Apparatus for determining optimal stem cell culture medium combination, method for determining optimal combination by calculating gene expression levels for culture medium combinations, and computer program

Abstract

According to any one of the above-mentioned technical solutions, for combinations of a plurality of culture medium conditions input from a user, a combination of optimal culture medium conditions for stem cells can be determined using a result distribution diagram of result values such as gene expression levels for cells cultured with the combinations of the culture medium conditions.

Description

Device for determining optimal stem cell culture medium combinations, method for determining optimal combinations by calculating gene expression levels for culture medium combinations, and computer program

[0001] An embodiment of the present disclosure relates to a device for determining an optimal stem cell culture medium combination, a method for calculating gene expression levels for culture medium combinations to determine an optimal combination, and a computer program, wherein, for a plurality of combinations of culture medium conditions input by a user, the method compares a plurality of combinations of culture medium conditions using gene expression levels in tumor cells cultured with the combinations of culture medium conditions to determine an optimal combination of culture medium conditions.

[0002] Cancer is one of the most common causes of death worldwide. Approximately 10 million new cases occur each year, accounting for approximately 12% of all deaths, making it the third leading cause of death.

[0003] To develop treatments for cancer, analysis and research on stem cells are necessary, so sample stem cells are cultured and used for research and analysis.

[0004] Cultured stem cells can be used to analyze the genetic or molecular characteristics of tumors. Furthermore, cultured stem cells can be utilized for new drug testing, studying tumor growth and metastasis mechanisms, and screening assays for drug discovery and testing. Furthermore, cultured stem cells can be used to create cancer models for in vitro and in vivo experiments, such as patient-derived organoids and xenografts.

[0005] Culture medium conditions are crucial for culturing stem cells, and the amount of cells cultured varies greatly depending on these conditions. Stem cell culture medium conditions vary depending on various factors, including stem cell type, desired experimental results, and resource availability. Stem cells cultured in inappropriate medium conditions cannot be used for research and / or analysis. In particular, gene expression levels also vary between stem cells cultured under different medium conditions.

[0006] In cells, overexpression of genes has the effect of increasing the regenerative capacity of the cells, and suppression of gene expression prevents the growth of malignant cells. In clinical treatment, suppression of expression is also used for treatment.

[0007] Therefore, there is a growing need to provide optimized culture medium conditions for the collected stem cells that match the expression level of a specific gene with the expression level of a reference gene.

[0008] The embodiments disclosed herein aim to provide a device and method for determining a combination of optimal culture medium conditions for cells by using a result distribution for result values ​​such as gene expression levels of cells cultured with a combination of multiple culture medium conditions input by a user.

[0009] According to embodiments of the present disclosure, a method may include a step in which a decision device obtains data from a user, including growth factors for first stem cells, a combination of first culture medium conditions including concentration values ​​of the growth factors, a combination of second culture medium conditions, a combination of third culture medium conditions, and a reference gene expression level of a first gene; a step in which the decision device inputs the combinations of the first to third culture medium conditions and information about the first stem cells into an optimal condition determination model, compares result distributions for the combinations of the first to third culture medium conditions, and outputs an optimal condition combination based on the gene expression level of the first gene; and a step in which the decision device outputs output data that visually outputs an optimal condition combination in which the first stem cells are cultured while being expressed at the reference gene expression level of the first gene.

[0010] The above optimal condition determination model may be a model trained using deep learning or machine learning, with inputs of result distributions generated based on combinations of multiple culture medium conditions for multiple stem cells and gene expression levels of the first gene in cells cultured under multiple culture medium conditions.

[0011] The above optimal condition determination model can generate a result distribution including the gene expression level of the first gene by sampling the experimental results cultured with each combination of culture medium conditions.

[0012] The above optimal condition decision model can generate distribution maps of results sampled using the Thompson sampling method.

[0013] The above optimal condition determination model may be a reinforcement learning model through a process of relearning using result values ​​including the gene expression level value of the first gene obtained by increasing the number of sampling times for culturing the first stem cell under the first to third culture medium conditions.

[0014] The above result distribution may be a visual representation of the gene expression levels obtained for a combination of culture medium conditions.

[0015] According to embodiments of the present disclosure, a device includes a memory, a communication unit, and a processor, wherein the processor obtains data from a user, including growth factors for first stem cells, a combination of first culture medium conditions including concentration values ​​of the growth factors, a combination of second culture medium conditions, a combination of third culture medium conditions, and a reference gene expression level of a first gene, inputs the combination of the first to third culture medium conditions and information on the first stem cells into an optimal condition determination model, compares result distributions for combinations of the first to third culture medium conditions, outputs an optimal condition combination based on the gene expression level of the first gene, and executes commands to output output data that visually outputs the optimal condition combination in which the first stem cells are cultured while being expressed at the reference gene expression level of the first gene.

[0016] A computer program according to an embodiment of the present invention can be stored in a medium to execute any one of the methods according to an embodiment of the present invention using a computer.

[0017] In addition, other methods for implementing the present invention, other systems, and computer-readable recording media recording a computer program for executing the above methods are further provided.

[0018] Other aspects, features and advantages other than those described above will become apparent from the following drawings, claims and detailed description of the invention.

[0019] According to any one of the aforementioned problem solving means, for a plurality of combinations of culture medium conditions input by a user, a combination of optimal culture medium conditions for stem cells can be determined by using a result distribution for results such as differentiation success rate, reproducibility, and gene expression level of cells cultured with the combinations of culture medium conditions.

[0020] Figure 1 is a block diagram of a device for determining an optimal stem cell culture medium combination according to embodiments of the present disclosure.

[0021] Figure 2 is a block diagram of the optimal condition determination unit.

[0022] Figure 3 is a block diagram of a learning device that learns an optimal condition decision model.

[0023] Figure 4 is an exemplary diagram of combinations of culture medium conditions for stem cells to be differentiated into vascular cells.

[0024] FIG. 5a, FIG. 5b, and FIG. 5c are exemplary drawings of result distributions generated according to embodiments of the present disclosure.

[0025] Figure 6 is a flowchart of an optimal condition output method according to embodiments of the present disclosure.

[0026] Figure 7 is a flowchart of an optimal condition output method according to other embodiments of the present disclosure.

[0027] The configuration and operation of the present invention will be described in detail with reference to embodiments of the present invention illustrated in the attached drawings below.

[0028] The present invention is capable of various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. The effects and features of the present invention, as well as the methods for achieving them, will become clearer with reference to the embodiments described in detail below, along with the drawings. However, the present invention is not limited to the embodiments disclosed below and can be implemented in various forms.

[0029] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. When describing with reference to the drawings, identical or corresponding components are given the same drawing reference numerals, and redundant descriptions thereof will be omitted.

[0030] Hereinafter, the term “upper” or “upper” may include not only things that are directly above in contact, but also things that are above in a non-contact manner.

[0031] In the following examples, the terms first, second, etc. are not used in a limiting sense, but are used for the purpose of distinguishing one component from another.

[0032] In the examples below, singular expressions include plural expressions unless the context clearly indicates otherwise.

[0033] In the following examples, terms such as “include” or “have” mean that a feature or component described in the specification is present, and do not preclude the possibility that one or more other features or components may be added.

[0034] For convenience of explanation, the sizes of components in the drawings may be exaggerated or reduced. For example, the sizes and thicknesses of each component shown in the drawings are arbitrarily indicated for convenience of explanation, and thus the present invention is not necessarily limited to what is shown.

[0035] Additionally, terms such as “…part”, “…area”, etc. described in this specification may mean a unit that processes at least one function or operation.

[0036] Stem cell culture can be used in the treatment of diseases such as cancer. Cultured stem cells can be used to study characteristics such as tumor initiation, growth, and resistance. Cultured stem cells can be used to conduct high-throughput drug screening to study compounds that can target cancer cells. Cultured stem cells can be used to establish patient-derived cancer cell lines and conduct tests for drug sensitivity. Cultured stem cells can also be used to study intratumoral heterogeneity.

[0037] Stem cells, which are utilized in various ways as described above, can exhibit varying levels of gene expression in differentiated cells depending on changes in culture conditions. Because stem cells can differentiate into different cell types depending on growth factors, concentrations, and media, culture conditions for stem cells can include the composition of the culture medium, temperature, pH, oxygen concentration, growth factors, hormones, cytokines, media type, and the timing of factor administration.

[0038] In this specification, genomic data may include WES (Whole exome sequencing) and sequencing data, and may include data obtained by various analysis methods.

[0039] WES (Whole exome sequencing) sequences approximately 50 Mbp at approximately 100X, specifically obtaining only the exome region, and can obtain a relatively greater read depth than WGS (Whole genome sequencing). It is a method of analyzing only the genome of the exon region that directly codes for proteins. In the case of humans, exons account for less than 2% of the entire genome, so it is cheaper to acquire than WGS and the data capacity is also reduced. Even when analyzing WGS, variants in non-coding regions are difficult to interpret and may produce highly meaningful results. To solve this problem, analysis using WES is more economical and can obtain information at a higher depth.

[0040] RNA is an abbreviation for ribonucleic acid, and together they are called nucleic acids. Nucleic acids are polymers composed of multiple nucleotides. The nucleotides that make up DNA (deoxyribonucleic acid) are specifically called deoxyribonucleotides, and the nucleotides that make up RNA (ribonucleic acid) are specifically called ribonucleotides.

[0041] FIG. 1 is a block diagram of an optimal stem cell culture medium combination determination device (100, hereinafter, determination device) according to embodiments of the present disclosure.

[0042] The decision device (100) according to embodiments of the present disclosure can determine optimal culture medium conditions for stem cells. The decision device (100) can determine optimal culture medium conditions by considering cells to be differentiated and the gene expression levels in the cells. A user may want to culture cells with a reference gene expression level. Conventionally, a user could culture a large number of tumor cells to be used in research, and then conduct research with cells with a high gene expression level of the desired gene, or conduct research by adjusting the gene expression level to a desired range by inserting a gene from another cell using genetic scissors. The decision device (100) according to embodiments of the present disclosure can determine culture medium conditions for culturing cells in which the gene expression level of the gene to be used in research matches the reference gene expression level. At this time, the decision device (100) can output optimal culture medium conditions for stem cells considering the gene expression level by utilizing an optimal condition determination model learned to determine culture medium conditions.

