Analysis device for carrying out tumor single cell analysis by utilizing artificial intelligence and use method of analysis device

Abstract

The single cell analysis apparatus for tumor analysis using artificial intelligence according to various embodiments of the present invention may include a memory, a communication unit, and a processor. The processor may be configured to: acquire single cell data on a tumor of a specific patient through the communication unit; preprocessing the single cell data of the tumor through a data preprocessing module; performing type classification based on the preprocessed data through a cell type definition module; predicting, by a single cell status prediction module, a phenotype inferrable from the RNA expression of the specific patient's tumor; materializing the heterogeneity of the specific patient tumor by a tumor analysis module; and generating a report according to the analysis result of the tumor of the specific patient through an analysis result generation module, and providing the report.

Description

Technical Field

[0001] The present invention relates to a device and method for analyzing the phenotype that can be inferred from the RNA expression of tumor cells in tumor samples collected from patients on a single-cell basis and providing result values, thereby achieving more accurate analysis. More specifically, the present invention relates to an analysis device and a method for using the same for single-cell tumor analysis using artificial intelligence. The present invention claims priority to Korean Patent Application No. 10-2023-0022552, filed on February 21, 2023, the entire contents of which are incorporated herein by reference. Background Art

[0002] Cancer is the most common disease among Koreans. While the rate of progression varies depending on the type of cancer, once it becomes malignant and metastasizes, treatment becomes extremely difficult.

[0003] In current clinical settings, tumor genotypes are classified based on the sum of all genes present in the cancer cells, immune cells, and other stromal cells that comprise the tumor. Furthermore, in the case of cancer cells, due to new mutations arising during division and changes in the environment, cancer cells with different phenotypes may coexist within the same tumor. This means that if cancer is classified into subtypes, grades, and other characteristics on a tumor-by-tumor basis, there are certain limitations in selecting appropriate treatments and drugs for cancer cells exhibiting different characteristics within the same tumor.

[0004] With the development of artificial intelligence technology, various attempts to apply it to the medical field have emerged. In particular, there is a need to utilize existing accumulated data to more accurately analyze tumor samples collected from cancer patients at the single cell level.

[0005] In this regard, there is Korean Patent Publication No. 10-2017-0072685. Summary of the Invention

[0006] Problems to be solved by the invention

[0007] The embodiments disclosed in the present invention are proposed to solve the above-mentioned problems. Their purpose is to divide the input data of tumor samples analyzed in single-cell units into more detailed subtype groups through machine learning, and to refine the classification accordingly to prescribe appropriate treatment methods and drugs based on the prognosis.

[0008] The technical problems to be solved by the present invention are not limited to the problems described above, and other problems not explicitly mentioned can be clearly understood by those skilled in the art through the following description.

[0009] Means used to solve problems

[0010] To achieve the above-mentioned technical problems, one aspect of the present invention provides a single-cell analysis device for performing tumor analysis using artificial intelligence, wherein the device includes a memory, a communication unit, and a processor, wherein the processor is configured to: obtain single-cell data about a specific patient's tumor through the communication unit; preprocess the single-cell data of the tumor through a data preprocessing module; perform type classification based on the preprocessed data through a cell type definition module; predict a phenotype that can be inferred from the RNA expression of the specific patient's tumor through a single-cell state prediction module; concretize the heterogeneity of the specific patient's tumor through a tumor analysis module; and generate a report of the analysis results of the specific patient's tumor through an analysis result generation module and provide it.

[0011] Furthermore, a single cell analysis method of the present invention can be provided.

[0012] In addition, a system including a server and a device for executing the method of the present invention may also be provided.

[0013] Effects of the Invention

[0014] According to the problem-solving means described in the present invention, it can be confirmed that a single patient with cancer may simultaneously have multiple cancer phenotypes that can be subdivided into different subtypes, and the patient can be predicted and classified into a completely new group that did not exist previously, or predicted and classified into an existing group through machine learning, thereby providing analysis result data containing information such as treatment methods and drugs corresponding to the group.

[0015] Furthermore, according to various embodiments of the present invention, by identifying and diagnosing tumors on a single-cell basis, the grade can be determined more accurately, which not only creates opportunities for multiple treatment options and multiple drugs, but also enables learning to predict metastasis information, providing customized information for individual patients.

[0016] The effects of the present invention are not limited to the effects described above, and other effects not mentioned can be clearly understood by those skilled in the art through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic block diagram of a single cell analysis system according to various embodiments of the present invention.

[0018] Figure 2 Schematic diagram of a single cell analysis device according to various embodiments of the present invention.

[0019] Figure 31 is a schematic flowchart of a method for performing single-cell tumor analysis using artificial intelligence according to various embodiments of the present invention.

[0020] Figure 4 is a schematic block diagram of a process for obtaining single-cell data of a specific patient's tumor according to various embodiments of the present invention.

[0021] Figure 5 4 is a flow chart of a data preprocessing process of a single-cell analysis device according to various embodiments of the present invention.

