Method for providing information on major depressive disorder on basis of explainable artificial intelligence algorithm, and device for providing information on major depressive disorder using same

Abstract

The present invention provides a method for providing information on major depressive disorder, the method being implemented by a processor, and a device and a system using same. The method comprises the steps of: receiving electroencephalogram data corresponding to a plurality of electrode channels of an individual; performing an evaluation of the electroencephalogram data on the basis of an explainable artificial intelligence (XAI) algorithm; and selecting a major channel from among the plurality of electrode channels on the basis of the evaluation result, wherein the major channel is defined as a channel having a high contribution to the classification as to whether the individual has major depressive disorder.

Description

Method for providing information on major depressive disorder based on explainable artificial intelligence algorithm and device for providing information on major depressive disorder using the same

[0001] The present invention relates to a method for providing information on major depressive disorder and a device for providing information on major depressive disorder using the same, and more specifically, to a method for providing information on major depressive disorder and a device for providing information on major depressive disorder using the same, which provides information on major depressive disorder based on an explainable artificial intelligence algorithm.

[0002] Mental disorder can refer to a dysfunction manifested in the mind or behavior. Major depressive disorder can be caused by genetic, physical, and psychological factors, such as stress.

[0003] In particular, major depressive disorder, a type of mental disorder, usually develops gradually and can present with symptoms such as hyperactivity, separation anxiety disorder, or intermittent depressive symptoms over several years, including insomnia, sad feelings, obsession with the past, distraction, hopelessness, fatigue, and loss of appetite.

[0004] Major depressive disorder is on the rise in prevalence in modern society, driven by the increasing frequency of exposure to psychological stress. However, major depressive disorder shares many similar symptoms with other major depressive disorders, and the severity varies from person to person, making accurate distinctions difficult.

[0005] Thus, developing optimal classification criteria for major depressive disorder may be important for accurate diagnosis of major depressive disorder.

[0006] Therefore, there is a continuous need to develop new diagnostic criteria and information provision systems for major depressive disorder that can improve diagnostic accuracy.

[0007] Meanwhile, functional magnetic resonance imaging (fMRI), based on dynamic neural activity, has emerged as a diagnostic tool for major depressive disorder.

[0008] More specifically, fMRI analysis revealed that patients with major depressive disorder exhibited different neural responses when reading emotional text compared to healthy individuals. Therefore, fMRI analysis results can provide information for the accurate diagnosis of major depressive disorder.

[0009] Meanwhile, fMRI may cause patients to complain of anxiety or fear during the diagnostic process. Furthermore, fMRI still has many limitations when applied to the diagnosis of major depressive disorder, including the high cost of analysis and spatial and temporal constraints.

[0010] In particular, fMRI may have limitations in providing reliable information for an accurate diagnosis of major depressive disorder, as it only focuses on neural activity during emotional information processing and does not consider important pathologies such as altered cognitive processes.

[0011] Other diagnostic methods for major depressive disorder have been proposed, such as questionnaires or clinician-based questionnaires. However, questionnaires and questionnaires often rely on the subjective judgment of clinicians, which can lead to a high rate of misdiagnosis. Furthermore, patients may conceal their symptoms or fail to adequately describe them, which can hinder adequate symptom sharing with clinicians, potentially increasing the rate of misdiagnosis.

[0012] In this regard, a deep learning-based auxiliary diagnostic system capable of objectively diagnosing patients with major depressive disorder has been developed to reduce diagnostic errors arising from vague subjective diagnostic criteria, but several limitations still exist for practical use.

[0013] More specifically, for existing deep learning-based assistive diagnostic systems, it can be difficult to explain which data characteristics enable accurate diagnoses. Furthermore, to improve diagnostic performance, a large number of electrodes are used to measure EEG signals, which can require significant preparation time and increase patient discomfort.

[0014] The inventors of the present invention sought to develop an artificial neural network-based assistive diagnosis information providing system that applies explainable artificial intelligence technology to simultaneously improve the efficiency and performance of diagnosis.

[0015] In particular, the inventors of the present invention applied explainable artificial intelligence (XAI) technology to select features used in the diagnosis of major depressive disorder among the functional brain features of patients with major depressive disorder.