[0043] The decision device (100) obtains, through various methods, result values ​​such as differentiation success rate, reproducibility, and gene expression level success rate of cells cultured with combinations of one or more culture medium conditions, generates result distributions of result values ​​for combinations of multiple culture medium conditions, and uses these result distributions to determine an optimal combination of culture medium conditions among the combinations of culture medium conditions. The decision device (100) can sample result values ​​for combinations of culture medium conditions to generate result distributions for the combinations. The decision device (100) can determine an optimal combination of culture medium conditions for a given stem cell through a process of updating the result distribution by updating the result values ​​for the combinations of culture medium conditions. The result value for the combination may be a value for cells cultured with the culture medium conditions, and may be a value calculated by combining differentiation success rate, gene expression level success rate, and the like. The result values ​​for the combination can be calculated by a preset mathematical formula for calculating the success rate of the division, the success rate of the gene expression level, etc. At this time, the result device (100) can update the result distribution by applying compensation for the Bayesian probability distribution to the result values ​​cultured with combinations of culture medium conditions.

[0044] The decision device (100) can obtain data on a combination of culture medium conditions for a given stem cell and data on a gene from a user's input. Here, the information on the gene can include a gene, a reference gene expression level, etc. The gene can be at least one of genes expressed in the stem cell. The reference gene expression level is a value for the expression level at which the corresponding gene should be expressed, and can include a specific value, above the average value, below the average value, twice the average value, half the average value, etc. The gene expression level input by the user can be referred to as the reference gene expression level. The user can obtain data on a combination of stem cells to be cultured and culture medium conditions that determine the reference gene expression level of a specific gene so that the specific gene is expressed in the stem cell at the desired reference gene expression level through the decision device (100). In an optional embodiment, the decision device (100) can generate combinations of a plurality of culture medium conditions by changing the concentration values ​​of each growth factor based on the growth factors, which are conditions of the culture medium input by the user. At this time, the result distributions for the combinations of the first and fourth culture medium conditions input by the user can be compared to generate a new combination of fifth culture medium conditions by utilizing the concentration values ​​in the combinations of the first and second culture medium conditions that satisfy a specific condition. Finally, a process of comparing the result distributions, which are the results of culturing with the combinations of the first to fourth culture medium conditions and the newly generated fifth culture medium conditions, can be performed again. As illustrated in FIG. 5b, when condition 1, condition 2, condition 3, and condition 4 are input, condition 5, which is a new combination of culture medium conditions, can be generated, and the gene expression success rate, which is the result of culturing with condition 5, can be included in the result distribution.The decision device (100) can generate a result graph by removing the result graph of condition 3, which has the worst result value among the combinations of five conditions, from the result distribution (see FIG. 5c).

[0045] Culture medium conditions may include growth factors, concentration values ​​of growth factors, information on supplements added during culture, timing (time point) of addition of each growth factor, medium, etc. Additionally, culture medium conditions may further include temperature, pH, oxygen concentration, growth factors, hormones, cytokines, biochemical factors, etc. Culture medium conditions may further include numerical values ​​representing the degree, size, concentration, etc. of each of the above factors.

[0046] The decision device (100) can generate result distributions by acquiring result values ​​for cells cultured with combinations of culture medium conditions for stem cells from culture experiments or from a database. The decision device (100) can determine an optimal combination of culture medium conditions for stem cells using the result distributions generated in this way. At this time, the result distributions for a plurality of combinations of culture medium conditions can be generated through a process of resampling using result values ​​of additional culture experiments that are performed, predicted, or produced with each combination of culture medium conditions. The decision device (100) can determine an optimal combination of culture medium conditions for stem cells using an optimal condition determination model that is learned with result values ​​of culture experiments performed with combinations of culture medium conditions. As mentioned above, the optimal condition determination model can be designed to output a new combination of culture medium conditions as an optimal combination of culture medium conditions in addition to the combinations of input culture medium conditions. Here, the combination of new culture medium conditions may be derived from various combinations, such as the average value of values ​​included in previously input combinations, the weighted average value of values, etc.

[0047] The decision device (100) can store information about the differentiation process for cells to be differentiated in advance. The decision device (100) can generate a combination of additional conditions by setting the injection points of growth factors to predetermined points based on the information about the differentiation process. Since each cell differentiates into mature cells through a predetermined process, information about the differentiation process is required to infer a combination of culture medium conditions that express the differentiation success rate, gene expression level, etc. of stem cells at a reference gene expression level.

[0048] The optimal condition decision model can be trained to output a result distribution map for the culture medium conditions by taking as input a combination of conditions including information about stem cells, growth factors, concentration values ​​for the growth factors, injection points, and the result values ​​of cells cultured under a combination of culture medium conditions. The optimal condition decision model can be constructed to sample the result distribution map for each combination of conditions and output the optimal combination of conditions. The optimal decision model can periodically update the input data to update the result distribution map for each combination of culture medium conditions, and output the optimal combination of conditions using the result distribution map.

[0049] At this time, the learned optimal condition decision model can output a combination of culture medium conditions that control the gene expression level of a specific gene by inferring factors of culture medium conditions that are highly correlated with cell differentiation success rate, reproducibility, and gene expression levels. The factors of culture medium conditions with high correlation may vary depending on the type of gene. The learned optimal condition decision model can learn factors of culture medium conditions that are correlated with each of one or more genes expressed in the cell. For example, the learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of MYC in cells such as Burkitt's lymphoma, breast cancer, lung cancer, and colon cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of HER2 in cells such as breast cancer and stomach cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of EGFR in cells such as non-small cell lung cancer, colon cancer, and head and neck cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of BRAF in cells such as malignant melanoma, colon cancer, and thyroid cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of Cyclin D1 in cells such as breast cancer, B-cell lymphoma, and esophageal cancer.

[0050] Information about stem cells may include the type of stem cell, the type of cell to be differentiated (insulin-producing cells, retinal cells, fat cells, cardiac muscle cells, skin cells, bone cells, etc.), information related to the stem cell target (age, sex, etc.), information about mutations, the types of genes expressed in stem cells, and information about the expression level of each gene (average value, expression level at the time of cancer onset).

[0051] Examples of growth factors in the culture medium conditions may further include VEGF (Vascular Endothelial Growth Factor), EGF (Epidermal Growth Factor), BMPs (Bone Morphogenetic Proteins), fatty acids, temperature, pH value, oxygen concentration, glucose concentration, illumination conditions, and concentration values. For example, the culture medium conditions for controlling the expression level of a certain gene for cultured cells may include BDNF (Brain-Derived Neurotrophic Factor) and may include the timing of injection of BDNF (Brain-Derived Neurotrophic Factor). The differentiation process of stem cells into vascular cells may include a cell aggregation stage, a mesoderm induction stage, a vascular lineage induction stage, a vascular network formation stage, and a cell maturation stage. In order to obtain a desired gene expression level for the cultured cells, the timing of injection of growth factors may be selected by each of the stages of the cell aggregation stage, the mesoderm differentiation stage, the vascular lineage differentiation stage, the vascular network formation stage, and the cell maturation stage, and the injection date (first day, second day, etc.) among the stages. Not limited to the above examples, various conditions can be changed with respect to the injection time in order to control the gene expression level of the gene.

[0052] As illustrated in FIG. 1, the decision device (100) according to some embodiments may include a processor (110), an input / output unit (130), a memory (140), a communication unit (150), and an optimal condition determination unit (120). However, not all of the components illustrated in FIG. 1 are essential components of the decision device (100). The decision device (100) may be implemented with more components than the components illustrated in FIG. 1, or with fewer components than the components illustrated in FIG. 1. The decision device (100) may be a user terminal, a server, a network system, or a separate device.

[0053] According to one embodiment of the present disclosure, the processor (110) typically controls the overall operation of the decision device (100). For example, the processor (110) may control the components included in the decision device (100) by executing a program stored in the decision device (100).

[0054] According to one embodiment of the present disclosure, the processor (110) can determine optimal culture medium conditions for stem cells. The processor (110) can determine optimal culture medium conditions by considering cells to be differentiated and the level of gene expression in the cells. The processor (110) can output optimal culture medium conditions for stem cells by utilizing an optimal condition determination model learned to determine a combination of culture medium conditions for culturing cells in which a gene to be utilized in research is expressed at a reference gene expression level.

[0055] The processor (110) can obtain, for combinations of culture medium conditions input by a user, result values ​​such as differentiation success rate, reproducibility, gene expression level, and success rate of gene expression level of cells cultured with combinations of culture medium conditions by various methods such as obtaining them from a database, obtaining them experimentally, or predicting them from experimental values. The processor (110) can generate a result distribution including result values ​​for combinations of multiple culture medium conditions and determine a combination of culture medium conditions in which a gene expression level is cultured at a reference gene expression level among the combinations of culture medium conditions. The processor (110) can generate a result distribution by sampling result values ​​for combinations of culture medium conditions. The processor (110) can update the result values ​​for combinations of culture medium conditions to update the result distribution, and determine an optimal combination of culture medium conditions for stem cells expressed at a reference gene expression level based on the result distributions. Stem cells cultured with a combination of optimal culture medium conditions can include genes expressed at a reference gene expression level. At this time, the processor (110) can update the result distribution using compensation for the Bayesian probability distribution.

[0056] The processor (110) can obtain data on a combination of culture medium conditions for a given stem cell and data on a gene from a user's input. Here, the data on the gene can include a gene, a reference gene expression level, etc. The gene can be at least one of genes expressed in the stem cell. The reference gene expression level is a value for the expression level at which the corresponding gene should be expressed, and can include a specific value, above the average value, below the average value, twice the average value, half the average value, etc. The user can obtain data on a combination of culture medium conditions that determine the stem cell to be cultured and the gene expression level of a specific gene so that the specific gene is expressed in the stem cell at the desired reference gene expression level through the processor (110). The processor (110) can additionally generate a plurality of combinations of culture medium conditions by changing the concentration value of each growth factor based on the growth factors, which are conditions of the culture medium input by the user. The combination of culture medium conditions to be generated can be generated based on several combinations determined based on the result values ​​for the input culture medium conditions. At this time, by using the result distribution including the result values ​​for the combinations of the first and fourth culture medium conditions input by the user, the culture medium conditions, i.e., concentration values, injection times, chemical factors, etc., in the combination of the first and second culture medium conditions that satisfy a specific condition, for example, the combination with the best result value, can be variously utilized to generate a new combination of fifth culture medium conditions. Finally, a process of comparing the result distributions, which are the results of culture with the combinations of the first to fourth culture medium conditions and the newly generated combination of the fifth culture medium conditions, can be performed again. The culture medium conditions can include growth factors, concentration values ​​of the growth factors, information on supplements added during culturing, the injection timing (time point) of each growth factor, the medium, etc.Additionally, culture medium conditions may further include temperature, pH, oxygen concentration, growth factors, hormones, cytokines, biochemical factors, etc. The culture medium conditions may further include numerical values ​​representing the degree, size, concentration, etc. of each of the above factors.

