Cell culture experiment system, cell culture experiment method, and program

Abstract

The present invention shortens a period until an appropriate environmental parameter is obtained in a cell culture experiment.　According to the present invention, a control device sets a recipe for each cell sample, and controls a robot device to perform a culture experiment according to the recipe. A database device sequentially stores intermediate progress data for each sample during one culture experiment in which a culture result is determined. A recipe determination device analyzes tendencies of the culture on the basis of the intermediate progress data for each sample, and on the basis of the analysis result and difference as to time-series intermediate progress data of the sample, calculates posterior probability indicating a degree of probability leading to future success of the culture, with respect to values of a plurality of environmental parameters making up the recipe being set. The recipe determination device uses one or more environmental parameters having a relatively high posterior probability as a verification parameter, and modifies a value of the verification parameter for each sample to correct the recipe. The control device controls the robot device such that the experiment is continued according to the corrected recipe.

Description

Cell culture experiment system, cell culture experiment method, and program

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

[0002] In the life science industry, modalities such as regenerative medicine and cell manufacturing are still in their infancy. Before researchers can obtain the normal cells they desire, there are various disturbing factors, and it is necessary to conduct a huge number of repeated experiments at once to find the appropriate environmental parameters. The set of environmental parameters for cell manufacturing or cultivation is also called a recipe.

[0003] Bayesian experimental design is known as a common method used in experiments for producing or culturing such cells. For example, Patent Document 1 describes a method for predicting, in a cell culture experiment, one or more execution parameter items and their search ranges that will affect the evaluation performance results based on one or more past execution performance results and evaluation performance results, based on a given feature extraction rule, and generating multiple variable parameter values ​​included in the search range based on one or more past execution performance results and evaluation performance results, in accordance with Bayesian optimization or the like.

[0004] Japanese Patent Application Laid-Open No. 2023-035238

[0005] Bayesian experimental design can improve the efficiency of experiments by utilizing experimental results to plan the next experiment, thereby reducing the number of experiments to be conducted, i.e., the amount of data required for an experiment. However, Bayesian experimental design cannot plan the next experiment until experimental results are obtained, and it cannot shorten the time required for an experiment itself. For example, it takes a long time, such as about four months, from the start of cell culture to the determination of the results as normal cells, and this period cannot be shortened.

[0006] For this reason, it takes four months to obtain accurate results and learn the appropriate intermediate environmental parameters, and it takes another four months in principle for the next results to be obtained using the learned intermediate environmental parameters, which significantly reduces research efficiency. Furthermore, even when the search period is effectively shortened by conducting a huge number of experiments at once, each experiment must be started in order according to the priority of the experiment.

[0007] An object of the present invention is to shorten the time required to obtain appropriate environmental parameters in cell culture experiments.

[0008] One representative embodiment of the present invention comprises a recipe determination device that sets a recipe to be applied to each of a plurality of cell samples to be cultured, a robot device, a control device that controls the operation of the robot device so that the robot device performs a culture experiment on the plurality of samples according to the set recipe, and a database device that stores interim data for each of the plurality of samples in chronological order during one culture experiment in which the culture results of the plurality of samples are determined, and the recipe determination device analyzes the tendency of the culture of each of the samples based on the interim data for each of the samples stored in the database device, and determines, for each of the samples, the tendency of the culture of the sample. This cell culture experimental system calculates a posterior probability representing the degree of likelihood that the parameter values ​​of multiple environmental parameters constituting the set recipe will cause future successful cultivation of the sample based on the trend analysis results and the difference in the time-series intermediate data for the sample, determines one or more environmental parameters with a relatively high calculated posterior probability as verification parameters, modifies the recipe for each sample by changing the parameter values ​​of the determined verification parameters, and the control device controls the operation of the robotic device so that the robotic device continues to perform the multiple culture experiments in accordance with the modified recipe.

[0009] a database device that stores interim data for each sample in chronological order during one culture experiment in which culture results of the multiple samples are determined; the recipe determination device analyzes trends in the culture of each sample based on the interim data for each sample stored in the database device; and, for each sample, calculates a posterior probability representing the degree of likelihood that parameter values ​​of multiple environmental parameters constituting the set recipe will cause future success in culturing the sample based on a difference between the analysis result of the culture trend for the sample and the chronological interim data for the sample; determines one or more environmental parameters with a relatively high calculated posterior probability as verification parameters; modifies the recipe for each sample by changing the parameter value of the determined verification parameter; and controls the operation of the robotic device so that the robotic device continues to perform the multiple culture experiments in accordance with the modified recipe.

[0010] One representative embodiment of the present invention is a program for causing one or more computers or processors to function as a recipe determination device that sets a recipe to be applied to each of a plurality of cell samples to be cultured, a control device that controls the operation of the robot device so that the robot device performs a culture experiment on the plurality of samples according to the set recipe, and a database device that stores interim data for each of the plurality of samples in chronological order during one culture experiment in which the culture results of the plurality of samples are determined, wherein the recipe determination device analyzes the trends in the culture of each of the samples based on the interim data for each of the samples stored in the database device, and for each sample, the program calculates a posterior probability representing the degree of possibility that the parameter values ​​of the multiple environmental parameters constituting the set recipe will cause future successful cultivation of the sample based on the analysis results of the trend in the cultivation of the sample and the difference in the chronological progress data of the sample, determines one or more environmental parameters with a relatively high calculated posterior probability as verification parameters, modifies the recipe for each sample by changing the parameter values ​​of the determined verification parameters, and the control device controls the operation of the robotic device so that the robotic device continues to perform the multiple cultivation experiments in accordance with the modified recipe.

[0011] According to a representative embodiment of the present invention, a technique is provided that can shorten the time required to obtain appropriate environmental parameters in a cell culture experiment.

[0012] 1 is a diagram schematically illustrating an example of the configuration of a cell culture experimental system according to embodiment 1. FIG. 2 is a diagram illustrating an example of the external appearance of main parts of the cell culture experimental system according to embodiment 1. FIG. 3 is a diagram illustrating an example of the configuration of data stored in a database device. FIG. 4 is a diagram (part 1) illustrating an example of a processing flow in the cell culture experimental system according to embodiment 1. FIG. 5 is a diagram (part 2) illustrating an example of a processing flow in the cell culture experimental system according to embodiment 1. FIG. 6 is a diagram (part 3) illustrating an example of a processing flow in the cell culture experimental system according to embodiment 1. FIG. 7 is a diagram illustrating an example of the operation of a robot device according to operation data. FIG. 8 is a diagram illustrating a detachment process of adherent cells in a cell culture process. FIG. 9 is a diagram illustrating how multiple devices operate using a common clock signal. FIG. 10 is a diagram illustrating an example of multiple video data that are part of the interim data. FIG. 11 is a diagram illustrating a force sensor that is an example of a sensor unit of a detection device. FIG. 12 is a diagram illustrating a specific example of contact of an object. FIG. 13 is a diagram illustrating an example of data on pressure applied to an object in each process that is part of the interim data. FIG. 14 is a diagram illustrating an example of a part of the interim data. FIG. 15 is a diagram illustrating an example of a bar graph showing the distribution of a likelihood function. FIG. 16 is a diagram illustrating an example of a table showing a likelihood function that represents the relationship between a prior probability distribution and likelihood. FIG. 10 is a diagram showing an example of a bar graph of posterior probabilities.

[0013] <Background of the Study by the Inventors> Before describing the embodiments of the present invention, the problems that arise when applying Bayesian experimental design to cell culture will be described using a specific example.

[0014] A typical Bayesian experimental design involves conducting an experiment using post-hoc results. For example, suppose there is newly developed cultured cell line X. To verify the effects of X, assume that the results are verified for 20 randomly selected samples out of a total of 40 samples assumed in advance. These 20 samples are called the validation group. The remaining 20 samples are then injected with dummy cells. Dummy cells are cells that are not expected to have any effect. These 20 samples are called the efficiency group. As a result, 4 samples in the validation group developed cancer (16 samples were successful), and 16 samples in the efficiency group developed cancer (4 samples were successful).