[0022] Figure 6 FIG. 4 is a schematic block diagram of a cell type definition process performed by a single-cell analysis apparatus according to various embodiments of the present invention.

[0023] Figure 7A and Figure 7B 1 is a flowchart and schematic diagram of the operation of a single cell state prediction module of a single cell analysis device according to various embodiments of the present invention.

[0024] Figure 8 Flowchart of a process for embodying heterogeneity through a tumor analysis module in a single-cell analysis device according to various embodiments of the present invention.

[0025] Figure 9 Schematic diagram of the process of embodying heterogeneity through a tumor analysis module in a single-cell analysis device according to various embodiments of the present invention. DETAILED DESCRIPTION

[0026] Throughout the specification of the present invention, the same reference numerals represent the same constituent elements. The present invention does not describe all the elements of each embodiment one by one, and omits the conventional content in the art or the content repeated between the embodiments. The terms "portion", "module", "component", "block" and the like used in this specification can be implemented by software or hardware, and, depending on the embodiment, multiple "portions", "modules", "components", "blocks" can be implemented as one constituent element, or one "portion", "module", "component", "block" can also include multiple constituent elements.

[0027] In this specification, when a certain part is described as being “connected” to another part, it includes not only a case of direct connection but also a case of indirect connection including a connection through a wireless communication network.

[0028] Furthermore, when a part is described as “comprising” a certain constituent element, unless otherwise explicitly described to the contrary, the description does not exclude other constituent elements but may further include other constituent elements.

[0029] In this specification, when a component is described as being “on” another component, it includes not only a case where the component is in direct contact with the other component but also a case where other components are present therebetween.

[0030] The terms “first”, “second”, etc. are only used to distinguish different components, and do not mean that the components are limited by these terms.

[0031] Unless clearly defined otherwise by the context, terms expressed in the singular form in the specification may also include the plural form.

[0032] The identifiers used in each step are only for convenience of description and are not used to limit the order in which the steps are to be performed. Unless a specific order is clearly defined from the context, the steps may be performed in a different order than described.

[0033] The working principle and embodiments of the present invention will be described below with reference to the accompanying drawings.

[0034] In this specification, "devices according to the present invention" include various devices that can perform computations and provide results to a user. For example, devices according to the present invention may include computers, server devices, and portable terminals, or any of these.

[0035] Here, the computer may include, for example, a notebook computer equipped with a web browser, a desktop computer, a laptop computer, a tablet computer, a touch-pad computer, etc.

[0036] The server device is a server that communicates with external devices and processes information, and may include an application server, a computing server, a database server, a file server, a game server, a mail server, a proxy server, and a web server.

[0037] The portable terminal may be, for example, a wireless communication device with portability and mobility, and may include: PCS (Personal Communication System), GSM (Global System for Mobile Communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT-2000 (International Mobile Telecommunication-2000), CDMA-2000 (Code Division Multiple Access-2000), W-CDMA (Wideband Code Division Multiple Access), WiBro (Wireless Broadband Internet) terminals, smart phones, and other handheld wireless communication devices, as well as wearable devices such as watches, rings, bracelets, anklets, necklaces, glasses, contact lenses, or head-mounted devices (HMD).

[0038] According to various embodiments of the present invention, a single-cell analysis device can be a server that utilizes artificial intelligence to perform a single-cell analysis process for tumors. Specifically, the single-cell analysis device can be a platform operating server that provides single-cell analysis process services. For example, the single-cell analysis device can include a device that uses artificial intelligence to analyze single cells from multiple cells in a tumor sample obtained from, for example, a hospital doctor.

[0039] The single-cell analysis device according to various embodiments of the present invention may include a high-performance CPU and / or GPU as a component. The high-performance CPU and GPU are not limited to the specifications that the various processes executed in the present invention can perform calculations without latency. In addition, the single-cell analysis device according to the present invention can also be a master control server that operates and manages a platform that provides single-cell analysis process services. Therefore, the single-cell analysis device of the present invention can learn prototypes (e.g., reference samples) of tumors (especially tumors containing cancer cells) through various machine learning models, and can classify and analyze tumor single-cell data of specific patients based on the prototype. The machine learning can update the model through repeated learning, and the analysis device can further perform supervised learning and / or unsupervised learning through artificial intelligence to explore new subtypes and grades related to cancer cells.

[0040] Figure 1 is a schematic block diagram of a single cell analysis system according to various embodiments of the present invention.

[0041] Reference Figure 1 The single cell analysis system 10 includes a single cell analysis device 100 and an external device 200. Each node can send and receive data with other different nodes, and each node can be connected through a network.

[0042] According to various embodiments of the present invention, single-cell analysis device 100 is a device capable of utilizing artificial intelligence to provide single-cell analysis results for tumors. Single-cell analysis device 100 can transmit or receive analysis result data to or from other connected devices via wired and / or wireless connections. Single-cell analysis device 100 can receive various data within single-cell analysis system 10 and provide the data via an application or web service.

