Ai agent-based method and computer system for data analysis - linking ai response area and workspace area to support analysis planning, execution, validation visualization, and automatic result provision

Abstract

A data analysis method performed by a computer system is provided. The data analysis method includes analyzing a user's input prompt and the data to be analyzed to automatically determine an optimal statistical analysis model, visually configuring multiple nodes corresponding to the analysis procedure of the model, and automatically setting parameters, while guiding the user on the progress of the analysis by linking the response area of ​​an AI agent (first area) and the workspace area (second area), which is the node display area, and outputting the final analysis result.

Description

Technology Field

[0001] The present disclosure relates to a data analysis method and a computer system that links the response area and the workspace area of ​​an AI agent to provide analysis progress and analysis results. Background Technology

[0002] Data-driven decision-making is used in various fields, particularly in the medical sector, to analyze vast amounts of data and derive insights. Specialized statistical analysis software, such as SPSS and SAS, is used for the statistical analysis of this data.

[0003] These traditional statistical analysis softwares provide analysis functions such as regression analysis, survival analysis, and exploratory data analysis (EDA). However, most software is based on a graphical user interface (GUI), requiring users to personally locate the desired analysis method within a complex menu tree and manually set numerous statistical parameters.

[0004] This approach makes it difficult for users who do not possess deep knowledge of statistics—that is, users who are knowledgeable in the field but are not statistical experts. Consequently, there was an issue of inefficiency where time and resources were wasted due to repetitive trial and error during the analysis process.

[0005] Therefore, there is a need for a data analysis platform that guides even non-statistical experts to easily start data analysis through natural language queries, while presenting the data analysis process through a visual and intuitive interface.

[0006] Korean Patent Publication No. 10-2025-0067105 discloses a data classification and analysis and report generation system.

[0007] The information described above is for illustrative purposes only and may include content that does not constitute part of the prior art. The problem to be solved

[0008] An embodiment may provide a data analysis method that automatically determines an optimal statistical analysis model by analyzing a user's input prompt and data to be analyzed, visually configures multiple nodes corresponding to the analysis procedure of the model, automatically sets parameters, guides the user on the progress of the analysis by linking the response area of ​​the AI ​​agent (first area) and the workspace area (second area), which is a node display area, and outputs the final analysis result. means of solving the problem

[0009] In one aspect, a data analysis method performed by a computer system comprises: receiving data to be analyzed from a user terminal; receiving an input prompt for an AI (Artificial Intelligence) agent from the user terminal to request analysis of the data to be analyzed, wherein the input prompt is input through a first area of ​​the user terminal and the input prompt and the response of the AI ​​agent to the input prompt are displayed in the first area; determining a statistical analysis model to analyze the data to be analyzed from among a plurality of statistical analysis models based on the analysis results of the data to be analyzed and the input prompt by a Large Language Model (LLM) associated with the AI ​​agent; displaying a plurality of nodes corresponding to pre-set analysis procedures of the determined statistical analysis model in a second area of ​​the user terminal distinct from the first area, wherein each of the nodes includes at least one parameter that must be set to analyze the data to be analyzed in each analysis procedure; and automatically setting at least one parameter included by at least one of the plurality of nodes based on the data to be analyzed. A data analysis method is provided, comprising the step of outputting an analysis result of analyzing the data to be analyzed based on the parameters set above, and further comprising the step of linking the status of the progress of the operation displayed in the first area with the display of a component displayed in the second area while the operation to perform analysis on the data to be analyzed is in progress.

[0010] The progress of the above operation includes at least one of the determination of the statistical analysis model, the automatic setting of the parameters, the completion of setting the parameters to generate the analysis results, or the completion of generating the analysis results, and the component may be at least one of each node, the parameters included by each node, or the user interface included by each node.

[0011] The above at least one parameter may be configured to be manually set by the user of the user terminal through the second area, or configured to be set through an additional prompt entered by the user through the first area.

[0012] If the progress of the above work is determined by the statistical analysis model, the linking step may link the display of the first area and the second area by displaying text describing the target statistical analysis performed through the determined statistical analysis model in the first area and displaying the nodes corresponding to the analysis procedures for performing the target statistical analysis in the second area.

[0013] The progress of the above operation is the automatic setting of the above parameter, and the linking step can link the display of the first area and the second area by visually highlighting a node including the automatically set parameter or at least one component among the automatically set parameters in the second area as the value of the automatically set parameter is displayed in the first area, or by applying a visual effect to the at least one component.

[0014] The progress of the above operation is the completion of setting the parameters for generating the above analysis results, and the linking step can link the display of the first area and the second area by visually highlighting or applying a visual effect thereto at least one component among the nodes in the second area that is related to the generation of the above analysis results or an element included in the above data processing node, as text indicating the completion of setting the above parameters is displayed in the first area.

[0015] The status of the progress of the above operation is the completion of the generation of the above analysis result, and the step of linking is such that, as text indicating the completion of the generation of the above analysis result and a first user interface for viewing the above analysis result are displayed in the first area, a second user interface for viewing the above analysis result is displayed in the second area in a display form common to the first user interface at a data processing node among the nodes related to the generation of the above analysis result, thereby linking the displays of the first area and the second area, and the step of outputting the above analysis result is such that, as the first user interface or the second user interface is selected, the above analysis result can be output through a separate window distinguished from the first area and the second area at the user terminal.

[0016] The above statistical analysis models may include at least one of an exploration analysis model for performing exploratory data analysis (EDA); a survival analysis model for performing survival analysis; or a regression analysis model for performing regression data analysis.

[0017] The above statistical analysis models include an exploratory analysis model and a survival analysis model, and if the determined statistical analysis model is an exploratory analysis model, the step of displaying the nodes includes displaying a category selection node, a data selection node, a basic statistical variable selection node, an advanced statistical variable selection node, a distribution and outlier selection node, a plot selection node, and a data processing node, and if the determined statistical analysis model is a survival analysis model, the step of displaying the nodes may include displaying a category selection node, a data selection node, and a data processing node.