[0015] If it can be determined from these results that cultured cell X is effective, a rational decision can be made with 50% efficiency in man-hours, without having to verify all 40 samples. First, the posterior probability distribution is calculated using the following Bayes' theorem.

[0016]

[0017] Although probability is essentially a continuous value, for simplicity, we will assume that the value of p (the probability that, when one of the specimens develops cancer, that specimen belongs to the validation group) takes on discrete values ​​in increments of 0.1, such as 0.1, 0.2, 0.3, ..., 0.9 (i.e., we will use a discrete probability distribution).

[0018] Furthermore, in Bayesian estimation, the prior probability distribution uses a subjective probability (uniform distribution). Generally speaking, Bayesian statistics is a method of calculating the posterior probability distribution (posterior) from the prior probability distribution (prior) and likelihood (likelihood). ・The more data you have, the more accurate the posterior probability distribution becomes. ・When the data is large enough, the prior probability distribution becomes almost completely unproblematic.

[0019] In Bayesian statistics, it is known that once enough data is collected, the proportion of the prior probability distribution in the calculation will eventually become so insignificant that it can be ignored.

[0020] In the initial stage, we assume that we have no idea how effective cultured cell X will be. Therefore, we consider that nine models with p = 0.1 to p = 0.9 are equally likely, and the prior probabilities are uniformly set to 0.11 each (0.12 for p = 0.5 to make the sum 1).

[0021] f(0.1)=0.11 (*This means that the probability that p=0.1 is 0.11. The same applies below.) f(0.2)=0.11 f(0.3)=0.11 f(0.4)=0.11 f(0.5)=0.12 f(0.6)=0.11 f(0.7)=0.11 f(0.8)=0.11 f(0.9)=0.11

[0022] This probability distribution indicates that all prior probabilities are the same, that is, p = 0.1 to p = 0.9 are all equally likely, resulting in a uniform distribution (strictly speaking, to make the sum total 1, only p = 0.5 is set to 0.12).

[0023] Next, calculate the likelihood function. For example, the likelihood when the success rate is 0.5 is the same as the conditional probability that there will be 4 successes when 20 trials are performed, and can be calculated as follows:

[0024]

[0025] If the same calculation is performed for p = 0.1 to 0.9, the likelihoods can be calculated as follows:

[0026] p= 0.1 : 0.08977882814987155 p= 0.2 : 0.21819940194609996 p= 0.3 : 0.13042097437387037 p= 0.4 : 0.03499079040415815 p= 0.5 : 0.004620552062988271 p= 0.6 : 0.0002696861504765946 p= 0.7 : 5.007558331512445e-06 p= 0.8 : 1.3005697843199945e-08 p= 0.9 : 3.1788044999999754e-13

[0027] 16 is a diagram showing an example of a bar graph representing the distribution of likelihood functions. When the above-mentioned data are put together into a graph, the distribution of likelihood functions as shown in FIG.

[0028] The normalization constant is a constant used to adjust the sum of the probabilities of the posterior probability distribution to be calculated to 1. This is the sum of the products of the corresponding prior probability distributions and likelihood functions calculated so far.

[0029] 17 is a diagram showing an example of a table of likelihood functions that represent the relationship between the prior probability distribution and likelihood. First, by summarizing the likelihood functions that represent the relationship between the prior probability distribution and likelihood in a table, the table shown in FIG. 17 is obtained.

[0030] The normalization constant is calculated by multiplying these values ​​one by one for each column and adding them together: 0.11 x 0.0898 + 0.11 x 0.2182 + 0.11 x 0.1304 + 0.11 x 0.035 + 0.12 x 0.0046 + 0.11 x 0.0003 + 0.11 x 0.00 + 0.11 x 0.00 ≒ 0.053

[0031] From the above, we can use Bayes' theorem to find the posterior probability distribution. For example, to find the posterior probability when p = 0.5, we get the following:

[0032]

[0033] This is the posterior probability of the p = 0.5 model. This indicates that the probability that the cancer incidence rate is 0.5 when the number of cancer cases is 4 in 20 trials is only 1%.

[0034] Similarly, the remaining eight models from p=0.1 to p=0.9 are calculated as follows:

[0035] P(k=4| n=20, p= 0.1 ) : 0.1875450876140688 P(k=4| n=20, p= 0.2 ) : 0.45581154041134897 P(k=4| n=20, p= 0.3 ) : 0.27244522533562093 P(k=4| n=20, p= 0.4 ) : 0.07309463697920575 P(k=4| n=20, p= 0.5 ) : 0.010529656158194935 P(k=4| n=20, p= 0.6 ) : 0.0005633657039385877 P(k=4| n=20, p= 0.7 ) : 1.0460628472988208e-05 P(k=4| n=20, p= 0.8 ) : 2.716848495074202e-08 P(k=4| n=20, p= 0.9 ) : 6.640420472689636e-13

[0036] 18 is a diagram showing an example of a bar graph of the posterior probabilities. When the above-mentioned posterior probabilities are plotted on a bar graph, the graph shown in FIG.

[0037] Looking at the graph of the posterior probability distribution, we can see that there is an almost 100% probability that p<0.5 (probability of cancer development is 50% or less) (in fact, from the table in Figure 18, this can be calculated as 0.188 + 0.456 + 0.272 + 0.0731 = 0.989). In other words, there is a 98.9% chance that cultured cell X will be effective.

[0038] Among these probabilities, the probability that p = 0.2 is the highest at 0.456. This shows that the probability that the cancerous cells are from the validation group (0.2) is 1 / 4 of the probability that the cells are from the efficiency group (0.8) (i.e., the application of cultured cell X is most likely to reduce the cancer rate to 1 / 4).

[0039] The above is an example of a commonly used Bayesian experimental design, and in this case, it is possible to derive reasonable results even if the number of samples is reduced to 50% (i.e., half the labor required).

[0040] However, a problem with the culture environment to which the present invention is applied is that it takes a long time (e.g., four months) from the start of culture until the results of normal cells (or cancerous cells) are confirmed. As a result, it takes four months to obtain accurate results and derive an appropriate recipe, i.e., an intermediate environmental parameter set, through learning. In principle, it takes another four months before the next results can be obtained using the intermediate environmental parameter set derived through learning. This significantly reduces research efficiency.

[0041] Furthermore, because Bayesian experimental design requires setting all probability distributions at once, it is necessary to conduct a huge number of experiments at once and obtain a huge number of results. However, in reality, even if the search period is effectively shortened by conducting a huge number of experiments at once, each experiment must be started sequentially according to the priority of the experiment, which means that the results must also be compiled in order, which is an issue.

[0042] The present inventors have made the present invention as a result of extensive research in light of the above circumstances. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

[0043] In the drawings, the same elements are generally designated by the same reference numerals, and repeated explanations are omitted. When there are multiple identical or similar components, they may be described using the same reference numerals with different subscripts. In the drawings, the representation of each component may not represent its actual position, size, shape, scope, etc., in order to facilitate understanding of the invention, and the present invention is not necessarily limited to the position, size, shape, scope, etc., disclosed in the drawings. Identification information, identifiers, IDs, names, numbers, and other expressions for various data and information are interchangeable.

[0044] The functional blocks, circuits, computers, processing units, etc. that make up a device or system may be realized by electronic circuits combining electronic components, or they may be realized by an information processing device such as a processor or a computer including a processor executing a predetermined program.

[0045] A processor is composed of semiconductor devices such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an MCU (Micro Controller Unit). A processor is composed of devices or circuits capable of performing predetermined calculations. The processing is not limited to software program processing, but can also be performed by dedicated circuits. Examples of the dedicated circuits that can be used include FPGAs (Field Programmable Gate Arrays) and ASICs (Application Specific Integrated Circuits).