[0043] According to various embodiments of the present invention, external device 200 may be a device that transmits single-cell tumor cell data related to a specific patient to single-cell analysis device 100. For example, external device 200 may be a terminal, server, or desktop computer that uploads single-cell RNA sequencing data for each of a plurality of cells constituting a tumor sample from a specific patient to single-cell analysis device 100.

[0044] According to another embodiment, the external device 200 may be a device that receives analysis result data from the single-cell analysis apparatus 100. For example, the external device 200 may be a research institute server, desktop computer, or terminal that receives single-cell analysis result data related to tumors obtained by the single-cell analysis apparatus 100. In other words, the external device 200 generally refers to a device that is connected to the single-cell analysis apparatus 100 via a network and is capable of transmitting and receiving various data.

[0045] Figure 2 Schematic diagram of a single cell analysis device according to various embodiments of the present invention.

[0046] refer to Figure 2 The single-cell analysis device 100 may include, but is not limited to, a processor 110, a communication unit 120, and a memory 130 as internal components. Each node can transmit and receive data with different nodes, and each node can be directly or indirectly connected via a network via wired and / or wireless connections. The single-cell analysis device 100 of the present invention may utilize an independent server to perform the functions of the processor 110, replacing the processor 110.

[0047] The single-cell analysis device 100 according to various embodiments of the present invention analyzes tumor data collected from patients on a single-cell basis and provides analysis results. The single-cell analysis device 100 can be a server that operates and manages a service platform for providing analysis results to hospitals (or physicians) providing cancer-related clinical examination and analysis services, as well as pharmaceutical companies providing clinical trial support services.

[0048] refer to Figure 2 The processor 110 includes a data preprocessing module 111, a cell type definition module 112, a single-cell state prediction module 113, a tumor analysis module 114, and an analysis result generation module 115. Based on the single-cell data about a specific patient's tumor acquired via the communication unit 120, the processor 110 classifies the tumor-related single-cell phenotype of each patient's cancer and provides result data predicting its grade. The processor 110 may include data about new subtypes in the result data, and may also include data in the form of a report to facilitate explanation to patients by clients such as hospitals, or for patients themselves to easily understand the analysis results.

[0049] The processor 110 may include various modules (111 to 115). Here, each module (111 to 115) may represent a functional block corresponding to a function performed by the processor 110. Specifically, each module (111 to 115) may correspond to a functional block named according to the function of the processor 110.

[0050] The processor 110 according to an embodiment of the present invention can be implemented as at least one functional block that performs the aforementioned operations using data stored in the memory 130. The memory 130 is used to store data of an algorithm for controlling the operation of each component in the single-cell analysis apparatus 100 or a program for reproducing the algorithm. In this case, the processor 110 and the memory 130 can be implemented as separate chips, or they can be implemented as a single chip.

[0051] The processor 110 can implement the following in the single cell analysis device 100: Figures 3 to 7A and Figure 8 The various embodiments of the present invention are described in detail, and any one or more combinations of the multiple components described above are controlled.

[0052] According to an embodiment of the present invention, the communication unit 120 may include a device for communicating with an external device (eg Figure 1 One or more components for communicating with the external device 200 shown in the figure may include, for example, at least one of the following: a broadcast receiving module, a wired communication module, a wireless communication module, a short-range communication module, and a location information module.

[0053] The wired communication module may include a variety of wired communication modules, such as a local area network (LAN) module, a wide area network (WAN) module, or a value-added network (VAN) module, and may also include a variety of wired interface communication modules, such as a universal serial bus (USB), a high-definition multimedia interface (HDMI), a digital visual interface (DVI), a recommended standard 232 (RS-232), a power line communication module, or a plain old telephone service (POTS) module.

[0054] The wireless communication module may include a Wi-Fi module, a wireless broadband (Wireless Broadband, WiBro) module, and may also include a communication module that supports multiple wireless communication modes, such as Global System for Mobile Communications (GSM), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Universal Mobile Telecommunications System (UMTS), Time Division Multiple Access (TDMA), Long Term Evolution Technology (LTE), fourth generation (4G), fifth generation (5G), and sixth generation (6G) communication modules.

[0055] The short-range communication module is used for short-range communication and can support short-range communication through at least one of the following technologies: Bluetooth TM), Radio Frequency Identification (RFID), Infrared communication (Infrared Data Association, IrDA), Ultra Wideband (UWB), ZigBee, Near Field Communication (NFC), Wireless Fidelity (Wi-Fi), Wi-Fi Direct, and Wireless USB, Wireless Universal Serial Bus (Wireless USB) technology.

[0056] According to an embodiment of the present invention, an output unit (not shown) is used to generate output related to vision, hearing, or touch, and may include a display unit (not shown). The display unit may form an interlayer structure or an integrated structure with the touch sensor to realize a touch screen. The touch screen can not only serve as a user input unit providing an input interface between the single-cell analysis device 100 and the user, but also provide an output interface between the single-cell analysis device 100 and the user.

[0057] The display unit is used to display (output) information processed by the single-cell analysis apparatus 100. For example, the display unit can display execution screen information of an application (e.g., an app) running on the single-cell analysis apparatus 100, or display user interface (UI) or graphical user interface (GUI) information corresponding to the execution screen information.