[0018] The above data analysis method may further include the step of receiving an additional prompt requesting an interpretation of the analysis result through the first area; and the step of displaying an interpretation of the analysis result in the first area based on the analysis result by the LLM and the analysis result for the additional prompt. Effects of the invention

[0019] A data analysis method can be provided that enables users who have sufficient knowledge of the data to be analyzed but lack knowledge of statistical analysis to easily perform statistical analysis optimized for the data to be analyzed by entering prompts such as natural language input.

[0020] A data analysis method can be provided that enhances user convenience by automating the process of setting complex analysis parameters when performing statistical analysis on the data to be analyzed.

[0021] A data analysis method can be provided that visually outputs the statistical analysis process of target data using AI through the response area and workspace area of ​​interconnected AI agents. Brief explanation of the drawing

[0022] FIG. 1 illustrates a data analysis method according to one embodiment, wherein a first area and a second area of ​​a user terminal are linked and displayed as statistical analysis of the data to be analyzed is performed. FIG. 2 shows a computer system that performs a data analysis method according to one embodiment. FIG. 3 is a flowchart illustrating a data analysis method that displays a first area and a second area of ​​a user terminal in conjunction as statistical analysis of the data to be analyzed is performed according to one embodiment. FIGS. 4a to 14 illustrate a method of displaying a first area and a second area of ​​a user terminal in conjunction as statistical analysis of the data to be analyzed is performed according to one example. Figure 15 illustrates a method for outputting the results of an analysis of data to be analyzed interpreted by an LLM according to one example. Specific details for implementing the invention

[0023] Hereinafter, embodiments will be described in detail with reference to the attached drawings.

[0025] FIG. 1 illustrates a data analysis method according to one embodiment, wherein a first area and a second area of ​​a user terminal are linked and displayed as statistical analysis of the data to be analyzed is performed.

[0026] Referring to FIG. 1, a method for performing statistical analysis on data to be analyzed by a computer system (100) and outputting the analysis results through a user terminal (50) is described.

[0027] The data subject to analysis may include, for example, medical data, financial data, marketing data, etc. For instance, the data subject to analysis may be medical data, and such medical data may include patients' demographic information, clinical characteristics, diagnostic information, treatment history, genomic data, or test results.

[0028] The data subject to analysis is data processed for statistical analysis and may be structured data having a specific format and structure; for example, it may be data in CSV or Excel file format. As an example, the data subject to analysis may be data processed into a form that can be analyzed using professional statistical analysis software such as SPSS or SAS.

[0029] The data to be analyzed may consist of at least one table, and such a table may include multiple columns, wherein each column represents a specific variable or attribute of the data, and each row of the table may represent an individual observation. Each cell within the table may contain the value of a specific variable for a specific observation.

[0030] The data subject to analysis can be uploaded for analysis by a user of the user terminal (50), and the computer system (100) can statistically analyze the uploaded data subject to analysis to generate an analysis result.

[0031] The data subject to analysis in the implementation may include an AI (Artificial Intelligence) agent and a Large Language Model (LLM) (200) associated with the AI ​​agent.

[0032] The LLM (200) is a large-scale neural network model pre-trained with a vast amount of text data, capable of processing and inferring human language based on Natural Language Understanding (NLU) and Generation (NLG) capabilities. In particular, the LLM (200) can perform the role of identifying the intent contained in the user's natural language input prompt and analyzing the structure of the data to be analyzed to infer an optimal statistical analysis model. Thus, in the data analysis method of the embodiment, as the user uploads the data to be analyzed and inputs a natural language input prompt through the AI ​​agent, the computer system (100) can statistically analyze the data to be analyzed using the LLM (200) to generate an analysis result and display the analysis result on the user terminal (50).

[0033] In an embodiment, when an analysis of the data to be analyzed is performed, the progress status and the results of the analysis during the operation to perform the analysis of the data to be analyzed can be displayed on the screen of the user terminal (50).

[0034] The user terminal (50) may include a first area (10) and a second area (20) as an area for displaying the progress status and the results of the analysis.

[0035] The first area (10) may be, for example, a conversational interface that receives user input (prompt) and outputs the AI ​​agent's response as an area for interaction with an AI agent. This first area (10) may be an area that displays an input prompt for user input, a processing process for the prompt, and a processing result. The first area (10) may be composed of a prompt input area (14) for inputting a prompt and an area (12) that displays a processing process for the prompt and a processing result. Below, for convenience of explanation, each element (including UI, etc.) displayed in the areas (12, 14) of the first area (10) may be named as a component displayed in the first area (10).

[0036] Meanwhile, the second area (20) is an area corresponding to a workspace for performing analysis on the data to be analyzed, and can diagrammatically display a plurality of nodes (30) corresponding to analysis procedures for performing statistical analysis on the data to be analyzed. Each node may include at least one configurable parameter, and each parameter may be a setting variable used for statistical analysis of the data to be analyzed. For example, the 'Data Selection' node may include a configurable variable for specifying a specific column to be analyzed as a parameter, and the 'Basic Statistics' node may include a variable for selecting whether to calculate the average or median as a parameter. Below, for convenience of explanation, each node displayed in the second area (20), or the parameters and other elements (including UI, etc.) included in the node, may be named as a component displayed in the second area (20).

[0037] In performing an analysis of the data to be analyzed in the embodiment, while the work to perform the analysis of the data to be analyzed is in progress, the status of the work displayed in the first area (10) and the display of the component displayed in the second area (20) may be linked. That is to say, while the work to perform the analysis of the data to be analyzed is in progress, the component included in the first area (10) and the component included in the second area (20) may be displayed in conjunction with each other.

[0038] Thus, a user who wishes to statistically analyze the data to be analyzed can intuitively check the progress of the work to perform the analysis through the first area (10) and the second area (20).

[0039] In the embodiment, the computer system (100) may be an electronic device corresponding to a server. Meanwhile, the user terminal (50) is an electronic device used by a user to obtain analysis results for data to be analyzed, and may be, for example, a smartphone, a PC (personal computer), a notebook computer, a laptop computer, a tablet, an Internet of Things device, or a wearable computer.

[0040] The computer system (100) can provide a unified bidirectional workflow by organically integrating a first area (10), which is a natural language-based prompt interface, and a second area (20), which is a graph-based node analysis environment, as described.