[0046] The program may be pre-installed as data on the target computer, or may be distributed as data from a program source to the target computer and then installed. The program source may be a program distribution server on a communication network, or a non-transitory computer-readable storage medium. The program may be composed of multiple program modules. The computer system may be composed of multiple devices.

[0047] (Embodiment 1) <Configuration of Cell Culture Experimental System> Fig. 1 is a diagram showing an example of the configuration of a cell culture experimental system according to embodiment 1. As shown in Fig. 1, the cell culture experimental system 10 according to embodiment 1 includes equipment placed in an aseptic operation area SC1 where grade A is assumed, equipment placed in an aseptic operation area SC2 where grade B / C is assumed, equipment placed in a general control area SC3, and equipment placed in other area SC4.

[0048] The sterile operation area SC1 is equipped with a robot device 300, a detection device 400, and cell culture equipment 600. The sterile operation area SC2 is equipped with a control device 100 and a terminal device 201. The general management area SC3 is equipped with a terminal device 200. The other area SC4 is equipped with a recipe determination device 550 and a database device 500.

[0049] The equipment installed in the sterile operation area SC1 is connected to the equipment installed in the sterile operation area SC2 and the general control area SC3 via a local bus LB so that they can communicate with each other. Also, the equipment installed in the sterile operation area SC2 is connected to the equipment installed in the other area SC4 so that they can communicate with each other via a cloud network bus CB. Note that the connections between the equipment via the above buses are merely examples, and are not limited to the above connection methods.

[0050] The control device 100 has an operation control unit 101 and a detection control unit 102. The operation control unit 101 transmits control signals to the robotic device 300 so that the robotic device 300 performs operations related to a cell culture experiment. The detection control unit 102 transmits control signals to the detection device 400 so that the detection device 400 detects predetermined physical quantities or data, and acquires detection data from the detection device 400.

[0051] The terminal devices 200 and 201 have an input unit and an output unit. The terminal devices 200 and 201 accept input settings and operations from users, such as researchers, via the input unit, and transmit control signals or data corresponding to the input settings and operations to the control device 100 or the detection device 400. The terminal devices 200 and 201 also acquire output signals or data from the control device 100 or the detection device 400, and output them to the user via the output unit. Note that both the terminal devices 200 and 201 allow users to perform input settings and operations related to the operation control of the robot device 300. However, generally, the user performs input settings and operations related to the operation control of the robot device 300 using the terminal device 201. The terminal device 200 may be, for example, a personal computer, a tablet terminal, a personal digital assistant, or the like.

[0052] The robot device 300 has various mechanisms, motors or actuators connected to the various mechanisms, and operates these mechanisms based on received control signals to perform various steps related to the cell culture experiment. The robot device 300 also exchanges information with the cell culture equipment 600 and delivers samples of cells to be cultured.

[0053] The main steps in a cell culture experiment are, for example, as follows:

[0054] First, there is a process called "medium exchange." Cells use the nutrients they need from the culture medium and metabolize those nutrients. Therefore, the medium, which has a low nutrient content and a high metabolic product content, is discarded and replaced with new medium. This process is called "medium exchange."

[0055] The second step is called "cell passaging." This requires the collection of a portion of the cultured cells and transferring them to a new culture vessel. This process is also called subculture. Cells divide repeatedly during culture, increasing their number until they eventually fill up the culture vessel or the nutrients in the medium are depleted.

[0056] The third step is called "cell seeding." Based on the measured cell count / cell concentration, an appropriate amount of "culture medium" is added to dilute the cell suspension, and a predetermined amount of the cell suspension is then transferred to a new culture vessel. This process can also be described as the process of adjusting the cell concentration to the desired concentration and uniformly seeding the suspension in the culture vessel. Note that a "cell suspension" refers to cells (or microorganisms) being cultured suspended in a liquid medium.

[0057] Further, more specific work contents in various steps of a cell culture experiment, that is, operations performed by the robot device 300, include the following.

[0058] The first is the operation called "pipetting." This refers to the process of sucking up and discharging a liquid (cell suspension) with a pipette. In cell culture, repeated sucking and discharging of a cell suspension with a pipette can also serve to loosen the cells in the cell suspension.

[0059] The second is the operation called "spreading." This refers to the process of spreading the cell suspension evenly over the bottom of the flask. The third is the operation called "dispensing." This refers to the process of accurately transferring a specified amount of liquid, such as culture medium, drug solution, or cell suspension, onto a plate or chip using a pipette.

[0060] The detection device 400 detects information related to the operation of the robot device 300. The detection device 400 includes an imaging unit 401 and a sensor unit 402.

[0061] The imaging unit 401 is installed in the laboratory and captures images of operations performed by the robotic device 300, for example, actions corresponding to each step constituting a cell culture experiment by the robotic device 300. The imaging unit 401 is, for example, a digital video camera, and outputs video data obtained by capturing the images.

[0062] The sensor unit 402 has one or more force sensors. The one or more force sensors are provided in various mechanisms of the robot device 300, detect forces, pressures, distortions, twisting forces, etc. applied to the various mechanisms as the robot device 300 operates, and output data corresponding to the detected physical quantities. The sensor unit 402 also has one or more environmental sensors. The one or more environmental sensors are provided in the laboratory, detect the environmental temperature, humidity, atmospheric pressure, atmospheric composition, etc. in the laboratory, and output data corresponding to the detected physical quantities.

[0063] The detection control unit 102 controls the time or period for capturing images by the image capturing unit 401, and cooperates with the operation control unit 101, if necessary, to synchronize the timing of operations or processes.

[0064] The cell culture equipment 600 includes a culture unit 601 , a measurement unit 602 , and a chemical solution installation unit 603 .

[0065] The culture unit 601 includes an incubator and other components, and cultures cells at a constant temperature and humidity. The measurement unit 602 measures physical quantities that represent the culture state of all or part of the cultured cells. The chemical solution installation unit 603 stores culture media as nutrients for culturing cells, chemical solutions used for measurement, consumables for adding chemical solutions, instruments for aspirating and dispensing liquids such as culture solutions or chemical solutions, and their consumables.

[0066] The database device 500 has a data storage area such as a cloud system that stores various data, information, etc. sent from the terminal devices 200, 201, the detection device 400, the recipe judgment device 550, the cell culture equipment 600, etc., and passed through the control device 100. The data stored in the database device 500 includes, for example, video data obtained by the detection device 400, environmental data, judgment result data obtained by the recipe judgment device 550, history data of judgment recipes judged in the past, etc. Note that data acquired by each of the above devices during an experiment and stored in the database device 500 is referred to as interim data.

[0067] The recipe judgment device 550 sets a recipe to be applied to each of multiple cell samples to be cultured. To streamline experiments, the recipe judgment device 550 also stores the priorities of verification for multiple types of environmental parameters that make up the recipe. The top N environmental parameters in terms of priority are called a judgment recipe. The recipe judgment device 550 references intermediate data from the database device 500 to correct and update the judgment recipe according to a predetermined algorithm.

[0068] Cell culture experiments are performed through operations by the robot device 300. The robot device 300 performs experiments in cooperation with other devices or units. For example, the robot device 300 temporarily transfers the cultured cell workpiece to the culture unit 601 and retrieves it, or transfers it to the measurement unit 602 for measurement, then retrieves it and starts operation, or takes in chemical solutions from the chemical solution installation unit 603 and uses them on the cultured cells.

[0069] The operation control unit 101 instructs the robot device 300 on the priority of experiments, predetermined processes, etc. The recipe judgment device 550 determines in advance the operations to be performed by the robot device 300 through the operation control unit 101 based on recipes set for each sample. The recipe judgment device 550 selects verification parameters, which are environmental parameters to be verified. Alternatively, the recipe judgment device 550 assigns priorities to sampling and instructs the operation control unit 101 to perform experiments according to the priorities.