[0057] The processor (110) can acquire the result values ​​of the results of culturing with combinations of culture medium conditions of stem cells from a culture experiment or a database to generate a result distribution. The processor (110) can determine an optimal combination of culture medium conditions for stem cells using the generated result distributions. At this time, the result distributions for a plurality of combinations of culture medium conditions can be generated through a process of resampling using the result values ​​of additional culture experiments that are executed, predicted, or produced with each combination of culture medium conditions. The processor (110) can determine an optimal combination of culture medium conditions for stem cells using an optimal condition determination model that is learned with the result values ​​of culture experiments executed with combinations of culture medium conditions. As mentioned above, the optimal condition determination model can be designed to output a newly generated combination of culture medium conditions in addition to the input combinations of culture medium conditions as an optimal combination of culture medium conditions.

[0058] The processor (110) can store information about the differentiation process for cells in advance. The processor (110) can randomly change the injection times of growth factors based on the information about the differentiation process to generate combinations of set conditions. The combinations of conditions with changed injection times can also be determined based on the result values ​​for the combinations of input conditions. If the result values ​​for stem cells cultured at the first injection time point and the second injection time point correspond to the first and second ranks among all the result values, a combination that uses any time point among the first injection time point and the second injection time point as the injection time point can be additionally generated. The arbitrary time point can be determined by processing the first injection time point and the second injection time point by a method such as an average or a weighted average. Since each cell differentiates into mature cells through a set process, information about the differentiation process of the cells is required to infer culture medium conditions that increase the differentiation success rate and the gene expression success rate of stem cells.

[0059] The processor (110) is a component for controlling the overall operation of the decision device (100). Specifically, the processor (110) controls the overall operation of the decision device (100) using various programs stored in the storage medium (150) of the decision device (100). For example, the processor (110) may include a CPU, RAM, ROM, and a system bus. Here, the ROM is a component in which a command set for system booting is stored, and the CPU copies the stored operating system of the decision device (100) to the RAM according to the command stored in the ROM and executes the O / S to boot the system. When the system booting is complete, the CPU can copy various stored applications to the RAM and execute them to perform various operations. Although the service provision server (100) has been described as including only one CPU in the above, it may be implemented with multiple CPUs (or DSPs, SoCs, etc.) during implementation.

[0060] According to one embodiment of the present invention, the processor (110) may be implemented as a digital signal processor (DSP), a microprocessor, or a time controller (TCON) that processes digital signals. However, the present invention is not limited thereto, and may include one or more of a central processing unit (CPU), a micro controller unit (MCU), a micro processing unit (MPU), a controller, an application processor (AP), a communication processor (CP), or an ARM processor, or may be defined by the corresponding terms. In addition, the processor (110) may be implemented as a system on chip (SoC) or large scale integration (LSI) having a built-in processing algorithm, or may be implemented in the form of a field programmable gate array (FPGA).

[0061] According to one embodiment of the present disclosure, the input / output unit (130) can display an interface generated by the memory (140) of the decision device (100). According to one embodiment of the present invention, the input / output unit (130) can display a user interface for an input user input. The input / output unit (130) can output stored graphic data, visual data, auditory data, and vibration data under the control of the memory (140).

[0062] The input / output unit (130) may be implemented as a display panel of various forms. For example, the display panel may be implemented as a display technology of various forms, such as a liquid crystal display (LCD), an organic light emitting diode (OLED), an active-matrix organic light-emitting diode (AM-OLED), a liquid crystal on silicon (LCoS), or a digital light processing (DLP). In addition, the input / output unit (130) may be coupled to at least one of the front area, the side area, and the rear area of ​​the display panel in the form of a flexible display.

[0063] The input / output unit (130) may be implemented as a touch screen with a layer structure. The touch screen may have a function of detecting not only a display function but also a touch input location, a touched area, and even a touch input pressure, and may also have a function of detecting not only a real touch but also a proximity touch.

[0064] The input / output unit (130) may include a user interface for inputting various information into the decision device (100).

[0065] According to one embodiment of the present disclosure, the memory (140) may store a program for processing and controlling the processor (110) and / or the optimal condition determination unit (120), and may also store data input to or output from the determination device (100). According to one embodiment of the present disclosure, the memory (140) may store information on the differentiation process, information on genes, information on culture medium conditions of stem cells, information on growth factors, generated result distributions, etc. The memory (140) may include a database storing the above information.

[0066] According to one embodiment of the present disclosure, the memory (140) may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., SD or XD memory, etc.), a RAM (Random Access Memory), a SRAM (Static Random Access Memory), a ROM (Read-Only Memory), an EEPROM (Electrically Erasable Programmable Read-Only Memory), a PROM (Programmable Read-Only Memory), a magnetic memory, a magnetic disk, and an optical disk. In addition, according to one embodiment of the present disclosure, programs stored in the memory (140) may be classified into a plurality of modules according to their functions.

[0067] According to one embodiment of the present disclosure, the communication unit (150) can communicate with an external device of the processor (110). For example, the communication unit (150) can communicate with an external device, such as a payment server or an authentication server, under the control of the processor (110). In addition, the communication unit (150) can also obtain user information or user input through communication with an external interface.

[0068] The optimal condition determination unit (120) may include a data input unit (121), an optimal combination output unit (122), a result update unit (123), and an output data generation unit (124), as illustrated in FIG. 2.

[0069] The data input unit (121) can obtain combinations of multiple stem cell culture medium conditions, information on stem cells, information on genes, and information on reference gene expression levels from user input. The information on genes may include at least one of genes expressed in stem cells. The information on reference gene expression levels may include values ​​for expression levels at which the corresponding genes should be expressed, such as a specific value, above the average, below the average, twice the average, half the average, etc. The combination of culture medium conditions may include growth factors, concentration values, medium, injection timings of growth factors, etc. For example, the combination of culture medium conditions may include, but is not limited to, growth factors CHIR, BMP4, concentration values ​​of CHIR, concentration values ​​of BMP4, injection timings of CHIR, and injection timings of BMP4, as illustrated in FIG. 4, and may include various culture medium conditions.

[0070] The data input unit (121) can input data including combinations of multiple conditions and reference gene expression levels of genes. This data can include information about genes and information about the gene expression levels of the corresponding genes. The data input unit (121) can input data about combinations of conditions including growth factors. The data input unit (121) can input data about combinations of conditions including stem cell information and growth factors.

[0071] The optimal combination output unit (122) can input data including combinations of multiple stem cell culture medium conditions, gene expression levels of genes, etc. into the optimal condition determination model, and sample the result distributions for the gene expression levels of genes for the combinations. The optimal condition determination model can generate a large amount of data for the results of culturing with each combination of culture medium conditions. The optimal condition determination model can output the optimal condition combination based on the result data for each combination. The optimal condition determination model can compare the result distributions for the combinations of the first and fourth culture medium conditions input by the user, and can generate a new combination of fifth culture medium conditions by utilizing the culture medium conditions, for example, concentration values, in the combination of the first and second culture medium conditions that satisfy a specific condition. Finally, a process of comparing the result distributions, which are the results of culturing with the combinations of the first to fourth culture medium conditions and the newly generated combination of the fifth culture medium conditions, can be performed again. Through the above additional process, the optimal condition determination model can output an optimal combination of conditions in which the genes in the stem cells desired by the user are cultured at a set gene expression level.

[0072] The result update unit (123) can update the result distributions using the compensation data for the result values ​​for stem cells cultured under the optimal condition combination in the optimal decision model. In other words, the result update unit (123) can input the result distribution for the combination of the newly added fifth culture medium conditions into the optimal decision model and process it to update the optimal decision model. The result update unit (123) can update the result distributions for the result values ​​for stem cell culture through random sampling based on the Bayesian probability distribution. The result values ​​for stem cell culture can include the differentiation success rate of cultured stem cells (number of cells that successfully differentiated / total number of cells), the oncogenic mutation match rate with patient tissue, the reproducibility of stem cells (comparison of size, number, etc. with actual stem cells, correlation coefficient), tumor purity, gene expression level of a gene, etc. The result values ​​for stem cell culture can be at least one of the gene expression level value among the result values ​​for stem cells, the average value of the gene expression level values ​​for stem cells, and the combined value. Here, tumor purity can be determined using in-silico tumor purity estimation algorithms. More specifically, tumor purity can be calculated using DNA-seq-based methods such as ABSOLUTE, Sequenza, and FACETS, as well as RNA-seq (expression)-based methods such as ESTIMATE and PUREE.

[0073] Here, the cancer mutation match rate with patient tissue can be calculated based on mutant genes defined in a previously registered database. Examples of such databases include, but are not limited to, OncoKB and COSMIC DB.

[0074] The success rate and / or reproducibility of stem cell differentiation can be calculated based on the results obtained from the cultured stem cells. The success rate and / or reproducibility of stem cell differentiation can be calculated using established rules, mathematical formulas, etc.

[0075] The result update unit (123) can update the optimal condition determination model to update the result distributions based on the compensation data.

[0076] The optimal combination output unit (122) can output an optimal combination of conditions in which a specific gene is cultured to have a set gene expression level in stem cells input by a user using an optimal condition determination model. The optimal condition determination model can output result distributions generated based on the gene expression level for each combination of input conditions. The Y-axis of the result distribution can be determined by the size of the gene expression level or the degree of matching with the gene expression level requested by the user. The X-axis of the result distribution can be determined by a value for the combinations. The optimal condition determination model can compare these result distributions and output an optimal combination of conditions having a gene expression level set by the user. The optimal combination output unit (122) can generate result distributions for combinations of additional conditions in addition to the combinations of conditions input by the user. The optimal combination output unit (122) can generate a new combination of conditions based on the result distributions for the combinations of input conditions. The optimal combination output unit (122) can remove the result value for the combination with the lowest result value (based on the X-axis) from the result distribution. The optimal combination output unit (122) can repeat the process of adding result values ​​for new combinations and deleting the combination with the lowest result value. In the optimal condition determination model, a new combination of conditions can be generated based on the correlation for the result values ​​of gene expression levels. Machine learning can be used to learn how to process factors included in the input conditions.

[0077] The output data generation unit (124) can generate output data including the optimal condition combination from the optimal combination output unit (122).

[0078] The decision device (100) according to embodiments of the present disclosure can output an optimal condition combination from among combinations of multiple conditions devised by researchers and combinations of newly added conditions. When cultured using the optimal condition combination, stem cells input by the user can be cultured with a predetermined gene expression level. The decision device (100) can output a combination of culture medium conditions that include optimal values ​​for concentration values ​​of growth factors of conditions for stem cells and injection time points of growth factors (information on differentiation stage). Additionally, the decision device (100) can output a combination of culture medium conditions that include the injection time points of concentration values ​​of growth factors of conditions for stem cells as optimal time points. The decision device (100) can reapply the result data obtained by experimentally culturing with the combination of conditions to update the result distribution. The decision device (100) can output an optimal condition combination using the updated result distribution. The resulting distribution can be generated by sampling the resulting values ​​for combinations of conditions. The resulting distribution can be generated for each combination, or can include the resulting values ​​of the combinations. Methods for sampling the cultivated resulting values ​​include, but are not limited to, Thomson sampling, greedy methods, UCB (upper confidence bound), Bayesian optimization, Monte Carlo sampling, and bootstrapping. Various sampling methods can be used. The resulting distribution can be generated by beta-distributing the resulting values ​​(gene expression levels of genes) for combinations of conditions.