[0041] Existing data analysis tools separated node-based graphical workflows from natural language prompt interfaces, limiting users to unidirectional communication. In other words, users had to perform separate tasks, such as manually selecting analyses within the graphical user interfaces provided by complex software like SPSS and SAS, or generating code via prompts.

[0042] However, the computer system (100) of the embodiment can receive a user's natural language query through the first area (10), convert it into a structured command through the LLM (200), and transmit the command to the analysis nodes (30) of the second area (20) using an interaction module (e.g., an MCP server) to perform analysis.

[0043] In addition, the computer system (100) can enable bidirectional and repetitive conversation between the user and the analysis workflow by returning the analysis results to the user terminal (50) and accepting follow-up questions through additional prompts. Thus, the data analysis method of the embodiment can perform intuitive and conversation-oriented data analysis while ensuring visual traceability and reproducibility of each analysis procedure when performing analysis on the data to be analyzed.

[0044] In particular, through this bidirectional interaction, the user may select a specific node (30) or analysis result in the second area (20), and the computer system (100) may automatically generate a context-aware prompt in the first area (10) for refinement or further exploration based on this.

[0045] The specific method for linking the data analysis method of the embodiment and the display between the first area (10) and the second area (20) will be described in more detail with reference to FIGS. 2 to 15, which will be described later.

[0047] FIG. 2 is a block diagram showing a computer system that performs a data analysis method according to one embodiment.

[0048] Referring to FIG. 1 according to embodiments, the computer system (100) described above may include configurations such as those illustrated. For example, a computer program for implementing the method of the embodiment may be installed and run on the computer system (100), and the computer system (100) may perform the method of the embodiment under the control of the run computer program.

[0049] The data analysis method according to the embodiments may be performed through a program running on a computer or an application dedicated to a mobile terminal. For example, the method of the embodiments may be implemented in the form of a program that operates independently, or configured as an in-app form of a specific application so as to be able to operate on said specific application. Such a specific application may be configured to provide interaction with an AI agent to a user terminal (50) and to display a first area (10) and a second area (20).

[0050] The computer system (100) may be a server as an electronic device. Alternatively, the computer system (100) may be an electronic device such as a personal computer (PC), notebook computer, laptop computer, tablet, Internet of Things device, smartphone, or wearable computer.

[0051] Meanwhile, the user terminal (50) may also include components similar to those of the computer system (100), and redundant descriptions regarding this are omitted. The user terminal (50) may be, for example, a smartphone PC (personal computer), a notebook computer, a laptop computer, a tablet, an Internet of Things device, or a wearable computer.

[0052] As illustrated, the computer system (100) may include a memory (110), a processor (120), a communication interface (130), and an input / output interface (140) as components for executing the data analysis method of the embodiment.

[0053] Memory (110) is a computer-readable recording medium and may include a non-perishable mass storage device such as RAM (random access memory), ROM (read only memory), and a disk drive. Here, a non-perishable mass storage device such as a ROM and a disk drive may be included in the computer system (100) as a separate permanent storage device distinct from memory (110). Additionally, an operating system and at least one program code may be stored in memory (110). These software components may be loaded into memory (110) from a computer-readable recording medium separate from memory (110). This separate computer-readable recording medium may include a computer-readable recording medium such as a floppy drive, disk, tape, DVD / CD-ROM drive, or memory card. In another embodiment, software components may be loaded into memory (110) through a communication interface (130) rather than a computer-readable recording medium. For example, software components can be loaded into the memory (110) of the computer system (100) based on a computer program installed by files received through the network (160).

[0054] The processor (120) may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input / output operations. Instructions may be provided to the processor (120) via memory (110) or a communication interface (130). For example, the processor (120) may be configured to execute instructions received according to program code stored in a recording device such as memory (110).

[0055] That is, the processor (120) can manage the components of the computer system (100) and can execute programs or applications used by the computer system (100). For example, when an application for performing the data analysis method of the embodiment is executed on a user terminal (50), the processor (120) can enable the user terminal (50) to provide interaction with an AI agent and display the first area (10) and the second area (20). In addition, the processor (120) can process operations necessary for the execution of various programs or applications and the processing of data, and may be at least one processor of the computer system (100) or at least one core within the processor.

[0056] A communication interface (communication unit (130)) may provide a function for a computer system (100) to communicate with another computer system (not shown) through a network (160). For example, requests, commands, data, files, etc. generated by the processor (120) of the computer system (100) according to program code stored in a recording device such as memory (110) may be transmitted to another computer system through the network (160) under the control of the communication interface (130). Conversely, signals, commands, data, files, etc. from another computer system may be received by the computer system (100) through the communication interface (130) of the computer system (100) via the network (160). Signals, commands, data, etc. received through the communication interface (130) may be transmitted to the processor (120) or memory (110), and files, etc. may be stored in a storage medium (the permanent storage device described above) that the computer system (100) may further include. For example, communication The interface (130) may be a hardware module such as a network interface card, a network interface chip, and a networking interface port of a computer system (100), or a software module such as a network device driver or a networking program.

[0057] The input / output interface (140) may be a means for interfacing with an input / output device (150). For example, the input device may include a device such as a microphone, keyboard, or mouse, and the output device may include a device such as a display or speaker. As another example, the input / output interface (140) may be a means for interfacing with a device in which the functions for input and output are integrated into one, such as a touchscreen. The input / output device (150) may be composed of a computer system (100) and a single device.

[0058] Additionally, in other embodiments, the computer system (100) may include more components than those of FIG. 2. However, it is not necessary to clearly illustrate most of the prior art components. For example, the computer system (100) may be implemented to include at least some of the input / output devices connected to the input / output interface (140) described above, or may include other components such as a transceiver, a GPS (Global Positioning System) module, a camera, various sensors, a database, etc.

[0059] In the detailed description to be described later, for convenience of explanation, embodiments are described with the computer system (100) as the focus, and communication with the user terminal (50) and operation on the user terminal (50) side may be briefly explained or omitted.

[0060] Additionally, in the detailed description to be provided below, for convenience of explanation, operations (steps) performed by the configuration of the computer system (100) (e.g., processor (120), etc.) may be described as being performed by the computer system (100).

[0061] The description of the technical features described above with reference to FIG. 1 can be applied as is to FIG. 2, so redundant descriptions are omitted.