[0070] The recipe judgment device 550 corrects the judgment recipe based on a trend analysis of the state of cultured cells based on the interim data stored in the database device 500. The recipe judgment device 550 corrects past judgment recipes using result indicators stored in the database device 500. When the artificial intelligence in the recipe judgment device 550 finds a causal relationship between environmental parameters and predicted culture results that differs from conventional causal relationships in the interim data, the recipe judgment device 550 improves the weight judgment of the trend analysis of the sample cell culture based on the interim data, i.e., the criteria for determining the trend in the trend analysis.

[0071] Here, the main functions of each device constituting the cell culture experimental system 10 according to the first embodiment can be summarized as follows.

[0072] The control device 100 sets a recipe to be applied to each of a plurality of cell samples to be cultured. The control device 100 controls the operation of the robot device 300 so that the robot device 300 performs a culture experiment on a plurality of samples according to the set recipe. The database device 500 stores interim progress data for each sample in chronological order during one culture experiment in which the culture results of the plurality of samples are determined.

[0073] The recipe judgment device 550 analyzes the culture trend of each sample based on the interim data for each sample stored in the database device 500. For each sample, the recipe judgment device 550 calculates a posterior probability, which indicates the degree of likelihood that the parameter values ​​of multiple environmental parameters constituting the set recipe will cause future successful culture of the sample, based on the analysis result of the culture trend of the sample and the difference between the time-series interim data of the sample. The recipe judgment device 550 determines one or more environmental parameters with a relatively high calculated posterior probability as verification parameters. The recipe judgment device 550 modifies the recipe for each sample by changing the parameter values ​​of the determined verification parameters.

[0074] The control device 100 controls the operation of the robotic device 300 so that the robotic device 300 continues to perform multiple culture experiments according to the modified recipe.

[0075] FIG. 2 is a diagram showing an example of the external appearance of the main components of the cell culture experimental system according to the first embodiment. As shown in FIG. 2 , the system includes a control device 100, a terminal device 201, a robotic device 300, and a detection device 400. A table 620 is also provided, and multiple instruments 610 to be processed and operated by the robotic device 300 are placed on the table 620. The instruments 610 include test tubes 611, pipettes 612, containers 613, well plates 614, flasks 615, and the like. The test tubes 611, pipettes 612, and containers 613 included in the instruments 610 are each operated by the robotic device 300. Images of the robotic device 300 and the multiple instruments 610 are captured by the imaging unit 401. Environmental sensors included in the sensor unit 402 detect the ambient temperature, humidity, atmospheric pressure, and atmospheric composition in the laboratory. Pressure, twisting force, strain force, and the like generated in various mechanisms constituting the robotic device 300 are detected by force sensors included in the sensor unit 402.

[0076] The robot device 300 has a base placed on the floor. The robot device 300 also includes a rotatable joint connected to the base and an arm connected to the joint. A joint and an arm are connected to the end of the arm. A gripper for gripping the instrument 610 is connected to the end of the arm. The robot device 300 moves and manipulates the instrument 610 using the joint, arm, gripper, and the like. Images captured by the imaging unit 401 are displayed on the output unit of the terminal device 201 via the control device 100, for example. The robot device 300, table 620, instrument 610, and the like are placed in a sterile operation area SC1. Placing the instrument 610 in the sterile operation area SC1 prevents foreign matter from being mixed into samples and the like. Other devices, such as a device for transporting the instrument 610, may also be placed in the sterile operation area. In this example, the terminal device 201 is placed in a sterile operation area SC2 that is different from the sterile operation area SC1, and can be operated by a user. Note that the area in which the terminal device 201 is placed may be any other area that has a lower cleanliness grade than the sterile operation area SC1, and is not limited to this example.

[0077] The control device 100, the database device 500, and the recipe determination device 550 can be realized using a dedicated or general-purpose computer such as a personal computer (PC) or a workstation (WS), or an electronic device equipped with a computer. The terminal device 201 can be realized using a dedicated or general-purpose computer such as a PC, a smartphone, a mobile phone, a tablet terminal, or a personal digital assistant (PDA), or an electronic device equipped with a computer.

[0078] 3 is a diagram showing an example of the configuration of a computer. As shown in FIG. 3, the computer 90 includes a processor 91, a memory 92, a storage 93, an input unit 94, an output unit 95, a communication control unit 96, and a communication bus 97. The processor 91, the memory 92, the storage 93, the input unit 94, the output unit 95, and the communication control unit 96 are connected via the communication bus 97 so as to be able to transmit and receive signals to and from each other. A predetermined program PG is stored in the storage 93. The processor 91 reads the program PG, loads it into the memory 92, and executes it, so that the computer 90 functions as, for example, the control device 100, the database device 500, or the recipe determination device 550.

[0079] 4 is a diagram showing an example of the configuration of data stored in the database device 500. As shown in Fig. 4, the database device 500 stores, for each cell sample that is the subject of a culture experiment, a sample number (No.) assigned to the sample, progress data related to the sample, an experiment result indicator, and the like, in association with each other.

[0080] <Processing Flow in Cell Culture Experimental System> Here, an example of the processing flow in the cell culture experimental system 10 according to the first embodiment will be described.

[0081] 5 to 7 are diagrams showing an example of the flow of processing in the cell culture experimental system according to the first embodiment.

[0082] As shown in FIG. 5 , in step S101, the initial setting of the determination recipe is performed. Specifically, the user inputs information specifying the initial setting of the recipe to be set for a plurality of cell samples (hereinafter simply referred to as samples) from the terminal device 201. The plurality of samples are divided into a group called a "verification group" and a group called an "efficiency group." The terminal device 201 transmits the input information of the initial setting to the recipe determination device 550. Based on the received information of the initial setting, the recipe determination device 550 stores information of the recipe for each sample belonging to the verification group and information of the recipe for each sample belonging to the efficiency group in the storage unit of the recipe determination device 550.

[0083] Next, in step S102, likelihood is calculated. This "likelihood" corresponds to the "likelihood" used in Bayes' theorem. Specifically, in step S102, the recipe judgment device 550 accesses the database device 500 and determines whether or not there is interim data for multiple samples. If it is determined that there is no interim data, the recipe judgment device 550 uses artificial intelligence (AI) that has been trained based on the results of previous cell culture experiments to calculate the likelihood for the currently set recipe as an estimated value. On the other hand, if it is determined that there is interim data, the recipe judgment device 550 reads the interim data for multiple samples from the database device 500 and substitutes the read interim data into a predetermined calculation formula to calculate the likelihood for each sample.

[0084] Next, in step S103, a posterior probability is calculated. This "posterior probability" corresponds to the "posterior probability" used in Bayes' theorem, and here represents the probability that future culture will be successful if the sample is cultured according to the recipe. More precisely, the posterior probability is the success rate (e.g., score: 0% to 100%) of cell culture for each environmental parameter to be verified. Specifically, in step S103, the recipe determination device 550 uses the above-described artificial intelligence to calculate the posterior probability for each environmental parameter in the recipe for each sample, using Bayes' theorem, based on the calculated likelihood and the prior probability preset according to the recipe for each of the multiple samples.

[0085] Next, in step S104, the nth determination recipe is modified. As mentioned above, this "determination recipe" refers to the recipe to be determined that is set for each sample. The recipe is repeatedly rewritten and modified multiple times based on the progress data obtained during the culture experiment according to that recipe. Therefore, here, the recipe to be determined at the current time is called the determination recipe.

[0086] Specifically, in step S104, the recipe judgment device 550 uses the above-mentioned artificial intelligence to rewrite the judgment recipe for each sample in the verification group and the efficiency improvement group using Bayes' theorem based on the posterior probability calculated for that sample.