[0079] Figure 3 is a block diagram of a learning device (300) that learns an optimal condition decision model.

[0080] The learning device (300) can train an optimal condition determination model. The learning device (300) can create a Gaussian process model that defines the relationship between combinations of culture medium conditions and gene expression levels, and train a model that outputs an optimal condition combination using a predetermined acquisition function. At this time, the acquisition function may include a function such as the Upper Confidence Bound. The acquisition function can calculate the uncertainty of the prediction and output an output value based on the uncertainty. The learning device (300) can train a model that outputs an optimal condition combination using a Bayesian optimization method. The trained optimal condition determination model can generate result distributions for each combination of input conditions using various sampling methods, and compare the result distributions for the combinations to output an optimal condition combination. More specifically, the optimal condition determination model can generate result distributions using the Thomson sampling method. The trained optimal condition determination model may further include a module that generates a new combination of culture medium conditions using the combinations of input culture medium conditions. The module that generates these combinations can generate new concentration values ​​based on the concentration values ​​of growth factors, or can generate them using combinations of two culture medium conditions in order of best results. A new combination of culture medium conditions can be generated to include at least one of a new concentration value, a new injection time point, and a new growth factor. A new value can mean a value that is not included in the input data. The logic for generating a new combination of culture medium conditions can also be changed depending on the result distribution for the new combination of culture medium conditions. For example, if the result distribution for a new combination of culture medium conditions including the average value of the concentration values ​​has a result value greater than a predetermined reference value, a new combination of culture medium conditions can be generated by setting the average value of the concentration values.If the result distribution for a combination of new culture medium conditions including the midpoint of the injection time points has a result value greater than or equal to a predetermined reference value, a new combination of culture medium conditions can be created by setting the midpoint of the injection time points. If the result distribution for a combination of new culture medium conditions has a result value greater than or equal to a predetermined reference value by removing a specific growth factor from the combination of culture medium conditions having the lowest result value, a new combination of culture medium conditions can be created by removing the specific growth factor.

[0081] The result distribution may be a two-dimensional data representation of the result values ​​cultured with each combination. The result distribution may be generated as two-dimensional data with the x-axis representing the success rate for the gene expression level of the gene and the y-axis representing the number of occurrences (density), as illustrated in Fig. 5a. The result distribution may be expressed as a sampling of the result values ​​cultured with each combination. The result value for stem cells may include the expression success rate of cultured stem cells (the number of cells that successfully expressed / the total number of cells), etc. The expression success rate may be a value for the probability that the gene is expressed with the expression level desired by the user. The y-axis value of the result distribution may be at least one of the result values ​​for the stem cells, the average value, or the combined value of the result values ​​for the stem cells. Here, the tumor purity may be a value determined using in-silico-based tumor purity estimation algorithms. More specifically, tumor purity can be calculated using DNA-seq-based methods such as ABSOLUTE, Sequenza, and FACETS, and RNA-seq (expression)-based methods such as ESTIMATE and PUREE.

[0082] The dataset generation unit (310) can generate training data sets using combinations of conditions including information on stem cells, information on genes (genes, gene expression levels), growth factors and concentration values ​​for the growth factors, injection times of the growth factors, and culture result values. For example, training data sets can be generated using combinations of four combinations of conditions for stem cells of lung cancer cells, gene expression levels of genes, different concentration values ​​for growth factors for stem cells of breast cancer cells, injection times of growth factors, etc. The dataset generation unit (310) can generate training data sets using combinations of different growth factors according to the type of cells to be differentiated and the desired gene expression levels.

[0083] The model learning unit (320) uses training data sets as input and generates, for each combination, distributions of results sampled using a sampling method such as Thompson sampling, and can train an optimal condition determination model using the distributions of results for each combination generated in this way.

[0084] The optimal condition determination model can output an optimal condition combination by inputting combinations of conditions including information about stem cells, growth factors and their concentration values, and the timing of changes in the concentration values. The timing of change can include a value for the injection time.

[0085] The model learning unit (320) can learn an optimal condition determination model that generates a result distribution using data obtained by performing a culture experiment with a combination of input conditions for stem cells.

[0086] The data obtained through the culture experiment may be data stored in a registered database and / or data experimentally obtained by the research team.

[0087] The model learning unit (320) can learn a model that generates a result distribution by statistically sampling the result values ​​for the expression levels of genes in stem cells cultured under conditions including growth factors, concentration values ​​for growth factors, injection time points for growth factors, etc.

[0088] The model learning unit (320) can build an optimal condition determination model that compares the result distributions for each combination of conditions and outputs the optimal combination of conditions.

[0089] The model learning unit (320) can update the result distributions by changing the number of samplings performed for culture experiments based on a combination of conditions. For example, the number of samplings may increase as the training data set grows. Results from an increased number of samplings determine a combination of culture medium conditions for stem cells with a higher success rate of gene expression compared to results from a lower number of samplings. The success rate of gene expression may be a value calculated based on whether a given gene has a predetermined reference gene expression level.

[0090] As shown in Fig. 5a, the larger the x-axis value at the peak point of the result distribution and the larger the y-axis value, the more optimal the combination of culture medium conditions can be. In Fig. 5a, condition 1 is the optimal combination of conditions.

[0091] The model learning unit (320) can generate a combination of fifth culture medium conditions using combinations of the first to fourth culture medium conditions, and can generate result distributions for the combination of fifth culture medium conditions of FIG. 5b. According to the result distribution of FIG. 5b, considering the location of the peak, condition 1, which has the highest success rate and density, may be the optimal condition combination. The model learning unit (320) can generate a result distribution as in FIG. 5c, which compares the combinations of the first to fifth culture medium conditions and eliminates the combination of the third culture medium conditions with the worst results. For the result distribution of FIG. 5c, an optimal condition determination model that additionally performs a process of generating new culture medium conditions can be trained.

[0092] The model learning unit (320) can build an optimal condition determination model that outputs an optimal condition combination using the result distributions of Fig. 5c updated in this way.

[0093] The model learning unit (320) can periodically update the results of experimental cultivation using a combination of input conditions to update the result distribution.

[0094] The learning device (300) can generate an optimal condition determination model through the above learning process and transmit it to the optimal condition determination device (100) to output an optimal condition combination.

[0095] Figure 4 is an exemplary diagram of combinations of culture medium conditions for stem cells to be differentiated into vascular cells.

[0096] Figure 6 is a flowchart of an optimal condition output method according to embodiments of the present disclosure.

[0097] In S110, the decision device (100) can receive and acquire data including a combination of first culture medium conditions for first stem cells, a combination of second culture medium conditions, a combination of third culture medium conditions, and a reference gene expression level value of a first gene from a user. The combination of the first to third culture medium conditions may include one or more growth factors, concentration values ​​for each of the growth factors, injection time points of the growth factors, a medium, etc. Additionally, the culture medium conditions may further include temperature, pH, oxygen concentration, growth factors, hormones, cytokines, biochemical factors, etc. The culture medium conditions may further include numerical values ​​representing the degree, size, concentration, etc. of each of the above factors. The medium information may include information on one or more growth factors included in the medium. The information on the injection time points of the growth factors may include information on the injection time points in the differentiation stages.

[0098] In S120, the decision device (100) inputs combinations of first to third culture medium conditions and first stem cells into an optimal condition determination model, and can output a result distribution for combinations of the first to third culture medium conditions and one optimal condition combination among the combinations of the first to third culture medium conditions.

[0099] The optimal condition determination model can acquire result data from experimental culture of first stem cells using combinations of first to third culture medium conditions through various routes, and can sample the result data acquired in this manner to generate first to third result distributions. The optimal condition determination model can compare the first to third result distributions to determine a result distribution having a higher result value (success rate of gene expression level) and determine and output a combination of culture medium conditions of the result distribution as an optimal condition combination. The success rate of gene expression level can be a value that probabilistically calculates the degree of matching with a reference gene expression level. The gene expression level within a predetermined error range can be counted as the number of successful cells. The optimal condition determination model can update the first to third result distributions by updating the result data from experimental culture of first stem cells. The optimal condition determination model can learn factors of culture medium conditions that are correlated with each of one or more genes expressed in the cells. For example, the learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of MYC in cells such as Burkitt's lymphoma, breast cancer, lung cancer, and colon cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of HER2 in cells such as breast cancer and stomach cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of EGFR in cells such as non-small cell lung cancer, colon cancer, and head and neck cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of BRAF in cells such as malignant melanoma, colon cancer, and thyroid cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of Cyclin D1 in cells such as breast cancer, B-cell lymphoma, and esophageal cancer.

[0100] In S130, the decision device (100) can output output data that visually displays the optimal condition combination for the first stem cell. There may be two or more combinations of input culture medium conditions. If the reference gene expression level for the stem cell changes, the output optimal condition combination may also change.

[0101] Through this, the optimal condition determination method according to the embodiments of the present disclosure can determine a combination of culture medium conditions predicted to have an optimal result value (such as a success rate of gene expression level) among several combinations of culture medium conditions suggested by the user. The optimal condition determination method can determine a combination of culture medium conditions predicted to have an optimal result value by updating the result distribution of culture medium conditions by updating the results of experimental culture on the first stem cell.

[0102] Figure 7 is a flowchart of an optimal condition output method according to other embodiments of the present disclosure.

[0103] In S210, the decision device (100) can generate a combination of fourth culture medium conditions using combinations of the first to third culture medium conditions.

[0104] The decision device (100) can compare the result distributions for combinations of the first and third culture medium conditions input by the user, and can generate a new combination of third culture medium conditions by variously utilizing concentration values, injection times, chemical factors, etc. in combinations of the first and second culture medium conditions that satisfy specific conditions. Finally, a process of comparing the result distributions, which are the results of culture with combinations of the first to third culture medium conditions and the newly generated combination of the fourth culture medium conditions, can be performed again.

[0105] The decision device (100) can compare the result distributions for combinations of the first and third culture medium conditions input by the user using the optimal condition determination model, and can generate a new combination of fourth culture medium conditions by utilizing the concentration values ​​in the combination of the first and second culture medium conditions that satisfy a specific condition. Finally, a process of comparing the result distributions, which are the results of culture using the combinations of the first to third culture medium conditions and the newly generated fourth culture medium conditions, can be performed again. Through the above additional process, the optimal condition determination model can output an optimal combination of conditions in which a gene in a stem cell desired by the user is cultured at a predetermined gene expression level.