[0058] According to an embodiment, the memory 130 may be used to store data supporting various functions of the single-cell analysis apparatus 100, programs for operating the processor 110, input / output data (e.g., images, videos, etc.), and may also store multiple application programs (or applications) running in the single-cell analysis apparatus 100, as well as data and instructions for operating the single-cell analysis apparatus 100. At least a portion of the application programs may be downloaded from an external server via wireless communication.

[0059] Furthermore, the memory 130 may include at least one type of storage medium, such as a flash memory type, a hard disk type, a solid state drive type (SSD), a silicon disk drive type (SDD), a multimedia card micro type, a card type memory (such as SD or XD memory), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. Furthermore, the memory may be a database separate from the single-cell analysis apparatus 100 and connected via a wired or wireless connection.

[0060] Figure 2 The components shown may be added or deleted at least one component according to their performance. In addition, the positional relationship between the components may also be changed according to the performance or structure of the device, which is readily understood by those skilled in the art.

[0061] on the other hand, Figure 2 Each component shown is a portion that can be implemented by software and / or hardware components such as a field programmable gate array (FPGA) and an application specific integrated circuit (ASIC).

[0062] Figure 3 1 is a schematic flowchart of a method for performing single-cell tumor analysis using artificial intelligence according to various embodiments of the present invention.

[0063] In step S310, the processor (eg, Figure 2 The processor 110 in the embodiment may be connected to the communication unit (eg, Figure 2The communication unit 120 in the processor obtains single-cell tumor data for a specific patient. The single-cell tumor data for a specific patient may include single-cell RNA sequencing data of multiple cells present in the tumor extracted from the specific patient. The processor distributes and stores the single-cell tumor data and analysis results of the single-cell data obtained by the communication unit to the specific patient.

[0064] In step S320, the processor may use a single cell data pre-processing module (e.g., Figure 2 The single cell data preprocessing module 111 in the single cell data preprocesses the single cell data of the tumor. The single cell data preprocessing process can be roughly divided into two categories. For example, the single cell data preprocessing process may include a quality control process and an error correction process. According to the embodiment, if necessary, the single cell data preprocessing module may also perform other processes in addition to the above processes to preprocess the data. The detailed description of the preprocessing process will be given in Figure 5 Further described in .

[0065] In step S330, the processor may define the cell type through a cell type definition module (e.g., Figure 2 The cell type definition module 112 in the present invention performs type classification based on the preprocessed data. The cell type definition module can define the type of single cells in a specific patient's tumor by using existing single cell phenotypes and newly learned phenotypes, and classify them into grouped types for classification.

[0066] According to an embodiment, the phenotype that can be inferred from the tumor RNA expression of the specific patient may include but is not limited to at least one of the following: subtype, grade, aggressiveness, enriched pathways, biomarkers, copy number variation (CNV) and single nucleotide variant (SNV).

[0067] In step S340, the processor (eg Figure 2 The single-cell state prediction module 113 in the present invention can predict the tumor subtype and grade for a specific patient using the single-cell state prediction module. This step can correspond to a process of more accurately predicting the type of cancer suffering from the specific patient by analyzing the results of decomposing the multiple cells that make up the tumor into single-cell units.

[0068] In step S350, the processor (eg Figure 2 The tumor analysis module 114 in the example can be used to specify the tumor heterogeneity for a specific patient. Figure 1 The single-cell analysis device 100 in the present invention is a process of specifying the heterogeneity of the entire cancer tumor sample based on the single-cell subtype and its grade predicted by the machine learning model.

[0069] In step S360, the processor (eg Figure 2 The analysis result generation module 115 in the analysis result generation module can generate and provide a report of the tumor analysis results for a specific patient. For example, the analysis result can be data, and the analysis result data can be provided as a downloadable report format file. In addition, the analysis result data can include data regarding the tumor sample type, treatment methods, and other aspects of the specific patient's cancer. According to one example, the analysis result generation module can include and provide in the report content at least one of the current status results corresponding to the specific patient's tumor heterogeneity, metastasis prediction results, and treatment drug results.

[0070] According to an embodiment, the analysis result generation module can decide what information to include in the report content by self-learning or pre-setting. The data generated by the analysis result generation module contains data that is meaningful to the customer. For example, the analysis result data may include information about the subtype of a single cell, the proportion of the subtype in the entire tumor, the grade of each single cell, the proportion of the grade in the entire tumor, and heterogeneity information. As another example, the analysis result data may also include: personal information corresponding to a specific patient, tissue sample information of the specific patient being analyzed, overall information of the analyzed cancer, and information about each cell population analyzed in a specific patient. The analysis result data is not limited to the above examples, and the information required to be included can be set by the user using the single-cell analysis device, or can be adaptively set according to the cancer condition of a specific patient.

[0071] Figure 4 A schematic block diagram of a process for obtaining single-cell data of a specific patient's tumor according to various embodiments of the present invention is provided.