[0087] The evaluation recipe consists of a list of multiple environmental parameters to be verified and a priority for each environmental parameter. The priority is determined from the results of trend analysis of cell culture based on interim data. Trend analysis of interim data involves analyzing the trends in the interim progress of cell culture based on interim data, or identifying conditions such as acceleration, stagnation, and cancer of cell growth (proliferation). Cell culture often takes a long time. During the culture, growth may gradually stagnate, or at some point, cells may suddenly become cancerous, or in some cases, growth may begin to accelerate after stagnation. Trend analysis involves analyzing trends in such interim progress of cell culture.

[0088] Here, the recipe judgment device 550 analyzes the intermediate data using artificial intelligence, calculates the degree to which each environmental parameter is likely to affect the success or failure of cell culture, i.e., a priority, and assigns that priority to each environmental parameter.Then, the environmental parameters with the highest N priorities are set as verification parameters, which are parameters to be verified, and the verification parameters are assigned parameter values ​​and distributed among the recipes of multiple samples.In other words, the recipe judgment device 550 extracts verification parameters from the analysis results of the intermediate data, and assigns parameter values ​​to environmental parameters corresponding to the extracted verification parameters among the environmental parameters of the recipes of multiple samples.

[0089] In other words, rather than thoroughly verifying all patterns in the combination of values ​​of numerous environmental parameters, patterns that are determined to have high priority based on recipe judgment are extracted and the recipe for that pattern is verified. In other words, the contents of the culture process for each sample are modified. The above priority is a posteriori probability and is derived using artificial intelligence.

[0090] The artificial intelligence used here has been trained or continues to be trained to prioritize verification parameters so that samples set with a combination of environmental parameter types and parameter values ​​that are thought to be highly likely to contribute to successful cell culture will have a higher success rate, and conversely, samples set with a combination of environmental parameter types and parameter values ​​that are thought to be highly likely to contribute to cell culture failure will have a higher failure rate (lower success rate). In other words, the artificial intelligence is capable of making rational decisions to reduce the number of verification parameters, such as by prioritizing "only the top 20% of verification parameters need to be verified."

[0091] The rewriting, or modification, of the judgment recipe is performed so that the posterior probability becomes higher for samples belonging to the verification group, and so that the posterior probability becomes lower for samples belonging to the efficiency group. The "nth" in the nth judgment recipe starts from n=1 and increases with each rewriting, such as n=2, n=3, ...

[0092] Next, in step S105, it is determined whether or not there is a change in the cell state of the samples belonging to the verification group. That is, it is determined whether or not to reset the configuration of samples in the verification group and the efficiency improvement group. Specifically, the recipe determination device 550 determines whether or not there is a change in the verification group of a predetermined sample population so that the verification group becomes a group with a higher priority.

[0093] For example, the recipe determination device 550 determines whether or not there has been a change in the cell state of a sample belonging to the verification group based on information contained in the interim data, such as the number (density) of cells, the cell proliferation rate, and whether or not the cells have become cancerous. If it is determined that there has been a change in the cell state (S105: Yes), the processing step proceeds to step S106 to reset the configuration of the verification group and the efficiency improvement group. On the other hand, if it is determined that there has been no change in the cell state (S105: No), the processing step proceeds to step S107 to maintain the configuration of the samples in the verification group and the efficiency improvement group.

[0094] In step S106, the composition of samples in the verification group and the efficiency group is reset. Specifically, the recipe determination device 550 uses artificial intelligence to reset the composition of samples belonging to the verification group and the efficiency group based on changes in the state of the cells. More specifically, as described above, the artificial intelligence resets the composition of samples in the verification group and the efficiency group so that a group of samples set with a combination of environmental parameters and their parameter values ​​that are likely to contribute to successful cell culture will have a higher culture success rate, and conversely, a group of samples set with a combination of environmental parameters and their parameter values ​​that are likely to contribute to cell culture failure will have a lower culture success rate (higher failure rate). The processing then proceeds to step S108.

[0095] In step S107, the configuration of the samples in the verification group and the efficiency improvement group is maintained. Specifically, the recipe determination device 550 maintains the current configuration of the samples in the verification group and the efficiency improvement group without changing them. Thereafter, the processing proceeds to step S108.

[0096] In step S108, the experiment is started. Specifically, the operation control unit 101 starts a process of transmitting a control signal to the robot device 300 so that a culture experiment is started for each sample belonging to the verification group and the efficiency improvement group in descending order of the set priority and so that the culture experiment is carried out according to the recipe set for each sample. Furthermore, the operation control unit 101 transmits control signals to the cell culture equipment 600 as needed so that the cell culture equipment 600 performs preparations, condition settings, etc. required for the experiment. Furthermore, the detection control unit 102 transmits control signals to the detection device 400 as needed so that the detection device 400 performs imaging using the imaging unit 401 or data acquisition using the sensor unit 402.

[0097] In step S109, it is determined whether the final results of the culture experiment have been obtained. Specifically, the control device 100 accesses the database device 500 to determine whether information related to the final results of the culture experiment currently being conducted is stored, and based on this determination result, it determines whether the final results of the culture experiment currently being conducted have been obtained. If it is determined that the final results have been obtained (S109: Yes), the processing step proceeds to step S114 so that the culture results can be reviewed. On the other hand, if it is determined that the final results have not been obtained (S109: No), the processing step proceeds to step S110 so that interim data can be stored in the database device 500.

[0098] In step S110, progress data during a predetermined operation process period of the robot device 300 is acquired. Specifically, the detection device 400 acquires video data recording the operation of the robot device 300 for the operation process currently being performed among a plurality of predetermined operation processes of the robot device 300 (S110-1), and acquires environmental data such as the temperature, humidity, and atmospheric pressure in the laboratory (S110-k). The detection device 400 also acquires operation data, which is control data that defines the operation of the robot device 300 in each operation process, from the recipe determination device 550 or the control device 100.

[0099] In step S111, it is determined whether the acquired intermediate data has characteristics in light of a criterion. Specifically, the recipe determination device 550 stores criterion information for each type of intermediate data, and determines whether characteristics appear for each acquired intermediate data by comparing it with the criterion. Note that this determination may be based on a threshold value of an index value calculated based on the intermediate data, or may be based on the analysis results of the intermediate data by artificial intelligence.

[0100] For example, the recipe judgment device 550 uses artificial intelligence or a user's evaluation or input to determine whether or not video data recording the operation of the robot device 300 contains characteristic operations that differ from expected operations (S111-1). Also, for example, the recipe judgment device 550 uses artificial intelligence or a user's evaluation or input to determine whether or not environmental data contains characteristics such as temperature, humidity, air pressure, etc. that are different from expected reference values ​​by a predetermined level or more (S111-k). Also, for example, the recipe judgment device 550 uses artificial intelligence or a user's evaluation or input to determine whether or not operation data of the robot device 300 contains characteristic control.

[0101] The process proceeds to step S112 for processing the environmental parameters determined to be characteristic in step S111. On the other hand, the process proceeds to step S113 for processing the environmental parameters determined to be non-characteristic in step S111.

[0102] In step S112, markings indicating characteristics are applied to the intermediate data. Specifically, the recipe judgment device 550 applies markings indicating portions of the intermediate data that have been determined to have characteristics. For example, the recipe judgment device 550 applies markings to the moving image data that correspond to the start of a time period during which an operation determined to have characteristics is occurring in the moving image data (S112-1). Also, for example, the recipe judgment device 550 applies markings to the environmental data that correspond to the start of a time period during which a change determined to have characteristics occurs in the environmental data (S112-k). Once the application of the markings is complete, the processing proceeds to step S113.

[0103] In step S113, interim data for each sample is stored. Specifically, the control device 100 stores interim data for each sample in the culture experiment in a storage area of ​​the database device 500. A folder is prepared for each sample number in the database device 500, and the interim data for each sample is stored in the folder for that sample number. In other words, interim data corresponding to each operation step of the robot device 300 in the culture experiment for each sample is accumulated in the folder prepared for each sample. Once the storage of the interim data is complete, the processing returns to step S102.