[0106] In S220, the decision device (100) can generate result distributions for combinations of the first to fourth culture medium conditions, including combinations of the generated fourth culture medium conditions.

[0107] In this way, the decision device (100) can output an optimal condition combination among combinations of culture medium conditions determined by itself by determining the concentration value for each growth factor, the injection time for each growth factor, the removal or insertion of chemical factors, etc. for inputs including growth factors.

[0108] In S230, the decision device (100) can output output data including the optimal condition combination output by the above process.

[0109] In addition, although the preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above, and various modifications can be made by a person having ordinary skill in the art to which the present invention pertains without departing from the gist of the present invention claimed in the claims, and such modifications should not be understood individually from the technical idea or prospect of the present invention.

[0110] Therefore, the spirit of the present invention should not be limited to the embodiments described above, and not only the scope of the patent claims described below but also all scopes equivalent to or equivalently modified from the scope of the patent claims are considered to fall within the scope of the spirit of the present invention. Hereinafter, the configuration and operation of the present invention will be described in detail with reference to embodiments of the present invention illustrated in the attached drawings.

[0111] The present invention is capable of various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. The effects and features of the present invention, as well as the methods for achieving them, will become clearer with reference to the embodiments described in detail below, along with the drawings. However, the present invention is not limited to the embodiments disclosed below and can be implemented in various forms.

[0112] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. When describing with reference to the drawings, identical or corresponding components are given the same drawing reference numerals, and redundant descriptions thereof will be omitted.

[0113] Hereinafter, the term “upper” or “upper” may include not only things that are directly above in contact, but also things that are above in a non-contact manner.

[0114] In the following examples, the terms first, second, etc. are not used in a limiting sense, but are used for the purpose of distinguishing one component from another.

[0115] In the examples below, singular expressions include plural expressions unless the context clearly indicates otherwise.

[0116] In the following examples, terms such as “include” or “have” mean that a feature or component described in the specification is present, and do not preclude the possibility that one or more other features or components may be added.

[0117] For convenience of explanation, the sizes of components in the drawings may be exaggerated or reduced. For example, the sizes and thicknesses of each component shown in the drawings are arbitrarily indicated for convenience of explanation, and thus the present invention is not necessarily limited to what is shown.

[0118] Additionally, terms such as “…part”, “…area”, etc. described in this specification may mean a unit that processes at least one function or operation.

[0119] Stem cell culture can be used in the treatment of diseases such as cancer. Cultured stem cells can be used to study characteristics such as tumor initiation, growth, and resistance. Cultured stem cells can be used to conduct high-throughput drug screening to study compounds that can target cancer cells. Cultured stem cells can be used to establish patient-derived cancer cell lines and conduct tests for drug sensitivity. Cultured stem cells can also be used to study intratumoral heterogeneity.

[0120] Stem cells, which are utilized in various ways as described above, can exhibit varying levels of gene expression in differentiated cells depending on changes in culture conditions. Because stem cells can differentiate into different cell types depending on growth factors, concentrations, and media, culture conditions for stem cells can include the composition of the culture medium, temperature, pH, oxygen concentration, growth factors, hormones, cytokines, media type, and the timing of factor administration.

[0121] In this specification, genomic data may include WES (Whole exome sequencing) and sequencing data, and may include data obtained by various analysis methods.

[0122] WES (Whole exome sequencing) sequences approximately 50 Mbp at approximately 100X, specifically obtaining only the exome region, and can obtain a relatively greater read depth than WGS (Whole genome sequencing). It is a method of analyzing only the genome of the exon region that directly codes for proteins. In the case of humans, exons account for less than 2% of the entire genome, so it is cheaper to acquire than WGS and the data capacity is also reduced. Even when analyzing WGS, variants in non-coding regions are difficult to interpret and may produce highly meaningful results. To solve this problem, analysis using WES is more economical and can obtain information at a higher depth.

[0123] RNA is an abbreviation for ribonucleic acid, and together they are called nucleic acids. Nucleic acids are polymers composed of multiple nucleotides. The nucleotides that make up DNA (deoxyribonucleic acid) are specifically called deoxyribonucleotides, and the nucleotides that make up RNA (ribonucleic acid) are specifically called ribonucleotides.

[0124] FIG. 1 is a block diagram of an optimal stem cell culture medium combination determination device (100, hereinafter, determination device) according to embodiments of the present disclosure.

[0125] The decision device (100) according to embodiments of the present disclosure can determine optimal culture medium conditions for stem cells. The decision device (100) can determine optimal culture medium conditions by considering cells to be differentiated and the gene expression levels in the cells. A user may want to culture cells with a reference gene expression level. Conventionally, a user could culture a large number of tumor cells to be used in research, and then conduct research with cells with a high gene expression level of the desired gene, or conduct research by adjusting the gene expression level to a desired range by inserting a gene from another cell using genetic scissors. The decision device (100) according to embodiments of the present disclosure can determine culture medium conditions for culturing cells in which the gene expression level of the gene to be used in research matches the reference gene expression level. At this time, the decision device (100) can output optimal culture medium conditions for stem cells considering the gene expression level by utilizing an optimal condition determination model learned to determine culture medium conditions.

[0126] The decision device (100) obtains, through various methods, result values ​​such as differentiation success rate, reproducibility, and gene expression level success rate of cells cultured with combinations of one or more culture medium conditions, generates result distributions of result values ​​for combinations of multiple culture medium conditions, and uses these result distributions to determine an optimal combination of culture medium conditions among the combinations of culture medium conditions. The decision device (100) can sample result values ​​for combinations of culture medium conditions to generate result distributions for the combinations. The decision device (100) can determine an optimal combination of culture medium conditions for a given stem cell through a process of updating the result distribution by updating the result values ​​for the combinations of culture medium conditions. The result value for the combination may be a value for cells cultured with the culture medium conditions, and may be a value calculated by combining differentiation success rate, gene expression level success rate, and the like. The result values ​​for the combination can be calculated by a preset mathematical formula for calculating the success rate of the division, the success rate of the gene expression level, etc. At this time, the result device (100) can update the result distribution by applying compensation for the Bayesian probability distribution to the result values ​​cultured with combinations of culture medium conditions.

[0127] The decision device (100) can obtain data on a combination of culture medium conditions for a given stem cell and data on a gene from a user's input. Here, the information on the gene can include a gene, a reference gene expression level, etc. The gene can be at least one of genes expressed in the stem cell. The reference gene expression level is a value for the expression level at which the corresponding gene should be expressed, and can include a specific value, above the average value, below the average value, twice the average value, half the average value, etc. The gene expression level input by the user can be referred to as the reference gene expression level. The user can obtain data on a combination of stem cells to be cultured and culture medium conditions that determine the reference gene expression level of a specific gene so that the specific gene is expressed in the stem cell at the desired reference gene expression level through the decision device (100). In an optional embodiment, the decision device (100) can generate combinations of a plurality of culture medium conditions by changing the concentration values ​​of each growth factor based on the growth factors, which are conditions of the culture medium input by the user. At this time, the result distributions for the combinations of the first and fourth culture medium conditions input by the user can be compared to generate a new combination of fifth culture medium conditions by utilizing the concentration values ​​in the combinations of the first and second culture medium conditions that satisfy a specific condition. Finally, a process of comparing the result distributions, which are the results of culturing with the combinations of the first to fourth culture medium conditions and the newly generated fifth culture medium conditions, can be performed again. As illustrated in FIG. 5b, when condition 1, condition 2, condition 3, and condition 4 are input, condition 5, which is a new combination of culture medium conditions, can be generated, and the gene expression success rate, which is the result of culturing with condition 5, can be included in the result distribution.The decision device (100) can generate a result graph by removing the result graph of condition 3, which has the worst result value among the combinations of five conditions, from the result distribution (see FIG. 5c).

[0128] Culture medium conditions may include growth factors, concentration values ​​of growth factors, information on supplements added during culture, timing (time point) of addition of each growth factor, medium, etc. Additionally, culture medium conditions may further include temperature, pH, oxygen concentration, growth factors, hormones, cytokines, biochemical factors, etc. Culture medium conditions may further include numerical values ​​representing the degree, size, concentration, etc. of each of the above factors.

[0129] The decision device (100) can generate result distributions by acquiring result values ​​for cells cultured with combinations of culture medium conditions for stem cells from culture experiments or from a database. The decision device (100) can determine an optimal combination of culture medium conditions for stem cells using the result distributions generated in this way. At this time, the result distributions for a plurality of combinations of culture medium conditions can be generated through a process of resampling using result values ​​of additional culture experiments that are performed, predicted, or produced with each combination of culture medium conditions. The decision device (100) can determine an optimal combination of culture medium conditions for stem cells using an optimal condition determination model that is learned with result values ​​of culture experiments performed with combinations of culture medium conditions. As mentioned above, the optimal condition determination model can be designed to output a new combination of culture medium conditions as an optimal combination of culture medium conditions in addition to the combinations of input culture medium conditions. Here, the combination of new culture medium conditions may be derived from various combinations, such as the average value of values ​​included in previously input combinations, the weighted average value of values, etc.

[0130] The decision device (100) can store information about the differentiation process for cells to be differentiated in advance. The decision device (100) can generate a combination of additional conditions by setting the injection points of growth factors to predetermined points based on the information about the differentiation process. Since each cell differentiates into mature cells through a predetermined process, information about the differentiation process is required to infer a combination of culture medium conditions that express the differentiation success rate, gene expression level, etc. of stem cells at a reference gene expression level.

[0131] The optimal condition decision model can be trained to output a result distribution map for the culture medium conditions by taking as input a combination of conditions including information about stem cells, growth factors, concentration values ​​for the growth factors, injection points, and the result values ​​of cells cultured under a combination of culture medium conditions. The optimal condition decision model can be constructed to sample the result distribution map for each combination of conditions and output the optimal combination of conditions. The optimal decision model can periodically update the input data to update the result distribution map for each combination of culture medium conditions, and output the optimal combination of conditions using the result distribution map.

[0132] At this time, the learned optimal condition decision model can output a combination of culture medium conditions that control the gene expression level of a specific gene by inferring factors of culture medium conditions that are highly correlated with cell differentiation success rate, reproducibility, and gene expression levels. The factors of culture medium conditions with high correlation may vary depending on the type of gene. The learned optimal condition decision model can learn factors of culture medium conditions that are correlated with each of one or more genes expressed in the cell. For example, the learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of MYC in cells such as Burkitt's lymphoma, breast cancer, lung cancer, and colon cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of HER2 in cells such as breast cancer and stomach cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of EGFR in cells such as non-small cell lung cancer, colon cancer, and head and neck cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of BRAF in cells such as malignant melanoma, colon cancer, and thyroid cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of Cyclin D1 in cells such as breast cancer, B-cell lymphoma, and esophageal cancer.