[0063] FIG. 3 is a flowchart illustrating a data analysis method that displays a first area and a second area of ​​a user terminal in conjunction as statistical analysis of the data to be analyzed is performed according to one embodiment.

[0064] Referring to FIG. 3, a method in which a computer system (100) uses an LLM (200) and statistical analysis model(s) to analyze data to be analyzed and provides the situation and analysis results to a user terminal (50) while the work to perform the analysis is in progress is described in more detail.

[0065] Meanwhile, in this regard, FIGS. 4a to 14 illustrate a method of displaying a first area and a second area of ​​a user terminal in conjunction as statistical analysis of the data to be analyzed is performed according to one example.

[0066] In step (310), the computer system (100) can receive data to be analyzed from the user terminal (50).

[0067] The user can upload the data to be analyzed to the computer system (100) for analysis. For example, the user can upload the data to be analyzed stored in local storage space for analysis by clicking an upload button on the user interface (UI) displayed on the user terminal (50). As described above, the data to be analyzed may be structured data having a specific format and structure as data processed for statistical analysis, and may be, for example, data in the format of a CSV or Excel file. For example, the data to be analyzed may be data processed into a form that can be analyzed using professional statistical analysis software such as SPSS, SAS, etc.

[0068] The computer system (100) can load the uploaded data to be analyzed for statistical analysis.

[0069] In step (320), the computer system (100) may receive an input prompt from the user terminal (50). That is, the computer system (100) may receive an input prompt from the user terminal (50) for an AI agent to request analysis of the (uploaded) data to be analyzed. The input prompt may be composed of natural language.

[0070] Such input prompts can be input through a first area (10) of the user terminal (50) (e.g., a prompt input area (14) for). As described above, the first area (10) can be configured to display an input prompt and the AI ​​agent's response to the input prompt as an area for interaction with an AI agent.

[0071] In this regard, FIG. 4a illustrates a first area (10) and a second area (20) in a situation where an input prompt (430) is entered after the data to be analyzed has been uploaded. As illustrated, information (410) (such as a filename) regarding the uploaded data to be analyzed may be displayed in the first area (10), and additional details (420) regarding the data to be analyzed may be displayed in the second area (20). The information (410) may include a UI (such as a selection box) for selecting one if there are multiple uploaded data to be analyzed. The user may request a statistical analysis of the data to be analyzed by entering an input prompt (430) composed of natural language. For example, the input prompt (430) may be a sentence such as "I want to know the state of the variables of the data I uploaded by subtype," as in the illustrated example. Unlike the illustrated example, the input prompt (430) may be an incomplete sentence.

[0072] In step (330), the computer system (100) can determine a statistical analysis model to analyze the data among a plurality of statistical analysis models based on the analysis results of the data to be analyzed by the LLM (200) associated with the AI ​​agent and the input prompt. That is to say, the input prompt (430) and the uploaded data to be analyzed can serve as prompts to the LLM (200).

[0073] Each of the statistical analysis models may be an analysis module implemented to perform statistical analysis on the target data according to a specific predefined algorithm. Each statistical analysis model may be predefined to perform statistical analysis for a specific purpose and method. Each of the statistical analysis models may be a pre-trained artificial intelligence-based model. Each statistical analysis model may be pre-trained to perform statistical analysis in a specific manner.

[0074] For example, statistical analysis models may include at least one of, for example, an exploration analysis model for performing Exploratory Data Analysis (EDA); a survival analysis model for performing survival analysis; a regression analysis model for performing regression data analysis; or an analysis model for performing analysis through T-tests. In addition, statistical analysis models may include models pre-trained to perform statistical analysis provided by general specialized statistical analysis software such as SPSS, SAS, etc.

[0075] To explain each statistical analysis model in more detail, the exploration analysis model is designed to perform exploratory data analysis (EDA). Prior to full-scale analysis, it can be used to identify basic statistical characteristics such as data distribution, mean, and variance, and to discover hidden patterns or outliers through visualization. The survival analysis model is used to analyze factors influencing the time leading up to a specific event (e.g., patient survival or death) and can be configured to perform analyses such as Kaplan-Meier analysis or Cox regression. The regression analysis model can be used to identify the influence of one or more independent variables on a dependent variable and to predict the value of a specific variable. The T-test analysis model can be used to verify whether there is a statistically significant difference between the means of two groups.

[0076] LLM (200) can determine an appropriate statistical analysis model among multiple statistical analysis models through, for example, the following process. First, LLM (200) can analyze the intent of an input prompt entered by a user based on its natural language understanding (NLU) capabilities. It can extract the core intent of "survival analysis" from a prompt such as "find the variables that affect the survival rate." At the same time, LLM (200) can identify the structure of the uploaded data to be analyzed, that is, the names of each column (e.g., "age," "gender," "survival period," "event occurrence status") and the data type. Next, LLM (200) can select the optimal statistical analysis model that best matches the predefined purpose of each statistical analysis model by comprehensively considering the user's intent extracted from the prompt and the structural characteristics of the data. For example, if there is a keyword 'survival rate' in the prompt and columns such as 'survival period' and 'event occurrence' exist in the data, LLM (200) can determine the survival analysis model as the optimal statistical analysis model.

[0077] In this regard, FIG. 4b illustrates a case where, according to the input prompt (430), it is determined that the data to be analyzed requires exploratory analysis, and the exploratory analysis model is determined to be the optimal statistical analysis model.

[0078] In FIG. 4b, only a portion of the first area (10) is shown for convenience of explanation, and the same component(s) as in FIG. 4a may be displayed in the second area (20). Meanwhile, the component of the first area (10) may further include a UI (450) that allows the user to determine the optimal statistical analysis model. In the illustrated example, when the UI (450) is selected, an exploratory analysis model may be determined as the optimal statistical analysis model. Thus, exploratory analysis of the data to be analyzed can be performed.

[0079] In step (340), the computer system (100) may display a plurality of nodes corresponding to preset analysis procedures of the determined statistical analysis model in a second area (20) that is distinct from the first area (10) of the user terminal (50). As described above, the second area (20) is an area corresponding to a workspace for performing analysis on the data to be analyzed, and may be an area for graphically displaying a plurality of nodes (30) corresponding to analysis procedures for performing statistical analysis on the data to be analyzed.