[0104] In step S114, the final culture results are checked. Specifically, the control device 100 outputs the final culture results of each sample in the validation group and the efficiency improvement group to the user, for example by displaying them on the screen of the display unit of the terminal device 201. A user, such as a researcher, refers to the culture results and inputs success indicators for the culture experiment for each sample. Once the input is complete, the processing proceeds to step S115.

[0105] In step S115, it is determined whether the success index is within a preset range. Specifically, the recipe determination device 550 determines, for each sample, whether the value of the success index is within a preset range of m to n.

[0106] In this judgment, for a sample whose success index is determined to be within the set range (S115: Yes), the recipe judgment device 550 sets the result that the success index of the sample is within the set range as a result index and stores the result index in the folder of the target sample in the database device 500 (S116).

[0107] On the other hand, for a sample for which the success index is determined not to be within the set range (S115: No), the recipe judgment device 550 determines that the success index of that sample is outside the set range as the result index. A user, such as a researcher, examines the cause of the result and inputs information describing the cause through the terminal device 201 (S117). For example, the user extracts differences that resulted in extremely good or bad results from the progress data, operates the terminal device 201, and adds the extracted results to the result index and stores them in the database device 500. For example, the recipe judgment device 550 uses artificial intelligence to investigate the causal relationship between the cause and the progress data for that sample, adds the investigation results to the result index, and stores the result index in the target sample folder in the database device 500 (S118).

[0108] When the processing for each sample in steps S115 to S118 is completed, the processing proceeds to step S119.

[0109] In step S119, it is determined whether or not to terminate the experiment. Specifically, the control device 100 determines whether or not to terminate the culture experiment based on information input by the user through the terminal device 201 or based on information representing a preset experimental plan. If it is determined that the culture experiment is to be terminated (S119: Yes), the processing related to the culture experiment is terminated. On the other hand, if it is determined that the culture experiment is not to be terminated, i.e., to be continued (S119: No), the processing step returns to step S102, and the processing related to the culture experiment is continued.

[0110] 5 to 7, the database device 500 stores intermediate data and evaluation results for each operation step of the robot device 300 and for each sample. Based on the stored intermediate data, the recipes of each sample in the verification group and the efficiency group, i.e., the parameter sets of intermediate environmental parameters, are sequentially modified. As a result, the verification group will contain samples with recipes that have a high probability of successful cultivation, while the efficiency group will contain samples with recipes that have a low probability of successful cultivation.

[0111] The recipes of the samples in the validation group converged in this way reveal the types of environmental parameters and their conditions that are likely to lead to successful cultivation, while the recipes of the samples in the efficiency group converged in this way reveal the types of environmental parameters and their conditions that are likely to lead to cultivation failure, or the types of environmental parameters that are likely to have no effect on the success or failure of cultivation.

[0112] The definition of successful cell culture varies depending on the user of the system, and there are no specific conditions. Successful cell culture can be defined, for example, as the cell health and cell count meeting pre-set conditions after the culture process is complete. It can also be defined as the cell health and cell count meeting pre-set conditions after the culture process is complete, and the freezing process being successful. The user of the system can define this definition of successful culture and intuitively add the success indicator (score) for each sample as accompanying data of the result indicator.

[0113] One of the most distinctive features of the first embodiment is that it stores, as interim data, video data, operation data, data detected by sensors, and the like related to the operation of the robotic device 300 during an experiment. Research to date has shown that the success or failure of cell culture is influenced by operations such as how the container containing the cells is handled, how the chemical solution or cell solution is aspirated or dispensed with a pipette, and what kind of impact is applied to the cells. Therefore, storing data related to the operation of the robotic device 300 related to these operations as interim data is extremely important in searching for an appropriate recipe for successful culture, i.e., the types of environmental parameters and their conditions.

[0114] That is, the intermediate data may be, for example, information related to the operation of the robot device 300 that has been corrected for each sample as intended by the user. The information related to the operation of the robot device 300 may be, for example, information such as the trajectory of the movement of the tool being handled, the speed of the movement of the tool, the acceleration of the movement of the tool, the posture of the tool, the pressure applied by the tool to the object, and the twisting force applied by the tool to the object. This information is different from uniform data that has been determined in advance by teaching. Specific examples of intermediate data will be described below.

[0115] <Specific Example 1 of Intermediate Progress Data> Figure 8 is a diagram showing an example of the operation of the robot device related to the operation data. As shown in Figure 8, the operation data stored as the intermediate progress data may, for example, be data relating to the operation of the robot device 300 when a liquid 651, such as a drug solution or cell fluid, in a petri dish 650 is aspirated and then discharged using a pipette 652. The operation data includes data representing operations corresponding to various conditions, such as a discharge start position P1, a trajectory RT of the pipette during discharge, a discharge end position P2, the natural separation of cell clumps by applying the pipette 652 to the wall of the petri dish 650, the gentle mixing of the liquid 651 as it spills from right to left (F8a in Figure 8), and the natural rotation of the fluid as the liquid 651 spills from left to right (F8b in Figure 8).

[0116] <Specific Example 2 of Progress Data> The control device 100 operates the robot device 300 based on the created operation data, i.e., the scenario. The detection device 400 monitors the operation of the robot device 300 and outputs the data to the user via the terminal device 201. While the robot device 300 is operating based on the scenario, the control device 100 accepts input of a correction to the operation from the user via the terminal device 201, and corrects the scenario.

[0117] Fig. 9 shows the detachment process of adherent cells in the cell culture process. In the detachment process of adherent cells shown in Fig. 9, most of the cells can be properly recovered by scraping with a scraping tool, but some cells are damaged and die due to the impact during scraping.

[0118] For example, suppose that detection device 400 detects a scraping action on a cell over a certain time period in video data. The speed of the scraping action, when performed by hand, is approximately 10 cm / sec. If the video of the scraping action is important as progress data, it would naturally be important to capture the event from when the scraping action begins to have a physical effect on the cell in the vicinity until it comes into contact with the cell and some kind of event occurs, that is, a distance interval roughly equal to or greater than the size of the cell, approximately 100 μm.

[0119] In the above example, a simple calculation shows that the time required for the robot device 300 to scrape off a section of about 100 μm is at most 1 ms. Therefore, time consistency is required between multiple devices to ensure that a series of operations are performed within a short period of 1 ms.

[0120] FIG. 10 illustrates how multiple devices operate using a common clock signal. Generally, the control device 100, robot device 300, detection device 400, and terminal devices 200 and 201 are controlled by a common clock signal output from a clock device 103 with microsecond-order accuracy, as shown in FIG. 10 . This ensures the mutual time consistency between devices, as required in the above example. This allows the operations of each device to be synchronized for events occurring within 1 millisecond, making it possible to secure interim data, such as "a series of operations from xx minutes xx seconds 500 milliseconds to xx minutes xx seconds 501 milliseconds." Note that the time error representing the accuracy of the common clock signal output from the clock device 103, such as the error in the time interval between rising edges, is preferably 10 μS (microseconds) or less.

[0121] 11 is a diagram showing an example of multiple video data that are part of the progress data. In the above example, for each sample, multiple video data are acquired as progress data, as shown in FIG. 11. The example of FIG. 11 shows how multiple video data are acquired as progress data H11 for sample No. 1. Each piece of video data that makes up the progress data H11 is data in which, for example, progress data number, start time, end time, and video number are associated with each other.

[0122] If the time of each device were not consistent, for example, it would be impossible to determine which time period the robot device 300 was operated in to cause a detected movement, and the data would be meaningless as interim data. In this way, it is possible to secure a series of interim data that is coordinated between devices using time consistency. Note that, depending on the application in the research field, verification of more microscopic movements may be required, and in this case, it goes without saying that more precise time consistency is required.

[0123] <Specific Example 3 of Progress Data> Specific example 3 of progress data is an example in which measurement data of pressure generated in a culture tool, acquired by a force sensor, is used as progress data.

