of an environment in which the cell culture is performed; a related data acquisition step in which a related data acquisition unit acquires the culture-related data; an activity data acquisition step in which the activity data acquisition unit detects the activity of the cultured cells at predetermined intervals; a factor parameter extraction step in which a factor parameter extraction unit extracts, as a factor parameter, culture-related data that has a higher influence as a factor of a change in the cell activity level than other culture-related data; an optimal data extraction step in which an optimal data extraction unit extracts data of factor parameters when the cell activity is equal to or greater than a predetermined threshold as optimal data; A method for obtaining optimized cell culture conditions, comprising:

7. a program for causing a computer to operate as a cell culture optimization condition acquisition system that acquires, in each of predetermined observation periods for cell culture, conditions for optimizing the environment for cell culture from culture-related data that is environmental data on the environment in which the cell culture is being performed, The computer related data acquisition means for acquiring the culture-related data; an activity data acquisition means for detecting the activity of the cultured cells at predetermined intervals; a factor parameter extraction means for extracting, as a factor parameter, culture-related data that has a higher influence as a factor of a change in the cell activity level than other culture-related data; Optimal data extraction unit means for extracting data of factor parameters when the cell activity is equal to or greater than a predetermined threshold as optimal data A program to function as a

Images