[0080] Each node of the nodes (30) may include at least one parameter that must be set to analyze the data to be analyzed in each analysis procedure of the statistical analysis performed by the determined statistical analysis model. This parameter may be a settable parameter and may be a setting variable used for statistical analysis of the data to be analyzed.

[0081] In step (350), the computer system (100) can automatically set at least one parameter included in at least one of the multiple nodes displayed in the second area (20) based on the data to be analyzed.

[0082] Next, the computer system (100) can generate an analysis result after analyzing the data to be analyzed based on the set parameters using a determined statistical analysis model, and in step (360), output the analysis result of analyzing the data to be analyzed.

[0083] Additionally, as in step (370), the computer system (100) may obtain and output an interpretation of the analysis result from the LLM in accordance with an additional prompt from the user. For example, the computer system (100) may receive an additional prompt requesting an interpretation of the analysis result through the first area (10), and may display an interpretation of the analysis result in the first area (10) based on the analysis result by the LLM and the analysis result for the additional prompt.

[0084] Meanwhile, in the embodiment, as in step (305), the computer system (100) can synchronize the status of the progress of the work displayed in the first area (10) with the display of the component displayed in the second area (20) while the work to perform analysis on the data to be analyzed is in progress. That is to say, while the work to perform analysis on the data to be analyzed is in progress, the component included in the first area (10) and the component included in the second area (20) can be displayed in conjunction with each other.

[0085] Thus, through the data analysis method of the embodiment, even a user without statistical knowledge can easily perform in-depth analysis of data based on their domain knowledge by inputting natural language prompts. In addition, the entire process from determining the optimal analysis model to setting complex parameters is automated, thereby minimizing time consumption and potential errors associated with manual work and maximizing the efficiency of the analysis. Furthermore, the computer system (100) visually presents the work process of the AI ​​agent by linking the components of the first area (10) and the components of the second area (20), thereby allowing the user to intuitively understand the entire analysis process and secure confidence in the final result.

[0086] In this way, in the embodiment, while the analysis of the data to be analyzed is in progress, the components included in the first area (10) and the components included in the second area (20) can be displayed in conjunction with each other throughout the entire process, such as the registration of the data to be analyzed, the determination of an optimal statistical analysis model for analyzing the data to be analyzed, the setting of variables for performing statistical analysis on the data to be analyzed, and the generation and output of analysis results.

[0087] The progress of the operation to perform analysis on the data to be analyzed may include, for example, at least one of the determination of a (optimal) statistical analysis model, the automatic setting of parameters included in the node, the completion of setting the said parameters to generate analysis results for the data to be analyzed, or the completion of generating said analysis results.

[0088] The computer system (100) can synchronize the display of a component displayed in the second area (20) according to the progress status of the above task displayed in the first area (10). Here, the component included in the second area (20) may be at least one of each node included in the second area (20), a parameter included in each node, or a user interface included in each node. Thus, in the first area (10), the progress status of the above task can be described through a component including text, and accordingly, the display of the component included in the second area (20) can be synchronized.

[0089] Specific examples of how a computer system (100) links the progress of a first area (10) with the display of components in a second area (20) may be as follows. For example, depending on the progress of the narrative in the first area (10), the corresponding node may be visually highlighted by changing the outline color of the relevant node among the various nodes displayed in the second area (20) or by highlighting the background. Or / additionally, when a situation in which a column of data to be analyzed is set in the first area (10) is displayed, the computer system (100) may update the display state of a parameter input field within a specific node in the second area (20), for example, by indicating that a value according to the situation has been automatically filled in the field, and apply visual effects such as temporarily blinking the area so that the user can immediately recognize the automatic setting of the parameter value. Alternatively, when it is stated that the analysis of the data to be analyzed in the first area (10) is completed, a UI that allows viewing the analysis results in the corresponding node of the computer system (100) may be highlighted, or the display state of the UI may be changed from disabled to enabled, thereby intuitively guiding the user to proceed to the next step.

[0090] In this way, the computer system (100) can organically link the display of the first area (10) and the second area (20) by controlling the display of the component through various UI / UX techniques such as color change, highlighting, blinking, and activation state change, thereby enabling the user to clearly track and understand the flow of the analysis work automated by the AI ​​agent.

[0091] Meanwhile, in FIG. 5, the first area (10) and the second area (20) of the user terminal (50) are shown in the case where the optimal statistical analysis model for analyzing the data to be analyzed is an exploratory analysis model.

[0092] FIG. 5 may show, for example, the first area (10) and the second area (20) after the UI (450) is selected in FIG. 4b.

[0093] As illustrated, as the exploratory analysis model is determined as the optimal statistical analysis model, the exploratory analysis model (EXPLORATION) may be displayed at the top right of the second area (20). The user may also select a different statistical analysis model by adjusting the selection box in the area as needed. Accordingly, the display of components in the second area (20) and the first area (10) can be controlled in conjunction.

[0094] When the exploratory analysis model is determined to be the optimal statistical analysis model, the second area (20) can display nodes (530) as illustrated. The nodes (530) may correspond to pre-set analysis procedures for performing exploratory analysis. These displayed nodes (530) may include a category selection node, a data selection node, a basic statistics variable selection node, an advanced statistics variable selection node, a distribution and outlier selection node, a plot selection node, and a process data node.

[0095] The Category Selection node is the highest level and can be a node for selecting the major type of analysis to be performed, such as 'exploratory analysis' or 'regression analysis'. Here, the value of the variable selected as a parameter can be exploration (exploratory analysis model), kaplan_meier (survival analysis model), cox_regression (survival analysis model), etc., depending on the type of statistical analysis that the determined statistical analysis model can analyze.

[0096] The Data Selection node can be a node that specifies a particular range (column) to analyze within the data to be analyzed.

[0097] The Basic Statistics node can be a node for selecting statistical variables (statistics: mean, median, standard deviation, etc.) to be analyzed from the data to be analyzed.

[0098] The Advanced Statistics node can be a node for selecting descriptive statistical indicators that are more in-depth (to analyze) than basic statistics, such as minimum / maximum values, percentiles, and confidence intervals.