[0124] Fig. 12 is a diagram showing a force sensor, which is an example of a sensor unit of a detection device. The force sensor 402a has a configuration that utilizes a lever, for example, as shown in Fig. 12, and is a sensor that detects a pressure PA1 that an operating tool 402b held by a part 300a of the robot device applies to an object it comes into contact with by detecting a pressure PA2 that is generated according to the principle of action and reaction. The force sensor 402a is configured by combining a plurality of sensors called strain sensors that convert pressure into an electrical signal, for example, and is capable of quantitatively grasping pressure.

[0125] FIG. 13 illustrates specific examples of contact between objects. The first example relates to a process for detaching adherent cells. More specifically, as shown in FIG. 13 , reference symbol F13a, the first example relates to a process for detaching adherent cells from the bottom surface 661 of a Petri dish by pressing a scraping tool 663 against the bottom surface 661 and moving the scraping tool to detach cells 662 cultured while adhering to the bottom surface 661 of the Petri dish. To avoid damaging the cells 662 and ensure sufficient detachment, it is important to operate the scraping tool while applying a predetermined pressure to the bottom surface 661 of the Petri dish. Therefore, the amount of pressure PA3 applied to the bottom surface 661 of the Petri dish, i.e., the measured value of this pressure PA3, can serve as interim data.

[0126] The second example relates to a sample extraction process. More specifically, the second example relates to a process in which a sample 665 is extracted from a test tube 664 by pressing an instrument 666 against the inner wall of the test tube 664 and pulling it out, as shown in FIG. 13, by reference symbol F13b. In this case, the amount of pressure PA4 applied to the inner wall surface of the test tube 664, i.e., the measured value of this pressure PA4, can be used as interim data.

[0127] The third example relates to a test tube cap opening and closing process. More specifically, the third example relates to a process in which gripping mechanisms 310, 311 of a robot device grip and open / close cap 668 in an operation of rotating cap 668 of test tube 667, as shown in FIG. 13 and indicated by reference symbol F13c. In this case, the amount of pressure PA5 applied to cap 668 when opening or closing cap 668 of test tube 667, i.e., the measured value of this pressure PA5, can be used as interim data.

[0128] FIG. 14 is a diagram showing an example of data on pressure applied to an object in each step, which is part of the progress data. A force sensor can detect multiple parameters, such as axial pressure or torque pressure. For each sample, data on detected pressure along the x-, y-, and z-axes in each step is acquired as progress data, as shown in FIG. 14 . The example in FIG. 14 shows multiple pieces of detected pressure data acquired as progress data H14 for sample No. 1. The individual pieces of detected pressure data constituting the progress data H14 are, for example, data in which the progress data No., detected pressure along the x-axis, detected pressure along the y-axis, detected pressure along the z-axis, and case (step) are associated with each other. Examples of possible cases (steps) include detaching adherent cells, removing a specimen, opening a test tube lid, and closing a test tube lid.

[0129] In this way, it is possible to secure, as intermediate data, the type of force sense used in the operation of the robot device 300 in each step of the culture experiment.

[0130] <Other Specific Examples of Intermediate Data> Potential intermediate data are, of course, not limited to those described above, and include, for example, the following: the angle at which the pipette is held, the speed at which the pipette's flange is pushed and pulled, the timing at which the flange is pushed and pulled, the acceleration during flange movement, the conditions for the operation of the tip attached to the end of the pipette from contacting the liquid to removing it, the speed at which the petri dish is moved, the tilt of the petri dish when it is moved, the speed at which the incubator door is opened and closed, etc. While these can be intermediate data, they can also be considered environmental parameters that make up the recipe to be verified.

[0131] 15 is a diagram showing an example of a portion of the interim data. As shown in FIG. 15, as part of the interim data, for example, an interim data number, a timing of modification of the culture recipe, a modified culture recipe, a likelihood, and a posterior probability are stored in association with each other.

[0132] <Example of Effects of First Embodiment> As described above, according to the first embodiment, in a cell culture experiment for which post-test results have not been obtained or for which it will take a long time for the post-test results to be determined, the verification priority of the environmental parameters is determined based on a first determination recipe by using artificial intelligence that has learned the relationship between the environmental parameter settings and the culture results based on the results of many past experiments. The first determination recipe has intermediate data saved as supplementary information, and the first determination recipe is corrected based on the intermediate data. The corrected first determination recipe is then applied to the next similar determination recipe. In this way, by utilizing recipes that are successively corrected and updated during the culture experiment, it becomes possible to estimate the success or failure of the culture, or the environmental parameters and their conditions related to the success of the culture, before the final results of the culture experiment are obtained.

[0133] Furthermore, according to the first embodiment, if a difference occurs in the cell culture state due to a change in the recipe, the difference in the interim data previously attached to that recipe is extracted, and the experimental result is predicted at the point when a difference appears in this interim data. This makes it possible to shorten the experimental period, eliminate useless experiments, and discover causal information collection.

[0134] Let's consider a scenario in which a robotic cell culture experiment system is used without modifying a predetermined scenario, and optimal values ​​are sought for some of the numerous environmental parameters that make up a cell culture recipe. The environmental parameters to be sought include, for example, the drug concentration and drug addition time in step 1, the drug treatment time in step 2, the cell detachment amount and cell detachment strength, and the drug concentration and drug addition time in step 3. The timing for changing the environmental parameters to be sought is after the experiment is completed, and the environmental parameters to be sought can be changed automatically. However, if there are many environmental parameters, it may take a long time, such as four months, to complete all the experiments.

[0135] It would be fine if the number of environmental parameters to be explored was small enough that humans could predetermine them, but the vast majority of them are not. Therefore, with this type of system, it is necessary to repeatedly perform incubation experiments by varying all environmental parameters, feed back the results, and then perform another incubation experiment, which takes an extremely long time. In other words, exploration is practically difficult.

[0136] On the other hand, the cell culture experimental system according to the first embodiment analyzes the progress data, calculates the degree (priority) that each environmental parameter is likely to have on the success or failure of cell culture, and assigns that degree to each parameter. The top N parameters are then set as verification parameters, and the values ​​of the verification parameters are distributed and set across multiple samples. In other words, the contents of the culture process for each sample are modified. The degree is a posteriori probability derived by artificial intelligence.

[0137] If an environmental parameter is found that indicates an extreme value for the posterior probability (singular point: outside the range of 20% to 80%), there is a possibility that equipment may have been lost or an accident may have occurred during the experiment. In such cases, a human being will intervene to identify the causal relationship and add the identified cause to the interim data.

[0138] (Embodiment 2) A cell culture experiment method as an embodiment of the present invention will be described below. The cell culture experiment method is as follows.

[0139] The recipe judgment device sets a recipe to be applied to each of multiple cell samples to be cultured. The control device controls the operation of the robot device so that the robot device performs culture experiments on the multiple samples according to the set recipe. The database device stores interim progress data for each sample in chronological order during one culture experiment in which the culture results of the multiple samples are determined. The recipe judgment device analyzes the culture trend for each sample based on the interim progress data for each sample stored in the database device, and calculates, for each sample, a posterior probability representing the degree of likelihood that parameter values ​​of multiple environmental parameters constituting the set recipe will cause future successful culture of the sample based on the difference between the analysis result of the sample culture trend and the chronological interim progress data for the sample. The recipe judgment device determines one or more environmental parameters with a relatively high calculated posterior probability as verification parameters, and modifies the recipe for each sample by changing the parameter values ​​of the determined verification parameters. The control device controls the operation of the robot device so that the robot device continues to perform multiple culture experiments according to the modified recipe.

[0140] This cell culture experimental method has the same effect as the cell culture experimental system according to embodiment 1, that is, it is possible to determine and implement the next recipe to be evaluated without waiting for the actual experimental results, thereby shortening the time required to obtain an appropriate set of environmental parameters.

[0141] Third Embodiment A program as an example of the present invention will be described. The program is as follows.