[0133] Information about stem cells may include the type of stem cell, the type of cell to be differentiated (insulin-producing cells, retinal cells, fat cells, cardiac muscle cells, skin cells, bone cells, etc.), information related to the stem cell target (age, sex, etc.), information about mutations, the types of genes expressed in stem cells, and information about the expression level of each gene (average value, expression level at the time of cancer onset).

[0134] Examples of growth factors in the culture medium conditions may further include VEGF (Vascular Endothelial Growth Factor), EGF (Epidermal Growth Factor), BMPs (Bone Morphogenetic Proteins), fatty acids, temperature, pH value, oxygen concentration, glucose concentration, illumination conditions, and concentration values. For example, the culture medium conditions for controlling the expression level of a certain gene for cultured cells may include BDNF (Brain-Derived Neurotrophic Factor) and may include the timing of injection of BDNF (Brain-Derived Neurotrophic Factor). The differentiation process of stem cells into vascular cells may include a cell aggregation stage, a mesoderm induction stage, a vascular lineage induction stage, a vascular network formation stage, and a cell maturation stage. In order to obtain a desired gene expression level for the cultured cells, the timing of injection of growth factors may be selected by each of the stages of the cell aggregation stage, the mesoderm differentiation stage, the vascular lineage differentiation stage, the vascular network formation stage, and the cell maturation stage, and the injection date (first day, second day, etc.) among the stages. Not limited to the above examples, various conditions can be changed with respect to the injection time in order to control the gene expression level of the gene.

[0135] As illustrated in FIG. 1, the decision device (100) according to some embodiments may include a processor (110), an input / output unit (130), a memory (140), a communication unit (150), and an optimal condition determination unit (120). However, not all of the components illustrated in FIG. 1 are essential components of the decision device (100). The decision device (100) may be implemented with more components than the components illustrated in FIG. 1, or with fewer components than the components illustrated in FIG. 1. The decision device (100) may be a user terminal, a server, a network system, or a separate device.

[0136] According to one embodiment of the present disclosure, the processor (110) typically controls the overall operation of the decision device (100). For example, the processor (110) may control the components included in the decision device (100) by executing a program stored in the decision device (100).

[0137] According to one embodiment of the present disclosure, the processor (110) can determine optimal culture medium conditions for stem cells. The processor (110) can determine optimal culture medium conditions by considering cells to be differentiated and the level of gene expression in the cells. The processor (110) can output optimal culture medium conditions for stem cells by utilizing an optimal condition determination model learned to determine a combination of culture medium conditions for culturing cells in which a gene to be utilized in research is expressed at a reference gene expression level.

[0138] The processor (110) can obtain, for combinations of culture medium conditions input by a user, result values ​​such as differentiation success rate, reproducibility, gene expression level, and success rate of gene expression level of cells cultured with combinations of culture medium conditions by various methods such as obtaining them from a database, obtaining them experimentally, or predicting them from experimental values. The processor (110) can generate a result distribution including result values ​​for combinations of multiple culture medium conditions and determine a combination of culture medium conditions in which a gene expression level is cultured at a reference gene expression level among the combinations of culture medium conditions. The processor (110) can generate a result distribution by sampling result values ​​for combinations of culture medium conditions. The processor (110) can update the result values ​​for combinations of culture medium conditions to update the result distribution, and determine an optimal combination of culture medium conditions for stem cells expressed at a reference gene expression level based on the result distributions. Stem cells cultured with a combination of optimal culture medium conditions can include genes expressed at a reference gene expression level. At this time, the processor (110) can update the result distribution using compensation for the Bayesian probability distribution.

[0139] The processor (110) can obtain data on a combination of culture medium conditions for a given stem cell and data on a gene from a user's input. Here, the data on the gene can include a gene, a reference gene expression level, etc. The gene can be at least one of genes expressed in the stem cell. The reference gene expression level is a value for the expression level at which the corresponding gene should be expressed, and can include a specific value, above the average value, below the average value, twice the average value, half the average value, etc. The user can obtain data on a combination of culture medium conditions that determine the stem cell to be cultured and the gene expression level of a specific gene so that the specific gene is expressed in the stem cell at the desired reference gene expression level through the processor (110). The processor (110) can additionally generate a plurality of combinations of culture medium conditions by changing the concentration value of each growth factor based on the growth factors, which are conditions of the culture medium input by the user. The combination of culture medium conditions to be generated can be generated based on several combinations determined based on the result values ​​for the input culture medium conditions. At this time, by using the result distribution including the result values ​​for the combinations of the first and fourth culture medium conditions input by the user, the culture medium conditions, i.e., concentration values, injection times, chemical factors, etc., in the combination of the first and second culture medium conditions that satisfy a specific condition, for example, the combination with the best result value, can be variously utilized to generate a new combination of fifth culture medium conditions. Finally, a process of comparing the result distributions, which are the results of culture with the combinations of the first to fourth culture medium conditions and the newly generated combination of the fifth culture medium conditions, can be performed again. The culture medium conditions can include growth factors, concentration values ​​of the growth factors, information on supplements added during culturing, the injection timing (time point) of each growth factor, the medium, etc.Additionally, culture medium conditions may further include temperature, pH, oxygen concentration, growth factors, hormones, cytokines, biochemical factors, etc. The culture medium conditions may further include numerical values ​​representing the degree, size, concentration, etc. of each of the above factors.

[0140] The processor (110) can acquire the result values ​​of the results of culturing with combinations of culture medium conditions of stem cells from a culture experiment or a database to generate a result distribution. The processor (110) can determine an optimal combination of culture medium conditions for stem cells using the generated result distributions. At this time, the result distributions for a plurality of combinations of culture medium conditions can be generated through a process of resampling using the result values ​​of additional culture experiments that are executed, predicted, or produced with each combination of culture medium conditions. The processor (110) can determine an optimal combination of culture medium conditions for stem cells using an optimal condition determination model that is learned with the result values ​​of culture experiments executed with combinations of culture medium conditions. As mentioned above, the optimal condition determination model can be designed to output a newly generated combination of culture medium conditions in addition to the input combinations of culture medium conditions as an optimal combination of culture medium conditions.

[0141] The processor (110) can store information about the differentiation process for cells in advance. The processor (110) can randomly change the injection times of growth factors based on the information about the differentiation process to generate combinations of set conditions. The combinations of conditions with changed injection times can also be determined based on the result values ​​for the combinations of input conditions. If the result values ​​for stem cells cultured at the first injection time point and the second injection time point correspond to the first and second ranks among all the result values, a combination that uses any time point among the first injection time point and the second injection time point as the injection time point can be additionally generated. The arbitrary time point can be determined by processing the first injection time point and the second injection time point by a method such as an average or a weighted average. Since each cell differentiates into mature cells through a set process, information about the differentiation process of the cells is required to infer culture medium conditions that increase the differentiation success rate and the gene expression success rate of stem cells.

[0142] The processor (110) is a component for controlling the overall operation of the decision device (100). Specifically, the processor (110) controls the overall operation of the decision device (100) using various programs stored in the storage medium (150) of the decision device (100). For example, the processor (110) may include a CPU, RAM, ROM, and a system bus. Here, the ROM is a component in which a command set for system booting is stored, and the CPU copies the stored operating system of the decision device (100) to the RAM according to the command stored in the ROM and executes the O / S to boot the system. When the system booting is complete, the CPU can copy various stored applications to the RAM and execute them to perform various operations. Although the service provision server (100) has been described as including only one CPU in the above, it may be implemented with multiple CPUs (or DSPs, SoCs, etc.) during implementation.

[0143] According to one embodiment of the present invention, the processor (110) may be implemented as a digital signal processor (DSP), a microprocessor, or a time controller (TCON) that processes digital signals. However, the present invention is not limited thereto, and may include one or more of a central processing unit (CPU), a micro controller unit (MCU), a micro processing unit (MPU), a controller, an application processor (AP), a communication processor (CP), or an ARM processor, or may be defined by the corresponding terms. In addition, the processor (110) may be implemented as a system on chip (SoC) or large scale integration (LSI) having a built-in processing algorithm, or may be implemented in the form of a field programmable gate array (FPGA).

[0144] According to one embodiment of the present disclosure, the input / output unit (130) can display an interface generated by the memory (140) of the decision device (100). According to one embodiment of the present invention, the input / output unit (130) can display a user interface for an input user input. The input / output unit (130) can output stored graphic data, visual data, auditory data, and vibration data under the control of the memory (140).

[0145] The input / output unit (130) may be implemented as a display panel of various forms. For example, the display panel may be implemented as a display technology of various forms, such as a liquid crystal display (LCD), an organic light emitting diode (OLED), an active-matrix organic light-emitting diode (AM-OLED), a liquid crystal on silicon (LCoS), or a digital light processing (DLP). In addition, the input / output unit (130) may be coupled to at least one of the front area, the side area, and the rear area of ​​the display panel in the form of a flexible display.

[0146] The input / output unit (130) may be implemented as a touch screen with a layer structure. The touch screen may have a function of detecting not only a display function but also a touch input location, a touched area, and even a touch input pressure, and may also have a function of detecting not only a real touch but also a proximity touch.

[0147] The input / output unit (130) may include a user interface for inputting various information into the decision device (100).

[0148] According to one embodiment of the present disclosure, the memory (140) may store a program for processing and controlling the processor (110) and / or the optimal condition determination unit (120), and may also store data input to or output from the determination device (100). According to one embodiment of the present disclosure, the memory (140) may store information on the differentiation process, information on genes, information on culture medium conditions of stem cells, information on growth factors, generated result distributions, etc. The memory (140) may include a database storing the above information.

[0149] According to one embodiment of the present disclosure, the memory (140) may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., SD or XD memory, etc.), a RAM (Random Access Memory), a SRAM (Static Random Access Memory), a ROM (Read-Only Memory), an EEPROM (Electrically Erasable Programmable Read-Only Memory), a PROM (Programmable Read-Only Memory), a magnetic memory, a magnetic disk, and an optical disk. In addition, according to one embodiment of the present disclosure, programs stored in the memory (140) may be classified into a plurality of modules according to their functions.

[0150] According to one embodiment of the present disclosure, the communication unit (150) can communicate with an external device of the processor (110). For example, the communication unit (150) can communicate with an external device, such as a payment server or an authentication server, under the control of the processor (110). In addition, the communication unit (150) can also obtain user information or user input through communication with an external interface.

[0151] The optimal condition determination unit (120) may include a data input unit (121), an optimal combination output unit (122), a result update unit (123), and an output data generation unit (124), as illustrated in FIG. 2.