[0072] Reference Figure 4 , processors (for example: Figure 2 The processor 110 in the embodiment may be configured to communicate with the communication unit (e.g., Figure 2 The communication unit 120 in the embodiment of the present invention obtains single cell data of tumors of a specific patient. Figure 4The input data 410 in corresponds to single cell data.

[0073] According to an embodiment, single-cell tumor data for a specific patient may include RNA sequencing data for each of a plurality of cells present in tumor tissue extracted from the specific patient. The processor assigns and stores the single-cell tumor data and analysis results of the single-cell data obtained by the communication unit to the specific patient.

[0074] According to an embodiment, the single-cell tumor data about a specific patient may include omics data. Omics refers to a research field based on a holistic data set containing a single gene, transcript, protein, metabolite, etc. The data involved in omics are large-scale, large-capacity data and can be collectively referred to as omics data. For example, a genome refers to the entire genome sequence and the genetic information contained therein; a transcriptome refers to the sum of all transcripts expressed in a specific cell; an epigenome refers to the overall data of various modifications of DNA and histones that regulate gene transcription through epigenetic modification without changing the DNA sequence; a proteome refers to the sum of all proteins expressed in a specific cell or tissue; an interactome refers to the overall interaction between proteins observed directly or indirectly in a specific cell or tissue. In the present invention, the single-cell tumor data about a specific patient may include the omics data listed in the above examples, but are not limited thereto.

[0075] According to an embodiment, single-cell tumor data about a specific patient may refer to data obtained by performing single-cell RNA sequencing on multiple cells contained in a tumor sample collected from a specific patient by a hospital or pharmaceutical company, etc. after a biopsy is performed on the tumor sample.

[0076] According to the embodiments, there are differences between single-cell RNA sequencing and bulk sequencing. For example, when performing bulk sequencing, the heterogeneity of the tumor may not be taken into account. That is, according to the results of bulk sequencing, regardless of whether the tumor is heterogeneous, the RNA of the immune cells, stromal cells and cancer cells with the same or different genetic and epigenetic characteristics that constitute the tumor may be mixed and analyzed. The RNA sequencing used in the present invention can analyze RNA separately for each cell of the tumor, thereby taking into account the heterogeneity of the tumor. In other words, taking 100 cells as an example, by examining the transcriptome expression level of each single cell and sequencing the heterogeneity expressed in each single cell separately, single-cell RNA sequencing can be achieved. Such single-cell RNA sequencing can be implemented through a variety of processes. For example, it can be analyzed by amplifying cDNA and then using next-generation sequencing (Illumina NGS). The reception of single-cell RNA sequencing data in the present invention is not limited to data input through a specific process, as long as it is RNA sequencing data about single cells analyzed by an external agency.

[0077] Figure 5 This is a flow chart of a data pre-processing process of a single-cell analysis device according to various embodiments of the present invention.

[0078] In step S510, a data pre-processing module (e.g. Figure 2 The data preprocessing module 111 in the method can filter the single cell data based on multiple parameters. This filtering process can correspond to the quality control process.

[0079] According to an embodiment, the data preprocessing module can use multiple parameters to filter out meaningful data from the single cell data. For example, the parameter can be a parameter based on the amount of data. The parameter related to the amount of data can be used to filter out single cells that contain too little or too much information. In another example, the parameter can be a parameter based on the quality of the data. The parameter related to the quality of the data can be used to filter out single cells that are dead or dying. For another example, the parameter can be a parameter related to data repetition. The parameter related to data repetition can be used to eliminate data that may be read repeatedly. In this way, the data preprocessing module can filter out single cell data that may cause bias and have no practical significance.

[0080] In step S520, the data preprocessing module may perform an error correction process. This process may correspond to a batch error correction process. In other words, the data preprocessing module may compare the data with multiple other data sets based on factors that may cause data discrepancies, such as the environment in which the specific patient sample is located, the sequencing equipment, the library, etc., to correct errors.

[0081] According to one embodiment, the data pre-processing module can perform a calibration process, which is to compare the acquired tumor single cell data of a specific patient with the data stored in the memory (e.g., Figure 2 According to an embodiment, the data preprocessing module can be combined with the cell type definition module (e.g., Figure 2 The cell type definition module 112 in the figure performs the processing independently or collaboratively, and can be designed to assist in sequentially executing the functions required for the cell type definition process after pre-processing the single cell data.

[0082] Figure 6 FIG. 4 is a schematic block diagram of a cell type definition process performed by a single-cell analysis apparatus according to various embodiments of the present invention.

[0083] In step S610, the processor (eg, Figure 2 The processor 110 of FIG. 1 may define the cell type by a cell type definition module (e.g., Figure 2 The cell type definition module 112) distinguishes normal cells from cancer cells.

[0084] According to embodiments of the present invention, cell type differentiation is performed to classify cells into two categories: normal cells (e.g., epithelial cells, immune cells, etc.) and cancer cells (e.g., cancer epithelial cells, stromal cells, etc.), which can only be collected from cancer patients. Here, the genomic phenotype of a single cell may exhibit different characteristics depending on the cell type to which it belongs.