[0016] The inventors of the present invention were able to recognize that by applying XAI technology to an information provision system, the number of electrodes used for diagnosis can be reduced through selection of major channels that directly affect the diagnosis of patients with major depressive disorder, thereby increasing the possibility of practical application of the diagnosis system.

[0017] As a result, the inventors of the present invention developed a new information provision system that integrates an XAI technology-based automatic channel selection algorithm and an artificial neural network-based prediction model.

[0018] The inventors of the present invention expected that by providing a new system, it would be possible to confirm which neurophysiological characteristics of patients were extracted and used for diagnosis, thereby providing reliable information when assisting diagnosis.

[0019] Furthermore, the inventors of the present invention expected that the practicality and reliability of the diagnostic system could be increased by providing a new system based on a channel selection algorithm.

[0020] Accordingly, the problem to be solved by the present invention is to provide a method, device, and system for providing information on major depressive disorder, which is configured to provide information related to the onset of major depressive disorder by evaluating brain wave data acquired from an individual, selecting key data according to individual characteristics, and applying the data to the learning of an artificial neural network-based prediction model.

[0021] The tasks of the present invention are not limited to the tasks mentioned above, and other tasks not mentioned will be clearly understood by those skilled in the art from the description below.

[0022] In order to solve the above-described problem, a method for providing information on major depressive disorder according to an embodiment of the present invention is provided. The method for providing information according to an embodiment of the present invention is implemented by a processor and includes the steps of receiving brain wave data corresponding to a plurality of electrode channels of an individual, performing an evaluation based on an explainable artificial intelligence (XAI) algorithm on the brain wave data, and selecting a major channel among the plurality of electrode channels based on the evaluation result, wherein the major channel is defined as a channel with a high contribution to classifying whether or not a person has major depressive disorder.

[0023] According to a feature of the present invention, the step of performing the evaluation may include a step of evaluating the contribution of the received brainwave data in the process of the prediction model classifying normal or major depressive disorder by using an artificial neural network-based prediction model trained to classify normal or major depressive disorder using brainwave data as input.

[0024] According to another feature of the present invention, the receiving step may include receiving brain wave data corresponding to a plurality of electrode channels from a subject with major depressive disorder and a normal control subject. The evaluating step may include calculating a relevance score for each of the received brain wave data in a process of classifying a subject with major depressive disorder or a control subject using a predictive model. In this case, the selecting step may include selecting a major channel based on the relevance score.

[0025] According to another feature of the present invention, the selecting step may include a step of sorting the received brainwave data based on the contribution score, and a step of selecting a main channel from the received brainwave data within a predetermined rank.

[0026] According to another feature of the present invention, the method may further include, after the step of selecting a major channel, a step of training the prediction model to classify normal or depressive disorder by inputting brainwave data corresponding to the major channel, a step of receiving new brainwave data for the subject, and a step of using the trained prediction model to determine whether the subject has major depressive disorder based on the new brainwave data.

[0027] According to another feature of the present invention, the prediction model may be a model based on one of the algorithms Shallow ConvNet, EEGNet, EEGNet Fusion, and MI-EEGNet.

[0028] According to another feature of the present invention, the algorithm may be a Layer-wise relevance propagation (LRP) algorithm.

[0029] According to another feature of the present invention, the subject may be a male with no history of drug use.

[0030] According to another feature of the present invention, the main channel may be at least one of O1, CB1, PO3, POZ and PO4.

[0031] According to another feature of the present invention, the main channel may be an electrode channel corresponding to a frontal lobe region and an occipital lobe region.

[0032] To address the aforementioned challenges, a device for providing information on major depressive disorder according to an embodiment of the present invention is provided. The device comprises a communication unit configured to receive brain wave data corresponding to multiple electrode channels of an individual, and a processor functionally connected to the communication unit. The processor is configured to perform an evaluation of the brain wave data based on an explainable artificial intelligence algorithm, and to select a primary channel among the multiple electrode channels based on the evaluation results.

[0033] According to a feature of the present invention, the processor may be further configured to evaluate the contribution of the received brainwave data in the process of the prediction model classifying normal or major depressive disorder by using an artificial neural network-based prediction model trained to classify normal or major depressive disorder using brainwave data as input.