[0099] The Distribution & Outlier node may be a node that analyzes how data is spread (distribution) and selects options to detect extreme values ​​(outliers) that fall outside the normal range.

[0100] The Plot node can be a node that selects the type of graph (e.g., histogram, bar graph, etc.) to visualize the analysis results.

[0101] The Process Data node may be an execution node that performs a final analysis (Run Analysis) and checks the generated results (View Result) based on all options set in the preceding nodes.

[0102] Meanwhile, as another example, if the determined statistical analysis model is a survival analysis model, the nodes displayed in the second area (20) may include a category selection node, a data selection node, and a data processing node. In this regard, FIG. 11 illustrates an example in which the survival analysis model is determined as the optimal statistical analysis model according to the input prompt (1110), and the UI (1120) is selected to confirm the optimal statistical analysis model, after which the nodes are displayed in the second area (20). Here, the input prompt (1110) may be the first input after the data to be analyzed is uploaded, or it may be an additional prompt (i.e., an additional prompt) that is additionally input after the analysis results are generated and output according to FIGS. 4a to 10. Thus, in the embodiment, after the data to be analyzed is uploaded once, various statistical analyses can be performed on the data to be analyzed using various statistical analysis models.

[0103] Referring again to FIG. 5, an embodiment is described as follows: a component of the first region (10) may display a message (510) (an example of a component) indicating that exploratory analysis is being performed, and then may output a message (520) indicating that parameters necessary to perform exploratory analysis must be set. That is, the message (520) may indicate that at least one parameter included in a necessary node among the illustrated nodes (530) is automatically set. When the UI (512) containing the message (520) is selected, the automatic setting of these parameters may begin.

[0104] Meanwhile, as illustrated, at least one parameter included in the node may be configured to be manually set by the user of the user terminal (50) through the second area (20), or configured to be set through an additional prompt entered by the user through the first area (10). That is to say, the parameter may be set automatically, but an unset parameter may be modified by the user, and even a parameter that has already been set may be modified to a different value by the user. In this way, the second area (20) may function as a workspace that the user can interact with.

[0105] Returning to FIG. 5, as in FIG. 4b, if the progress of the work displayed in the first area (10) is determined by a statistical analysis model, the computer system (100) can link the displays of the first area (10) and the second area (20) by displaying text (message (510)) describing the target statistical analysis (exploration analysis in the illustrated example) performed through the statistical analysis model determined in the first area (10), and by displaying nodes (530) corresponding to the analysis procedures for performing the target statistical analysis in the second area (20).

[0106] As illustrated in FIG. 6, according to the message (510) displayed in the first area (10), the value of the parameter 'exploration' can be automatically selected in the category selection node (610), and the data to be analyzed can be automatically set in the data selection node (620).

[0107] Meanwhile, when UI (512) is selected, automatic settings for additional parameters can be performed to perform analysis on the data to be analyzed.

[0108] In the case where the status of the work for analyzing the data to be analyzed is the automatic setting of the above parameters, a message indicating that the automatic setting of the parameters is being performed may be displayed in the first area (10). At this time, as the value of the automatically set parameter is displayed in the first area (10), the computer system (100) may link the display of the first area and the second area by visually highlighting a node containing the automatically set parameter or at least one component of the automatically set parameter in the second area (20), or by applying a visual effect to the at least one component.

[0109] For example, as illustrated in FIG. 7, a message (702) indicating that a column of the subject of analysis is set among the data included in the data to be analyzed may be displayed in the first area (10). The message (702) may display the column of the subject of analysis (i.e., the value of the automatically set parameter). At this time, in the second area (20) which is linked to the display in the first area (10), the parameter setting area (725) that specifies a specific range (column) to be analyzed for the related node, the data selection node (710), may be visually distinguished from other parameter setting areas. Additionally, the column to be analyzed (age, bmi...) may also be displayed within the node (710). Meanwhile, the highlight display of the parameter setting area (725) illustrated in FIG. 7 (e.g., the perimeter expressed as a solid line may be highlighted in blue) may be displayed only while the corresponding parameter is being automatically set (i.e., only while the analysis column is being set), and may not be displayed after such automatic setting is completed. In this way, in the second area (20), not only are the parameter values ​​reflected after the automatic setting is completed displayed, but it can also be visualized which parameters are currently being automatically set.

[0110] Additionally, as illustrated in the example in FIG. 8, a message (802) indicating that a statistical variable to be analyzed is set for the data to be analyzed (i.e., the column to be analyzed) may be displayed in the first area (10). The message (802) may indicate the type of statistical variable to be analyzed (e.g., Sample Size, Mean, etc.). At this time, in the second area (20) which is linked to the display in the first area (10), the part corresponding to this automatically set statistical variable (parameter) of the related node, the basic statistical node (810), may be visually distinguished from the rest (parameters). For example, as illustrated, values ​​corresponding to the set statistical variable may be checked. Meanwhile, the user may change the setting of the statistical variable to be analyzed by entering an additional prompt in the first area (10) or by selecting an additional statistical variable in the node (810).

[0111] Next, when the progress of the work for analyzing the data to be analyzed is complete in setting parameters for generating analysis results, a message (804) indicating that the setting of parameters for generating analysis results has been completed may be displayed in the first area (10). As such, as text (message (804)) indicating the completion of parameter setting is displayed in the first area (10), the computer system (100) can synchronize the display of the first area (10) and the second area (20) by visually highlighting or applying visual effects thereto at least one component among the nodes in the second area (20) that is related to the generation of analysis results, such as a data processing node (820) or an element (UI, button, other visual element, etc.) included in the data processing node (820). As illustrated in FIG. 8, the message (804) may indicate that the parameter setting has been completed and the preparation for analysis is complete, and accordingly, a UI (Run Analysis) to start analysis according to the set parameters may be activated in the node (820). Analysis may start when such a UI is selected. However, depending on the embodiment, analysis may be performed automatically even if the UI is not selected.