[0085] In step S620, the cell type definition module performs a process of grouping cancer cells according to phenotype. At this time, the cell type definition module can group cancer cells according to phenotype based on the expression level of genes.

[0086] According to an embodiment, the cell type definition module defines cell types by clustering cells with similar phenotypes. Subsequently, the cell type definition module determines the cell type of a cell group or individual cell by determining the expression level of genes (e.g., EPCAM (epithelial cell adhesion molecule)) in each cell.

[0087] In step S630, the cell type definition module performs a process of defining the cell types in the pre-processed data. The cell type definition module may map the grouped data to the cancer cell type benchmark data, thereby defining the cell types in the pre-processed data.

[0088] According to an embodiment, the cell type definition module can infer the cell type by mapping the cell type data in the data defined as the standard to the acquired single cell data. The reason for defining the cell type by the cell type definition module may be that different cells have diverse functions and phenotypes. For example, it may not make sense to mix the cancer cells of a specific patient with the immune cells of other patients for analysis. Therefore, the single cell analysis device of the present invention (e.g. Figure 1 The single-cell analysis device 100 shown performs analysis on a single-cell basis by clearly defining the cell type.

[0089] According to an embodiment, a single-cell analysis device can distinguish between normal cells and cancer cells contained in a tumor and group the cancer cells, thereby obtaining more accurate and meaningful information. This process is based on the principle that separating cancer cells from normal cells and analyzing them separately can achieve higher accuracy. If the detailed phenotypes in the cancer cells are further analyzed, more accurate and meaningful information can be derived. In addition, the single-cell analysis device can also compensate for errors that may occur during the collection of tumor samples during the analysis process. For example, when performing a tissue examination to obtain a tumor sample, it may not be possible to accurately target cancer cells for collection. In this case, the single-cell analysis device of the present invention can perform a process to clearly distinguish between normal cells and cancer cells.

[0090] Figure 7A and Figure 7B 1 is a flowchart and schematic diagram of the operation of a single cell state prediction module of a single cell analysis device according to various embodiments of the present invention.

[0091] Reference Figure 7A In step S710, the processor (eg Figure 2 The processor 110 in the embodiment of the present invention can be used to predict the state of a single cell by a single cell state prediction module (e.g. Figure 2 The single cell state prediction module 113 in the machine learning model is generated. However, the process of generating a model by the single cell state prediction module can also be replaced by obtaining an already generated model. Figure 7B The schematic diagram is described in detail later.

[0092] In step S720, the single cell state prediction module may execute a process for predicting the subtype and grade of the tumor. The single cell state prediction module may sequentially execute a subtype prediction process and a grade prediction process for a single cell based on the cell.

[0093] Regarding the subtypes and grading of cancer cells, breast cancer is used as an example for explanation. Breast cancer has four unique subtypes based on gene expression, including Luminal A, Luminal B, HER2, and TNBC. Depending on the subtype, the prognosis caused by its gene expression may also be different. For example, Luminal A has the best prognosis, as shown by a higher expression rate of estrogen and progesterone receptors, and in some cases chemotherapy can even be omitted. For example, Luminal B has a relatively poor prognosis compared to Luminal A, with a lower expression ratio of estrogen and progesterone receptors, so chemotherapy is usually required. For another example, although HER2 is relatively rare, it is usually associated with a poor prognosis, so targeted therapy is required in most cases. The single-cell state prediction module of the present invention can target these subtypes, helping to achieve more accurate diagnosis and, combined with the corresponding prognostic information, assisting in the formulation of an appropriate treatment plan.

[0094] Regarding the grading of cancer cells, in actual clinical practice, doctors usually use it as an important basis for judging the prognosis of breast cancer. The grading can be divided into three categories: well differentiated (grade 1), moderately differentiated (grade 2), and poorly differentiated (grade 3). In this case, the larger the number, the worse the prognosis. When graded according to G1, G2, and G3, the prognostic index of G1 is significantly better than that of G2, so a corresponding treatment plan can be formulated according to different grades. The single cell state prediction module of the present invention can be analyzed for these grades, thereby helping to determine a more appropriate treatment method.

[0095] refer to Figure 7B The single-cell state prediction module can be built using random forest classification. This is a supervised learning method based on labeled cancer cell subtypes and grades. It constructs multiple decision trees and uses the most frequently occurring result in each tree as the basis for determining the cancer subtype.

[0096] According to an embodiment, the single cell state prediction module can use the genes of a single cell as features, compare them with a preset threshold based on the labeled tumor pathology examination result data, and make predictions using a repeatedly trained model. Figure 7B, the single-cell state prediction module can obtain available information by analyzing the covariance between each feature and the target label during the operation of the decision tree. At this time, the single-cell state prediction module performs subgrouping according to the preset threshold corresponding to each feature. The module will repeat the above process until the subgroup in the decision tree corresponds to a unique target label. In this case, processing mechanisms such as overfitting can also be added, and the value corresponding to the label with the highest frequency of occurrence in the data set will be used as the subtype and grade of the single cell.