[0152] The data input unit (121) can obtain combinations of multiple stem cell culture medium conditions, information on stem cells, information on genes, and information on reference gene expression levels from user input. The information on genes may include at least one of genes expressed in stem cells. The information on reference gene expression levels may include values ​​for expression levels at which the corresponding genes should be expressed, such as a specific value, above the average, below the average, twice the average, half the average, etc. The combination of culture medium conditions may include growth factors, concentration values, medium, injection timings of growth factors, etc. For example, the combination of culture medium conditions may include, but is not limited to, growth factors CHIR, BMP4, concentration values ​​of CHIR, concentration values ​​of BMP4, injection timings of CHIR, and injection timings of BMP4, as illustrated in FIG. 4, and may include various culture medium conditions.

[0153] The data input unit (121) can input data including combinations of multiple conditions and reference gene expression levels of genes. This data can include information about genes and information about the gene expression levels of the corresponding genes. The data input unit (121) can input data about combinations of conditions including growth factors. The data input unit (121) can input data about combinations of conditions including stem cell information and growth factors.

[0154] The optimal combination output unit (122) can input data including combinations of multiple stem cell culture medium conditions, gene expression levels of genes, etc. into the optimal condition determination model, and sample the result distributions for the gene expression levels of genes for the combinations. The optimal condition determination model can generate a large amount of data for the results of culturing with each combination of culture medium conditions. The optimal condition determination model can output the optimal condition combination based on the result data for each combination. The optimal condition determination model can compare the result distributions for the combinations of the first and fourth culture medium conditions input by the user, and can generate a new combination of fifth culture medium conditions by utilizing the culture medium conditions, for example, concentration values, in the combination of the first and second culture medium conditions that satisfy a specific condition. Finally, a process of comparing the result distributions, which are the results of culturing with the combinations of the first to fourth culture medium conditions and the newly generated combination of the fifth culture medium conditions, can be performed again. Through the above additional process, the optimal condition determination model can output an optimal combination of conditions in which the genes in the stem cells desired by the user are cultured at a set gene expression level.

[0155] The result update unit (123) can update the result distributions using the compensation data for the result values ​​for stem cells cultured under the optimal condition combination in the optimal decision model. In other words, the result update unit (123) can input the result distribution for the combination of the newly added fifth culture medium conditions into the optimal decision model and process it to update the optimal decision model. The result update unit (123) can update the result distributions for the result values ​​for stem cell culture through random sampling based on the Bayesian probability distribution. The result values ​​for stem cell culture can include the differentiation success rate of cultured stem cells (number of cells that successfully differentiated / total number of cells), the oncogenic mutation match rate with patient tissue, the reproducibility of stem cells (comparison of size, number, etc. with actual stem cells, correlation coefficient), tumor purity, gene expression level of a gene, etc. The result values ​​for stem cell culture can be at least one of the gene expression level value among the result values ​​for stem cells, the average value of the gene expression level values ​​for stem cells, and the combined value. Here, tumor purity can be determined using in-silico tumor purity estimation algorithms. More specifically, tumor purity can be calculated using DNA-seq-based methods such as ABSOLUTE, Sequenza, and FACETS, as well as RNA-seq (expression)-based methods such as ESTIMATE and PUREE.

[0156] Here, the cancer mutation match rate with patient tissue can be calculated based on mutant genes defined in a previously registered database. Examples of such databases include, but are not limited to, OncoKB and COSMIC DB.

[0157] The success rate and / or reproducibility of stem cell differentiation can be calculated based on the results obtained from the cultured stem cells. The success rate and / or reproducibility of stem cell differentiation can be calculated using established rules, mathematical formulas, etc.

[0158] The result update unit (123) can update the optimal condition determination model to update the result distributions based on the compensation data.

[0159] The optimal combination output unit (122) can output an optimal combination of conditions in which a specific gene is cultured to have a set gene expression level in stem cells input by a user using an optimal condition determination model. The optimal condition determination model can output result distributions generated based on the gene expression level for each combination of input conditions. The Y-axis of the result distribution can be determined by the size of the gene expression level or the degree of matching with the gene expression level requested by the user. The X-axis of the result distribution can be determined by a value for the combinations. The optimal condition determination model can compare these result distributions and output an optimal combination of conditions having a gene expression level set by the user. The optimal combination output unit (122) can generate result distributions for combinations of additional conditions in addition to the combinations of conditions input by the user. The optimal combination output unit (122) can generate a new combination of conditions based on the result distributions for the combinations of input conditions. The optimal combination output unit (122) can remove the result value for the combination with the lowest result value (based on the X-axis) from the result distribution. The optimal combination output unit (122) can repeat the process of adding result values ​​for new combinations and deleting the combination with the lowest result value. In the optimal condition determination model, a new combination of conditions can be generated based on the correlation for the result values ​​of gene expression levels. Machine learning can be used to learn how to process factors included in the input conditions.

[0160] The output data generation unit (124) can generate output data including the optimal condition combination from the optimal combination output unit (122).

[0161] The decision device (100) according to embodiments of the present disclosure can output an optimal condition combination from among combinations of multiple conditions devised by researchers and combinations of newly added conditions. When cultured using the optimal condition combination, stem cells input by the user can be cultured with a predetermined gene expression level. The decision device (100) can output a combination of culture medium conditions that include optimal values ​​for concentration values ​​of growth factors of conditions for stem cells and injection time points of growth factors (information on differentiation stage). Additionally, the decision device (100) can output a combination of culture medium conditions that include the injection time points of concentration values ​​of growth factors of conditions for stem cells as optimal time points. The decision device (100) can reapply the result data obtained by experimentally culturing with the combination of conditions to update the result distribution. The decision device (100) can output an optimal condition combination using the updated result distribution. The resulting distribution can be generated by sampling the resulting values ​​for combinations of conditions. The resulting distribution can be generated for each combination, or can include the resulting values ​​of the combinations. Methods for sampling the cultivated resulting values ​​include, but are not limited to, Thomson sampling, greedy methods, UCB (upper confidence bound), Bayesian optimization, Monte Carlo sampling, and bootstrapping. Various sampling methods can be used. The resulting distribution can be generated by beta-distributing the resulting values ​​(gene expression levels of genes) for combinations of conditions.

[0162] Figure 3 is a block diagram of a learning device (300) that learns an optimal condition decision model.

[0163] The learning device (300) can train an optimal condition determination model. The learning device (300) can create a Gaussian process model that defines the relationship between combinations of culture medium conditions and gene expression levels, and train a model that outputs an optimal condition combination using a predetermined acquisition function. At this time, the acquisition function may include a function such as the Upper Confidence Bound. The acquisition function can calculate the uncertainty of the prediction and output an output value based on the uncertainty. The learning device (300) can train a model that outputs an optimal condition combination using a Bayesian optimization method. The trained optimal condition determination model can generate result distributions for each combination of input conditions using various sampling methods, and compare the result distributions for the combinations to output an optimal condition combination. More specifically, the optimal condition determination model can generate result distributions using the Thomson sampling method. The trained optimal condition determination model may further include a module that generates a new combination of culture medium conditions using the combinations of input culture medium conditions. The module that generates these combinations can generate new concentration values ​​based on the concentration values ​​of growth factors, or can generate them using combinations of two culture medium conditions in order of best results. A new combination of culture medium conditions can be generated to include at least one of a new concentration value, a new injection time point, and a new growth factor. A new value can mean a value that is not included in the input data. The logic for generating a new combination of culture medium conditions can also be changed depending on the result distribution for the new combination of culture medium conditions. For example, if the result distribution for a new combination of culture medium conditions including the average value of the concentration values ​​has a result value greater than a predetermined reference value, a new combination of culture medium conditions can be generated by setting the average value of the concentration values.If the result distribution for a combination of new culture medium conditions including the midpoint of the injection time points has a result value greater than or equal to a predetermined reference value, a new combination of culture medium conditions can be created by setting the midpoint of the injection time points. If the result distribution for a combination of new culture medium conditions has a result value greater than or equal to a predetermined reference value by removing a specific growth factor from the combination of culture medium conditions having the lowest result value, a new combination of culture medium conditions can be created by removing the specific growth factor.

[0164] The result distribution may be a two-dimensional data representation of the result values ​​cultured with each combination. The result distribution may be generated as two-dimensional data with the x-axis representing the success rate for the gene expression level of the gene and the y-axis representing the number of occurrences (density), as illustrated in Fig. 5a. The result distribution may be expressed as a sampling of the result values ​​cultured with each combination. The result value for stem cells may include the expression success rate of cultured stem cells (the number of cells that successfully expressed / the total number of cells), etc. The expression success rate may be a value for the probability that the gene is expressed with the expression level desired by the user. The y-axis value of the result distribution may be at least one of the result values ​​for the stem cells, the average value, or the combined value of the result values ​​for the stem cells. Here, the tumor purity may be a value determined using in-silico-based tumor purity estimation algorithms. More specifically, tumor purity can be calculated using DNA-seq-based methods such as ABSOLUTE, Sequenza, and FACETS, and RNA-seq (expression)-based methods such as ESTIMATE and PUREE.

[0165] The dataset generation unit (310) can generate training data sets using combinations of conditions including information on stem cells, information on genes (genes, gene expression levels), growth factors and concentration values ​​for the growth factors, injection times of the growth factors, and culture result values. For example, training data sets can be generated using combinations of four combinations of conditions for stem cells of lung cancer cells, gene expression levels of genes, different concentration values ​​for growth factors for stem cells of breast cancer cells, injection times of growth factors, etc. The dataset generation unit (310) can generate training data sets using combinations of different growth factors according to the type of cells to be differentiated and the desired gene expression levels.

[0166] The model learning unit (320) uses training data sets as input and generates, for each combination, distributions of results sampled using a sampling method such as Thompson sampling, and can train an optimal condition determination model using the distributions of results for each combination generated in this way.

[0167] The optimal condition determination model can output an optimal condition combination by inputting combinations of conditions including information about stem cells, growth factors and their concentration values, and the timing of changes in the concentration values. The timing of change can include a value for the injection time.

[0168] The model learning unit (320) can learn an optimal condition determination model that generates a result distribution using data obtained by performing a culture experiment with a combination of input conditions for stem cells.

[0169] The data obtained through the culture experiment may be data stored in a registered database and / or data experimentally obtained by the research team.

[0170] The model learning unit (320) can learn a model that generates a result distribution by statistically sampling the result values ​​for the expression levels of genes in stem cells cultured under conditions including growth factors, concentration values ​​for growth factors, injection time points for growth factors, etc.

[0171] The model learning unit (320) can build an optimal condition determination model that compares the result distributions for each combination of conditions and outputs the optimal combination of conditions.