[0112] Next, when the status of the work for analyzing the data to be analyzed is that the generation of the analysis result has been completed, as shown in FIG. 9, text (message (902)) indicating the completion of the generation of the analysis result and a first UI (904) for viewing the analysis result may be displayed in the first area (10). The message (902) may include text indicating that the analysis has been successfully completed, as illustrated, and asking whether to view the analysis result. In this way, as the text (message (902)) indicating that the generation of the analysis result has been completed is displayed in the first area (10), the computer system (100) can synchronize the display of the first area (10) and the second area (20) by displaying a second UI (912) for viewing the analysis result in the second area (20) at the data processing node (910) related to the generation of the analysis result.

[0113] At this time, the first UI (904) and the second UI (912) can perform the same function of viewing analysis results. Additionally, the first UI (904) and the second UI (912) may have a common display form. A common display form may include at least one of the same color, same shape, same size, same icon, or same font. In FIG. 9, the first UI (904) and the second UI (912) are shown to have a common display form of the same color. The second UI (912) can be changed from an inactive state to an active state when the generation of analysis results is completed.

[0114] The user can recognize that the analysis is completed (i.e., the generation of the analysis result is completed) in both the first area (10) and the second area (20) through the first UI (904) and the second UI (912), and can view the analysis result by selecting either the first UI (904) or the second UI (912).

[0115] The computer system (100) can output an analysis result through a separate window distinct from the first area (10) and the second area (20) on the user terminal (50) as the first UI (904) or the second UI (912) is selected. According to an embodiment, the analysis result may be displayed in the first area (10) or the second area (20) (or part thereof).

[0116] In FIG. 10, an example of an analysis result (1000) displayed through a separate window is illustrated. The illustrated analysis result may be the result of frequency analysis as an example of an exploratory analysis result. As illustrated, the analysis result (1000) may be downloaded or added to a report depending on the user's choice.

[0117] Meanwhile, FIGS. 11 to 14 illustrate a case where the statistical analysis model determined in the aforementioned step (330) is a survival analysis model. In this case, the nodes displayed in the second area (20) may include a category selection node, a data selection node, and a data processing node.

[0118] In FIG. 11, an example is illustrated in which a survival analysis model is determined as the optimal statistical analysis model according to an input prompt (1110), and after the UI (1120) is selected and the optimal statistical analysis model is confirmed, nodes are displayed in the second area (20). Here, the input prompt (1110) may be the first input after the data to be analyzed is uploaded, or it may be an additional prompt (i.e., an additional prompt) that is additionally input after the analysis result in FIG. 10 is generated and output. As such, in the embodiment, once the data to be analyzed is uploaded, various statistical analyses can be performed on the data to be analyzed using various statistical analysis models.

[0119] In Fig. 11, after the survival analysis model was selected, Cox Regression analysis was recommended based on the analysis by LLM (200), and as the UI (1120) was selected and the statistical analysis model was confirmed, the parameter 'cox_regression' was automatically selected in the category selection node.

[0120] In FIG. 12, a message (1210) indicating that the Cox Regression analysis is completed is shown (after the completion of automatic parameter settings for analysis, with the city and description omitted). Meanwhile, as an additional prompt requesting additional analysis is entered by the user, the computer system may repeat the aforementioned steps (330 to 350) to perform additional analysis. Message (1220) shows the additional prompt entered by the user, and accordingly, Kaplan-Meier analysis for pCR variable analysis may be performed. Message (1230) may indicate that a survival analysis model is determined for Kaplan-Meier analysis and may include a UI for confirming the survival analysis model.

[0121] In Fig. 13, a method is illustrated in which parameters are automatically set at each node to perform Kaplan-Meier analysis after the survival analysis model is determined, and the analysis results of the Kaplan-Meier analysis are generated.

[0122] As illustrated in the example, parameters indicating the type of analysis (Kaplan Meier analysis) can be automatically set in the category selection node, and the data file to be analyzed and the columns to be analyzed of the data to be analyzed can be automatically set in the data selection node. After the parameters for the analysis are automatically set, Kaplan Meier analysis can be performed, and after the Kaplan Meier analysis is completed and the analysis results are generated, a UI for viewing the analysis results can be displayed (activated) in both the first area (10) and the second area (20). When the UI is selected, the analysis results of the Kaplan Meier analysis can be displayed, for example, in a separate window.

[0123] Figure 14 shows the results of the Kaplan-Meier analysis in the form of a graph. As shown, the analysis results can be downloaded or added to a report depending on the user's choice.

[0124] Meanwhile, as described above with reference to the aforementioned step (370), the user can request an interpretation of the analysis result from the LLM (200) by entering an additional prompt.

[0125] In this regard, FIG. 15 illustrates a method for an LLM to output an analysis result of data to be analyzed according to one example. For instance, a user may input an additional prompt in the first area (10) asking to "interpret the survival analysis result," and the LLM (200) may generate an analysis result by using this additional prompt and the previously generated analysis result. This analysis result may be displayed in the first area (10). In this way, the user can obtain additional insights into the analysis result based on the statistical analysis model.

[0126] The description of the technical features described above with reference to FIGS. 1 and FIG. 2 can be applied as is to FIGS. 3 to FIGS. 15, so redundant descriptions are omitted.

[0128] The computer system (100) of the embodiment can perform the data analysis method of the embodiment through interaction between a prompt interface module related to the display of the first area (10), a node-based analysis graph interface module related to the display of the second area (20), and a module (interaction module) that controls the linked display between the two areas.

[0129] To explain the data analysis method of the embodiment again with this in mind, the user can initiate an analysis query (input prompt) composed of natural language through the first area (10). When the computer system (100) receives such user input, it can convert the natural language into a structured command (e.g., a JSON object) that adheres to a predefined schema using an LLM (200). The structured command thus generated can be transmitted to an interaction module. Beyond simple data transmission, the interaction module can perform the function of a hub that validates whether the received command conforms to a predefined schema, manages user permissions for the requested operation, and routes the command to an appropriate backend analysis service. Additionally, the interaction module can perform the function of receiving and processing analysis results from the backend analysis service and returning the results back to the first area (10).

[0130] The node-based user interface of the second area (20) can be implemented using a framework such as ReactFlow, for example. When the interaction module transmits a command to a backend analysis service to execute an analysis (e.g., Kaplan-Meier analysis or Cox regression analysis, etc.), the node among the nodes (30) in the graph of the second area (20) corresponding to the analysis task can be updated to reflect its state and display the analysis results.