[0097] Figure 8 FIG. 4 is a flow chart illustrating a method for embodying heterogeneity through a tumor analysis module of a single-cell analysis device according to various embodiments of the present invention.

[0098] Reference Figure 8 In step S810, the processor (eg Figure 2 The processor 110 in the embodiment of the present invention can be used by the tumor analysis module (e.g. Figure 2 The tumor analysis module 114 in the embodiment confirms the grouping based on the genotype. Figure 1 The single cell analysis device 100 in the embodiment can superimpose the single cell data of each cancer cell in the tumor of multiple patients and cluster them, thereby forming cell groups with similar genotypes. Among these formed groups, meaningful groups are selected and, if they are clearly distinguished from existing genotypes, they can be defined as new subtypes for classification. This can correspond to forming genotype groups by performing unsupervised clustering on single cell data. According to an embodiment, the tumor analysis module can be as follows: Figure 7A The workflow shown here confirms grouping based on the phenotypes inferred from RNA expression in tumor cells.

[0099] According to an embodiment, in step S820, the tumor analysis module may derive a new subtype based on the generated group that is distinguishable from existing subtypes. The term "subtype" herein is merely an example of a new phenotype. In step S830, the tumor analysis module may compare and analyze the tumor phenotype of a specific patient with the new phenotype. In step S840, the tumor analysis module may determine the group to which each single cell belongs based on the tumor phenotype of the specific patient. This means that when a new phenotype emerges, the tumor phenotype of the specific patient is compared and analyzed with the new phenotype. If a new phenotype is derived, the single-cell genetic phenotype of the specific patient's tumor has been classified as a new phenotype. In other words, if a different prognosis is observed in a specific patient due to the new phenotype, and this difference is predicted, the analysis device may derive and classify the new phenotype through single-cell analysis, thereby enabling more appropriate treatment.

[0100] According to an embodiment, if the tumor analysis module determines that a single tumor cell from a specific patient corresponds to a new subtype, the prediction model result value for that cell can be output to obtain relevant information. This approach breaks through the paradigm of traditional cancer subtype classification by including not only existing subtypes but also newly classified subtypes, thereby including a processing flow that facilitates the processing of these new subtypes.

[0101] In step S850 , the tumor analysis module may analyze at least one or more phenotypes that can be inferred from the tumor RNA expression of a specific patient based on the phenotypes of each group, thereby performing a process of specifying heterogeneity.

[0102] According to the heterogeneity of this embodiment, it refers to the different characteristics of the cell composition that constitutes a particular patient's tumor. In particular, cancer cells may have different genetic and epigenetic variations during division, that is, there may be cancer cells with different phenotypes in a tumor. Traditional pathological diagnosis methods usually match one diagnostic name for one patient. According to the single-cell analysis method of the present invention, single-cell data can be used to match two or more phenotypes for one patient. In other words, the single-cell analysis device of the present invention can match a corresponding diagnostic name for each cell or each cell population. The diagnostic name can be provided by referring to the result value of the prediction model and confirming the gene phenotype expressed in the subtype.

[0103] Figure 9 FIG is a schematic diagram of a tumor analysis module of a single cell analysis device for specifically processing tumor heterogeneity according to various embodiments of the present invention. Figure 9 , which is understandable Figure 8 The described tumor analysis module (e.g. Figure 2 The tumor analysis module 114 in the heterogeneity specific processing process is performed. Figure 8Steps S810 to S840 can be performed by Figure 9 910 and 920 in the description, and Figure 8 Step S850 can be achieved by Figure 9 920 to 940 are described below.

[0104] Reference Figure 9 , the tumor analysis module can predict subtypes and grades through precise subclustering analysis. In this embodiment, the tumor analysis module focuses on the significantly different phenotypes of cancer cells in different patients, thereby increasing the sample-to-sample variance. In this case, since it is difficult to identify common patterns between cells from different patients, the tumor analysis module will perform a normalization process on the sample covariance as shown in 910. Afterwards, the tumor analysis module confirms whether the clonality of the entire single cell can be identified to determine whether the normalization is completed correctly. Subsequently, as shown in 920, the tumor analysis module clusters the data through unsupervised learning.

[0105] According to this embodiment, the tumor analysis module can perform dimensionality reduction before or simultaneously with clustering. As shown in diagram 930, the tumor analysis module determines which dimensions to use based on the covariance between features. This process can be performed before clustering. The tumor analysis module then identifies biologically significant cell populations and classifies these newly defined populations by phenotype and genotype to map their subtypes and grades.

[0106] According to an embodiment, the tumor analysis module is capable of specifying the single-cell heterogeneity of a specific patient's tumor and predicting metastasis information for that subtype. This can be performed based on information previously confirmed by standard parameters. In other words, the tumor analysis module of the present invention uses machine learning to infer possible phenotypes based on the RNA expression of a specific patient's tumor and confirms them using existing models. Furthermore, if the analysis results indicate that the phenotype does not fall within the existing range of phenotypes, it can be identified as a new phenotype, and metastasis information and appropriate treatment information for this new phenotype can be processed and provided.