[0172] The model learning unit (320) can update the result distributions by changing the number of samplings performed for culture experiments based on a combination of conditions. For example, the number of samplings may increase as the training data set grows. Results from an increased number of samplings determine a combination of culture medium conditions for stem cells with a higher success rate of gene expression compared to results from a lower number of samplings. The success rate of gene expression may be a value calculated based on whether a given gene has a predetermined reference gene expression level.

[0173] As shown in Fig. 5a, the larger the x-axis value at the peak point of the result distribution and the larger the y-axis value, the more optimal the combination of culture medium conditions can be. In Fig. 5a, condition 1 is the optimal combination of conditions.

[0174] The model learning unit (320) can generate a combination of fifth culture medium conditions using combinations of the first to fourth culture medium conditions, and can generate result distributions for the combination of fifth culture medium conditions of FIG. 5b. According to the result distribution of FIG. 5b, considering the location of the peak, condition 1, which has the highest success rate and density, may be the optimal condition combination. The model learning unit (320) can generate a result distribution as in FIG. 5c, which compares the combinations of the first to fifth culture medium conditions and eliminates the combination of the third culture medium conditions with the worst results. For the result distribution of FIG. 5c, an optimal condition determination model that additionally performs a process of generating new culture medium conditions can be trained.

[0175] The model learning unit (320) can build an optimal condition determination model that outputs an optimal condition combination using the result distributions of Fig. 5c updated in this way.

[0176] The model learning unit (320) can periodically update the results of experimental cultivation using a combination of input conditions to update the result distribution.

[0177] The learning device (300) can generate an optimal condition determination model through the above learning process and transmit it to the optimal condition determination device (100) to output an optimal condition combination.

[0178] Figure 4 is an exemplary diagram of combinations of culture medium conditions for stem cells to be differentiated into vascular cells.

[0179] Figure 6 is a flowchart of an optimal condition output method according to embodiments of the present disclosure.

[0180] In S110, the decision device (100) can receive and acquire data including a combination of first culture medium conditions for first stem cells, a combination of second culture medium conditions, a combination of third culture medium conditions, and a reference gene expression level value of a first gene from a user. The combination of the first to third culture medium conditions may include one or more growth factors, concentration values ​​for each of the growth factors, injection time points of the growth factors, a medium, etc. Additionally, the culture medium conditions may further include temperature, pH, oxygen concentration, growth factors, hormones, cytokines, biochemical factors, etc. The culture medium conditions may further include numerical values ​​representing the degree, size, concentration, etc. of each of the above factors. The medium information may include information on one or more growth factors included in the medium. The information on the injection time points of the growth factors may include information on the injection time points in the differentiation stages.

[0181] In S120, the decision device (100) inputs combinations of first to third culture medium conditions and first stem cells into an optimal condition determination model, and can output a result distribution for combinations of the first to third culture medium conditions and one optimal condition combination among the combinations of the first to third culture medium conditions.

[0182] The optimal condition determination model can acquire result data from experimental culture of first stem cells using combinations of first to third culture medium conditions through various routes, and can sample the result data acquired in this manner to generate first to third result distributions. The optimal condition determination model can compare the first to third result distributions to determine a result distribution having a higher result value (success rate of gene expression level) and determine and output a combination of culture medium conditions of the result distribution as an optimal condition combination. The success rate of gene expression level can be a value that probabilistically calculates the degree of matching with a reference gene expression level. The gene expression level within a predetermined error range can be counted as the number of successful cells. The optimal condition determination model can update the first to third result distributions by updating the result data from experimental culture of first stem cells. The optimal condition determination model can learn factors of culture medium conditions that are correlated with each of one or more genes expressed in the cells. For example, the learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of MYC in cells such as Burkitt's lymphoma, breast cancer, lung cancer, and colon cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of HER2 in cells such as breast cancer and stomach cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of EGFR in cells such as non-small cell lung cancer, colon cancer, and head and neck cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of BRAF in cells such as malignant melanoma, colon cancer, and thyroid cancer. The learned optimal condition decision model can be trained to infer factors of culture medium conditions that are highly correlated with the gene expression level of Cyclin D1 in cells such as breast cancer, B-cell lymphoma, and esophageal cancer.

[0183] In S130, the decision device (100) can output output data that visually displays the optimal condition combination for the first stem cell. There may be two or more combinations of input culture medium conditions. If the reference gene expression level for the stem cell changes, the output optimal condition combination may also change.

[0184] Through this, the optimal condition determination method according to the embodiments of the present disclosure can determine a combination of culture medium conditions predicted to have an optimal result value (such as a success rate of gene expression level) among several combinations of culture medium conditions suggested by the user. The optimal condition determination method can determine a combination of culture medium conditions predicted to have an optimal result value by updating the result distribution of culture medium conditions by updating the results of experimental culture on the first stem cell.

[0185] Figure 7 is a flowchart of an optimal condition output method according to other embodiments of the present disclosure.

[0186] In S210, the decision device (100) can generate a combination of fourth culture medium conditions using combinations of the first to third culture medium conditions.

[0187] The decision device (100) can compare the result distributions for combinations of the first and third culture medium conditions input by the user, and can generate a new combination of third culture medium conditions by variously utilizing concentration values, injection times, chemical factors, etc. in combinations of the first and second culture medium conditions that satisfy specific conditions. Finally, a process of comparing the result distributions, which are the results of culture with combinations of the first to third culture medium conditions and the newly generated combination of the fourth culture medium conditions, can be performed again.

[0188] The decision device (100) can compare the result distributions for combinations of the first and third culture medium conditions input by the user using the optimal condition determination model, and can generate a new combination of fourth culture medium conditions by utilizing the concentration values ​​in the combination of the first and second culture medium conditions that satisfy a specific condition. Finally, a process of comparing the result distributions, which are the results of culture using the combinations of the first to third culture medium conditions and the newly generated fourth culture medium conditions, can be performed again. Through the above additional process, the optimal condition determination model can output an optimal combination of conditions in which a gene in a stem cell desired by the user is cultured at a predetermined gene expression level.

[0189] In S220, the decision device (100) can generate result distributions for combinations of the first to fourth culture medium conditions, including combinations of the generated fourth culture medium conditions.

[0190] In this way, the decision device (100) can output an optimal condition combination among combinations of culture medium conditions determined by itself by determining the concentration value for each growth factor, the injection time for each growth factor, the removal or insertion of chemical factors, etc. for inputs including growth factors.

[0191] In S230, the decision device (100) can output output data including the optimal condition combination output by the above process.

[0192] In addition, although the preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above, and various modifications can be made by a person having ordinary skill in the art to which the present invention pertains without departing from the gist of the present invention claimed in the claims, and such modifications should not be understood individually from the technical idea or prospect of the present invention.

[0193] Therefore, the spirit of the present invention should not be limited to the embodiments described above, and not only the scope of the patent claims described below, but also all scopes equivalent to or equivalently modified from the scope of the patent claims are considered to fall within the scope of the spirit of the present invention.

Claims

1. A step in which a decision device obtains data including a combination of first culture medium conditions including growth factors for first stem cells, concentration values ​​of the growth factors, a combination of second culture medium conditions, a combination of third culture medium conditions, and a reference gene expression level of a first gene from a user; The step of the above decision device inputting the combinations of the first to third culture medium conditions and information about the first stem cell into the optimal condition decision model, comparing the result distributions for the combinations of the first to third culture medium conditions, and outputting the optimal condition combination based on the gene expression level of the first gene; and A method for determining an optimal combination by calculating gene expression levels for combinations of culture media, comprising: a step of outputting output data visually outputting an optimal combination of conditions in which the first stem cell is cultured with the reference gene expression level of the first gene; 2. In paragraph 1, The above optimal condition decision model is, A method for determining an optimal combination by calculating gene expression levels for culture medium combinations, using a model learned using deep learning or machine learning, by inputting result distributions generated based on gene expression levels of the first gene in cells cultured under a plurality of culture medium conditions and combinations of a plurality of culture medium conditions for a plurality of stem cells.

3. In paragraph 2, The above optimal condition decision model is, A method for determining an optimal combination by calculating gene expression levels for combinations of culture media, generating a result distribution including the gene expression level of a first gene by sampling experimental results cultured under each combination of culture media conditions.

4. In paragraph 3, The above optimal condition decision model is, A method for determining the optimal combination by calculating the gene expression level for culture medium combinations, generating distribution maps of the results of sampling using the Thomson sampling method.

5. In paragraph 2, The above optimal condition decision model is, A method for determining an optimal combination by calculating the survival rate for combinations of culture media, wherein the method involves reinforcement learning through a process of relearning with result values ​​including the gene expression level value of the first gene obtained by increasing the number of sampling times for culturing the first stem cell under the first to third culture media conditions.

6. In paragraph 1, The above result distribution is, A method for determining the optimal combination by calculating the survival rate for combinations of culture media, which is a visual representation of the amount of gene expression obtained for a combination of culture media conditions.

7. Including memory, communication unit, and processor, The above processor, Obtain data including growth factors for first stem cells from a user, a combination of first culture medium conditions including concentration values ​​of growth factors, a combination of second culture medium conditions, a combination of third culture medium conditions, and a reference gene expression level of the first gene, By inputting the combination of the first to third culture medium conditions and information on the first stem cell into the optimal condition determination model, the result distributions for the combinations of the first to third culture medium conditions are compared, and the optimal condition combination is output based on the gene expression level of the first gene. An optimal stem cell culture medium combination determination device that executes commands to output output data that visually outputs an optimal combination of conditions for culturing the first stem cell by expressing the first gene at a reference gene expression level.

8. In paragraph 7, The above optimal condition decision model is, An optimal stem cell culture medium combination determination device, which is a model learned using deep learning or machine learning, using as input the result distributions generated based on the gene expression level values ​​of the first gene in cells cultured under the plurality of culture medium conditions and combinations of the plurality of culture medium conditions for a plurality of stem cells.

9. In paragraph 8, The above optimal condition decision model is, An optimal stem cell culture medium combination determination device that samples experimental results cultured under each combination of culture medium conditions and generates a result distribution including the gene expression level of the first gene.

10. In paragraph 9, The above optimal condition decision model is, A device for determining the optimal stem cell culture medium combination, which generates distribution maps of sampling results using the Thompson sampling method.

11. In paragraph 8, The above optimal condition decision model is, An optimal stem cell culture medium combination determination device, which is reinforced through a process of re-learning using result values ​​including the gene expression level value of the first gene obtained by increasing the number of sampling times for culturing the first stem cell under the first to third culture medium conditions.

12. In paragraph 7, The above result distribution is, A device for determining the optimal stem cell culture medium combination, which visually represents the gene expression level values ​​obtained for a combination of culture medium conditions.

13. A computer program stored in a computer-readable storage medium for executing any one of the methods of clauses 1 to 6 using a computer.

Images