[0131] The data analysis method of the embodiment may feature a planning, execution, and validation process of data analysis via a prompt, and a bidirectional interaction loop between the first area (10) and the second area (20). The analysis results may also include statistical outputs, machine learning predictions, or natural language descriptions generated in real time by the LLM (200), and these results may be returned to the user immediately. Through this, the user may participate in an analysis process similar to a conversation, asking a question in natural language, for example, "analyze survival of breast cancer patients undergoing NAC," and receiving a response from the computer system (100) based on the analysis planning, execution, and validation.

[0132] In particular, the bidirectional interaction loop of this embodiment may include a function to automatically highlight a specific node or part of the analysis process in the second area (20). For example, if the computer system (100) automatically selects independent and dependent variables required for analysis and visually presents them in the second area (20), the user can verify and finally confirm them (Human in the loop). Through such a hybrid system, both the automation of analysis and the user's control can be secured.

[0134] The device described above may be implemented as a hardware component, a software component, and / or a combination of a hardware component and a software component. For example, the device and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. In addition, other processing configurations, such as parallel processors, are also possible.

[0135] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or instruct the processing unit independently or collectively. Software and / or data may be embodied in any type of machine, component, physical device, computer storage medium, or device so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0136] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. In this case, the medium may continuously store a program executable by a computer, or temporarily store it for execution or download. Furthermore, the medium may be various recording or storage means in the form of a single or several hardware combined, and is not limited to a medium directly connected to a computer system, but may also exist distributed over a network. Examples of media may include magnetic media such as hard disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media, and media configured to store program instructions including ROM, flash memory, etc. Additionally, other examples of media may include recording or storage media managed by app stores that distribute applications or sites and servers that supply or distribute various other software.

[0137] Although the embodiments have been described above with reference to limited examples and drawings, those skilled in the art can make various modifications and variations from the description above. For example, suitable results can be achieved even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0138] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

Claims

Claim 1 A data analysis method performed by a computer system comprises: receiving data to be analyzed from a user terminal; receiving an input prompt for an AI (Artificial Intelligence) agent from the user terminal to request analysis of the data to be analyzed, wherein the input prompt is input through a first area of ​​the user terminal and the input prompt and the response of the AI ​​agent to the input prompt are displayed in the first area; determining a statistical analysis model to analyze the data to be analyzed from among a plurality of statistical analysis models based on the analysis results of the data to be analyzed and the input prompt by a Large Language Model (LLM) associated with the AI ​​agent; displaying a plurality of nodes corresponding to pre-set analysis procedures of the determined statistical analysis model in a second area of ​​the user terminal distinct from the first area, wherein each of the nodes includes at least one parameter that must be set to analyze the data to be analyzed in each analysis procedure; and automatically setting at least one parameter included by at least one of the plurality of nodes based on the data to be analyzed. A data analysis method comprising the step of outputting an analysis result obtained by analyzing the data to be analyzed based on the parameters set above, and further comprising the step of linking the status of the progress of the operation displayed in the first area with the display of a component displayed in the second area while the operation to perform analysis on the data to be analyzed is in progress. Claim 2 A data analysis method according to claim 1, wherein the progress of the above operation includes at least one of the determination of the statistical analysis model, the automatic setting of the parameters, the completion of setting the parameters for generating the analysis result, or the completion of generating the analysis result, and wherein the component is at least one of each node, the parameters included by each node, or the user interface included by each node. Claim 3 A data analysis method according to claim 1, wherein at least one parameter is configured to be manually set by the user of the user terminal through the second area, or configured to be set through an additional prompt input by the user through the first area. Claim 4 A data analysis method according to paragraph 2, wherein, if the progress of the above work is determined by the statistical analysis model, the linking step links the display of the first area and the second area by displaying text describing the target statistical analysis performed through the determined statistical analysis model in the first area and displaying the nodes corresponding to the analysis procedures for performing the target statistical analysis in the second area. Claim 5 A data analysis method according to claim 4, wherein the progress of the above operation is the automatic setting of the above parameter, and the linking step links the display of the first area and the second area by visually highlighting a node including the automatically set parameter or at least one component among the automatically set parameters in the second area as the value of the automatically set parameter is displayed in the first area, or by applying a visual effect to the at least one component. Claim 6 A data analysis method according to claim 5, wherein the progress of the above work is the completion of setting the above parameters for generating the above analysis result, and the linking step is to link the display of the first area and the second area by visually highlighting or applying a visual effect thereto at least one component among the nodes in the second area that is related to the generation of the above analysis result or an element included in the above data processing node, as text indicating the completion of setting the above parameters is displayed in the first area. Claim 7 A data analysis method according to claim 5, wherein the progress of the above operation is the completion of the generation of the above analysis result, and the step of linking is to link the display of the first area and the second area by displaying a second user interface for viewing the above analysis result in a display form common to the first user interface at a data processing node among the nodes related to the generation of the above analysis result in the second area as text indicating the completion of the generation of the above analysis result and a first user interface for viewing the above analysis result are displayed as a result of linking the display of the first area and the second area, and the step of outputting the above analysis result is to output the above analysis result through a separate window distinguished from the first area and the second area at the user terminal as the first user interface or the second user interface is selected. Claim 8 A data analysis method according to claim 1, wherein the statistical analysis models include at least one of an exploration analysis model for performing exploratory data analysis (EDA); a survival analysis model for performing survival analysis; or a regression analysis model for performing regression data analysis. Claim 9 In paragraph 2, the statistical analysis models include an exploratory analysis model and a survival analysis model, and if the determined statistical analysis model is an exploratory analysis model, the step of displaying the nodes is to display a category selection node, a data selection node, a basic statistical variable selection node, an advanced statistical variable selection node, a distribution and outlier selection node, a plot selection node, and a data processing node, and if the determined statistical analysis model is a survival analysis model, the step of displaying the nodes is to display a category selection node, a data selection node, and a data processing node, a data analysis method. Claim 10 A data analysis method according to claim 1, further comprising: receiving an additional prompt requesting an interpretation of the analysis result through the first area; and displaying an interpretation of the analysis result in the first area based on the analysis result by the LLM and the analysis result for the additional prompt.

Images