[0107] According to an embodiment, the tumor analysis module can predict drugs and treatments that are suitable for a particular patient's tumor. This can include a process in which the tumor analysis module makes a judgment by integrating information learned based on existing models with information obtained through new phenotypic grouping. Information about such drugs and treatments can be stored in a memory (e.g., Figure 2 In addition, according to various embodiments of the present invention, the single cell analysis device (e.g., Figure 1The single-cell analysis device 100 shown can perform a comprehensive analysis of information such as subtype, grade, and prognosis through repeated learning, thereby predicting and recommending at least one drug and treatment method.

[0108] On the other hand, embodiments of the present invention may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code and, when executed by a processor, generate a program module to perform the operations of embodiments of the present invention. The recording medium may be implemented as a computer-readable recording medium.

[0109] Computer-readable recording media include all types of recording media capable of storing instructions interpreted by a computer, such as read-only memory (ROM), random access memory (RAM), magnetic tapes, magnetic disks, flash memory, optical data storage devices, and the like.

[0110] As described above, the embodiments of the present invention have been described with reference to the accompanying drawings. It will be understood by those skilled in the art that the present invention may be implemented in a manner different from the embodiments without changing the technical concept or essential features of the present invention. The embodiments of the present invention are merely illustrative and should not be construed restrictively.

Claims

1. A single-cell analysis device for performing tumor analysis using artificial intelligence, characterized in that: The single-cell analysis device for performing tumor analysis using artificial intelligence includes: Memory; Ministry of Communications; and processor, The processor is configured to: acquiring, via the communication unit, single-cell data regarding a tumor of a specific patient; Preprocessing the single-cell data of the tumor by a data preprocessing module; Performing type classification based on the pre-processed data by a cell type definition module; Predicting a phenotype inferred from the RNA expression of the tumor of the specific patient using a single-cell state prediction module; Specifying the heterogeneity of the tumor of the specific patient by a tumor analysis module; And the analysis result generation module generates and provides a report of the analysis result of the tumor of the specific patient.

2. The single cell analysis device according to claim 1, characterized in that The single-cell data about the tumor of the specific patient includes RNA sequencing data of each single cell among a plurality of cells present in the tumor extracted from the specific patient, The processor is configured to: The single cell data on the tumor of the specific patient obtained through the communication unit and result data of analyzing the single cell data are distributed to the specific patient.

3. The single cell analysis device according to claim 2, characterized in that The data preprocessing module is configured to: Filtering the single cell data based on multiple parameters, and comparing the single cell data with the single cell data of other tumors stored in the memory for correction, The multiple parameters are parameters related to data quantity, data quality or data repeatability.

4. The single cell analysis device according to claim 3, characterized in that The cell type definition module is configured to: Distinguishing normal cells from cancer cells based on the single cell data, grouping the cancer cells by phenotype based on gene expression levels, The grouped data is mapped to cancer cell type benchmark data to define the cell type of the pre-processed data.

5. The single cell analysis device according to claim 4, characterized in that The single cell state prediction module is configured as follows: A machine learning prediction model is used to predict the phenotype that can be inferred from the RNA expression of the tumor.

6. The single cell analysis device according to claim 5, characterized in that The single cell state prediction module is configured as follows: Using single cell genes as features and based on annotated tumor pathology results, The state of the single cell is predicted by comparing it with a preset threshold and repeatedly training the model.

7. The single cell analysis device according to claim 6, characterized in that The tumor analysis module is configured to: By performing unsupervised clustering on the single cell data, Grouping was performed based on genotype. and derive new phenotypes that can be distinguished from existing phenotypes based on the formed groupings, When the new phenotype is derived, The tumor phenotype of the specific patient is compared to the novel phenotype for analysis.

8. The single cell analysis device according to claim 2, characterized in that The phenotype that can be inferred from the RNA expression of the patient's tumor includes at least one of the following: subtype; grade; aggressiveness; biological activity (enriched pathways); Biomarker (biological marker); copy number variation (CNV); and single nucleotide variants (SNVs).

9. The single cell analysis device according to claim 7, characterized in that The analysis result generating module is configured to: A report is generated and provided for at least one of a current status result, a metastasis prediction result, and a therapeutic drug result corresponding to the heterogeneity of the tumor of the specific patient.

10. A method for analyzing single tumor cells using artificial intelligence, characterized in that: The method for analyzing single tumor cells using artificial intelligence comprises the following steps: access to single-cell data on patient-specific tumors through the Department of Communications; Preprocessing the single-cell data of the tumor by a data preprocessing module; Performing type classification based on the pre-processed data by a cell type definition module; Predicting a phenotype inferred from the RNA expression of the patient-specific tumor using a single-cell state prediction module; Specifying the heterogeneity of the patient-specific tumor using a tumor analysis module; and And the analysis result generation module generates and provides a report of the analysis result of the specific patient's tumor.

Images