[0034] According to another feature of the present invention, the communication unit is configured to receive brain wave data corresponding to a plurality of electrode channels from a subject with major depressive disorder and a normal control subject, and the processor may be further configured to calculate a contribution score for each of the received brain wave data in the process of classifying a subject with major depressive disorder or a control subject using a prediction model, and to select a major channel based on the contribution score.

[0035] According to another feature of the present invention, the processor may be further configured to sort the received brainwave data based on the contribution score and select a primary channel from the received brainwave data within a predetermined ranking.

[0036] According to another feature of the present invention, the communication unit is further configured to receive new brain wave data for the subject, and the processor may be further configured to train a prediction model to classify normal or depressive disorder by inputting brain wave data corresponding to a main channel, and to determine whether the subject has major depressive disorder based on the new brain wave data using the trained prediction model.

[0037] In order to solve the above-described problem, a system for providing information on major depressive disorder according to an embodiment of the present invention is provided. The system includes an internal memory configured to store brain wave data corresponding to a plurality of electrode channels of an individual and an artificial intelligence algorithm capable of explaining the brain wave data, and a processing unit configured to access the internal memory, perform an evaluation based on the algorithm on the brain wave data, and select a major channel among the plurality of electrode channels based on the evaluation result.

[0038] Specific details of other embodiments are included in the detailed description and drawings.

[0039] The present invention can extract neurophysiological features of patients with major depressive disorder by applying XAI and confirm which features were used for diagnosis, thereby providing reliable information when assisting diagnosis.

[0040] In addition, the present invention can provide an information provision system integrated with XAI and an artificial neural network-based prediction algorithm, thereby enabling selection of major channels that have a significant influence on the diagnosis of each patient with major depressive disorder, thereby enabling customized diagnosis.

[0041] Furthermore, the present invention can increase practicality and reliability by providing an information provision system to which a channel selection algorithm is applied so as to enable the use of a minimum number of electrodes.

[0042] Furthermore, the present invention can screen for major channels that directly affect the diagnosis of not only major depressive disorder but also other mental illnesses, and thus has high potential for practical application when applied to computer-assisted diagnosis.

[0043] That is, the present invention can improve computational efficiency by reducing computational cost and time through a channel reduction strategy and reducing the possibility of overfitting, and can reduce the load on the user by minimizing the number of brainwave channels through the provision of a channel selection algorithm.

[0044] Accordingly, the present invention can overcome the limitations of analysis methods such as fMRI, which still have many limitations, such as providing information with low reliability, entailing high analysis costs, and spatial and temporal constraints.

[0045] Furthermore, the present invention can overcome the limitations of a method for diagnosing major depressive disorder based on questionnaires and surveys.

[0046] The effects according to the present invention are not limited to those exemplified above, and more diverse effects are included within the present invention.

[0047] FIG. 1 is a schematic diagram illustrating a system for providing information on major depressive disorder using brain wave data according to an embodiment of the present invention.

[0048] FIG. 2A is a block diagram showing the configuration of a user device according to one embodiment of the present invention.

[0049] FIG. 2b is a block diagram showing the configuration of a server of an information providing device according to one embodiment of the present invention.

[0050] Figure 3 illustrates a procedure of a method for providing information on major depressive disorder according to one embodiment of the present invention.

[0051] FIGS. 4a and 4b illustrate a procedure for selecting a major channel in a method for providing information on major depressive disorder according to one embodiment of the present invention.

[0052] FIG. 5 illustrates a prediction procedure for major depressive disorder based on a prediction model in a method for providing information on major depressive disorder according to one embodiment of the present invention.

[0053] Figures 6a and 6b illustrate learning data and evaluation procedures of a prediction model used in various embodiments of the present invention.

[0054] Figures 7a to 7e illustrate evaluation results for major channel selection based on contribution score.

[0055] Figures 8a to 8c illustrate topographic maps based on contribution scores for normal controls and major depressive disorder subjects.

[0056] The advantages and features of the present invention, and the methods for achieving them, will become clearer with reference to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms. These embodiments are provided solely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims. In connection with the description of the drawings, similar reference numerals may be used for similar components.