[0142] a recipe determination device that sets a recipe to be applied to each of a plurality of cell samples to be cultured; a control device that controls the operation of the robotic device so that the robotic device performs a culture experiment on a plurality of samples in accordance with the set recipe; and a database device that stores interim data for each sample in chronological order during a single culture experiment in which the culture results of the plurality of samples are determined, wherein the recipe determination device analyzes trends in the culture of each sample based on the interim data for each sample stored in the database device, and calculates, for each sample, a posterior probability that the parameter values ​​of a plurality of environmental parameters that make up the set recipe will likely cause future successful culture of the sample based on the analysis result of the trend in the culture of the sample and the difference between the chronological interim data for the sample, determines one or more environmental parameters with a relatively high calculated posterior probability as verification parameters, and modifies the recipe for each sample by changing the parameter value of the determined verification parameter, and the control device controls the operation of the robotic device so that the robotic device continues to perform a plurality of culture experiments in accordance with the modified recipe.

[0143] Furthermore, a computer-readable tangible non-transitory storage medium on which such a program is recorded is also an embodiment of the present invention.

[0144] According to such a program or storage medium, one or more computers or processors can read and execute the program, thereby achieving the same effect as the cell culture experimental system of embodiment 1, i.e., it becomes possible to determine and implement the next recipe to be evaluated without waiting for the actual experimental results, thereby shortening the time required to obtain an appropriate set of environmental parameters.

[0145] Thus, according to this embodiment, even during a cell culture experiment (round), the culture trends of each sample are analyzed based on the progress data obtained up to that point. Furthermore, the system estimates the probability of successful culture as a posterior probability based on the analysis results, and treats the estimated posterior probability as the actual experimental result and inputs it into the Bayesian estimation formula, thereby correcting the recipe for each sample (including parameters related to robotic operation). This system's processing makes it possible to determine and implement the next recipe to be evaluated without waiting for the actual experimental results, thereby shortening the time required to obtain an appropriate set of environmental parameters.

[0146] The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and are not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, or to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, the above-described configurations, functions, processing units, processing means, etc. may be implemented in hardware, in part or in whole, by designing, for example, an integrated circuit. Furthermore, the above-described configurations, functions, etc. may be implemented in software, by a processor interpreting and executing a program that realizes each function. Information such as programs, tables, and files that realize each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

[0147] 10... cell culture experimental system, 100... control device, 101... operation control unit, 102... detection control unit, 200, 201... terminal device, 300... robot device, 400... detection device, 401... imaging unit, 402... sensor unit, 500... database device, 550... recipe determination device, 600... cell culture equipment, 601... culture unit, 602... measurement unit, 603... chemical solution installation unit, 661... bottom surface of petri dish

Claims

1. A method for culturing a cell culture comprising: a recipe judgment device that sets a recipe to be applied to each of a plurality of cell samples to be cultured; a robot device; a control device that controls the operation of the robot device so that the robot device performs a culture experiment of the plurality of samples according to the set recipe; and a database device in which interim data for each of the samples is stored in chronological order during one culture experiment in which the culture results of the plurality of samples are determined, wherein the recipe judgment device analyzes a trend in the culture of each of the samples based on the interim data for each of the samples stored in the database device; calculates, for each of the samples, a posterior probability that indicates a degree of possibility that the parameter values ​​of a plurality of environmental parameters constituting the set recipe will cause future success in the culture of the sample based on a result of the analysis of the trend in the culture of the sample and a difference between the interim data of the sample in chronological order; determines one or a plurality of the environmental parameters with a relatively high calculated posterior probability as a verification parameter; and modifies the recipe for each of the samples by changing the parameter value of the determined verification parameter; and the control device controls the operation of the robot device so that the robot device continues to perform the culture experiment of the plurality of samples according to the modified recipe. Cell culture experimental system.

2. A cell culture experimental system as described in claim 1, further comprising a detection device for detecting information relating to the operation of the robotic device, and the intermediate progress data includes data representing the detected operation of the robotic device.

3. A cell culture experimental system as described in claim 2, wherein the data representing the operation of the robotic device includes data capable of identifying at least one of the trajectory of movement of an instrument handled by the robotic device, the speed of movement of the instrument, the acceleration of movement of the instrument, and the posture of the instrument.

4. A cell culture experimental system according to claim 3, wherein the cells are adherent cells, and the operation of the robotic device includes a step of detaching the adherent cells, a step of removing a specimen from a container, or a step of opening and closing a lid of a test tube.

5. A cell culture experimental system as described in claim 2, wherein the detection device is provided with an imaging unit that records the operation of the robotic device, and the intermediate progress data includes video data of the operation of the robotic device recorded by the imaging unit.

6. A cell culture experimental system as described in claim 1, comprising a detection device for detecting information relating to the operation of the robotic device, the detection device comprising a force sensor for detecting the magnitude of force accompanying the operation of the robotic device, and the intermediate progress data including data representing the magnitude of the force detected by the force sensor.

7. A cell culture experimental system according to claim 6, wherein the force sensor detects pressure when an instrument handled by the robot device comes into contact with an object.

8. A cell culture experimental system according to claim 2, wherein among the multiple devices constituting the cell culture experimental system, at least the control device, the robot device, and the detection device are synchronized based on a common clock signal.

9. The cell culture experimental system according to claim 8, wherein the time error representing the accuracy of the clock signal is 10 microseconds or less.

10. A cell culture experimental system as described in claim 1, wherein the recipe judgment device divides the plurality of samples into a verification group and an efficiency group, and reconfigures the composition of the samples in the verification group and the efficiency group so that the group of samples set with a combination of the type of environmental parameter and its parameter value that is considered to be more likely to contribute to successful cell culture based on the posterior probability of the samples has a higher culture success rate, and the group of samples set with a combination of the type of environmental parameter and its parameter value that is considered to be more likely to contribute to failure of cell culture has a lower culture success rate.

11. A cell culture experimental method, comprising: a recipe judgment device setting a recipe to be applied to each of a plurality of cell samples to be cultured; a control device controlling the operation of the robot device so that the robot device performs a culture experiment on the plurality of samples in accordance with the set recipe; a database device storing interim data for each of the plurality of samples in chronological order during one culture experiment in which the culture results of the plurality of samples are determined; the recipe judgment device analyzing a trend in the culture of each of the samples based on the interim data for each of the samples stored in the database device; calculating, for each of the samples, a posterior probability representing the degree of possibility that the parameter values ​​of a plurality of environmental parameters constituting the set recipe will cause future success in the culture of the sample based on a difference between a result of the analysis of the trend in the culture of the sample and the chronological interim data of the sample; determining, as a verification parameter, one or more of the environmental parameters for which the calculated posterior probability is relatively high; and modifying the recipe for each of the samples by changing the parameter value of the determined verification parameter; and the control device controlling the operation of the robot device so that the robot device continues to perform the culture experiment on the plurality of samples in accordance with the modified recipe.

12. A program for causing one or more computers or processors to function as: a recipe judgment device that sets a recipe to be applied to each of a plurality of cell samples to be cultured; a control device that controls the operation of the robot device so that the robot device performs a culture experiment on the plurality of samples according to the set recipe; and a database device in which interim data for each of the plurality of samples is stored in chronological order during one culture experiment in which the culture results of the plurality of samples are determined, wherein the recipe judgment device analyzes a trend in the culture of each of the samples based on the interim data for each of the samples stored in the database device; calculates, for each of the samples, a posterior probability representing the degree of possibility that the parameter values ​​of a plurality of environmental parameters constituting the set recipe will cause future success in the culture of the sample based on the analysis result of the trend in the culture of the sample and the difference between the chronological interim data of the sample; determines one or more of the environmental parameters for which the calculated posterior probability is relatively high as verification parameters; and modifies the recipe for each of the samples by changing the parameter value of the determined verification parameter. The control device controls the operation of the robotic device so that the robotic device continues to perform the incubation experiment on the multiple samples according to the modified recipe.

Images