[0057] In this document, the expressions "has," "may have," "includes," or "may include" indicate the presence of a feature (e.g., a number, function, operation, or component such as a part), but do not exclude the presence of additional features.

[0058] In this document, the expressions "A or B," "at least one of A and / or B," or "one or more of A or / and B" can include all possible combinations of the listed items. For example, "A or B," "at least one of A and B," or "at least one of A or B" can all refer to cases where (1) at least one A is included, (2) at least one B is included, or (3) at least one A and at least one B are included.

[0059] The terms "first," "second," "first," or "second," as used herein, may describe various components, regardless of order and / or importance, and are only used to distinguish one component from another, without limiting the components. For example, a first user device and a second user device may represent different user devices, regardless of order or importance. For example, without departing from the scope of the rights set forth in this document, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component.

[0060] When it is said that a component (e.g., a first component) is "(operatively or communicatively) coupled with / to" or "connected to" another component (e.g., a second component), it should be understood that the component is directly coupled to the other component, or can be connected via another component (e.g., a third component). Conversely, when it is said that a component (e.g., a first component) is "directly coupled to" or "directly connected to" another component (e.g., a second component), it should be understood that no other component (e.g., a third component) exists between the first component and the other component.

[0061] The expression "configured to" as used herein can be used interchangeably with, for example, "suitable for," "having the capacity to," "designed to," "adapted to," "made to," or "capable of." The term "configured to" does not necessarily mean something that is "specifically designed to" in hardware. Instead, in some contexts, the expression "a device configured to" can mean that the device, together with other devices or components, is "capable of." For example, the phrase "a processor configured (or set) to perform A, B, and C" may mean a dedicated processor (e.g., an embedded processor) for performing those operations, or a general-purpose processor (e.g., a CPU or application processor) that can perform those operations by executing one or more software programs stored in a memory device.

[0062] The terms used in this document are used only to describe specific embodiments and may not be intended to limit the scope of other embodiments. The singular expression may include the plural expression unless the context clearly indicates otherwise. Terms used herein, including technical or scientific terms, may have the same meaning as commonly understood by those of ordinary skill in the art described in this document. Terms defined in general dictionaries among the terms used in this document may be interpreted as having the same or similar meaning in the context of the relevant technology, and shall not be interpreted in an idealized or overly formal sense unless explicitly defined in this document. In some cases, even if a term is defined in this document, it cannot be interpreted to exclude the embodiments of this document.

[0063] The individual features of the various embodiments of the present invention can be partially or wholly combined or combined with each other, and as can be fully understood by those skilled in the art, various technical connections and operations are possible, and each embodiment can be implemented independently of each other or can be implemented together in a related relationship.

[0064] For clarity in the interpretation of this specification, the terms used in this specification are defined below.

[0065] As used herein, the term "major depressive disorder" may refer to a mental disorder in which a person experiences one or more major depressive episodes without manic or hypomanic episodes, which is one of the mood disorders.

[0066] Meanwhile, within the present specification, major depressive disorder may encompass “major depressive disorder” and further “depressive disorder.”

[0067] As used herein, the term "subject" may refer to a subject suspected of having major depressive disorder. Preferably, the subject in this specification refers to a subject without a history of drug use. More preferably, the subject in this specification refers to a male subject without a history of drug use. However, this is not a limitation.

[0068] As used herein, the term "brainwave data" may refer to electroencephalogram (EEG) signal values ​​recorded by a sensor that detects brain waves. More specifically, brainwave data may be obtained by measuring electrical signals generated in the brain from two or more electrode channels.

[0069] As used herein, the term "multiple electrode channels" may include at least two channels selected from FP1, FPZ, FP2, AF3, AF4, F7, F5, F3, F1, FZ, F2, F4, F6, F8, FT7, FC5, FC3, FC1, FCZ, FC2, FC4, FC6, FT8, T7, C5, C3, C1, CZ, C2, C4, C6, T8, TP7, CP5, CP3, CP1, CPZ, CP2, CP4, CP6, TP8, P7, P5, P3, P1, PZ, P2, P4, P6, P8, PO7, PO5, PO3, POZ, PO4, PO6, PO8, CB1, O1, OZ, O2 and CB2.

[0070] According to a feature of the present invention, the brain wave data may be brain wave data obtained in a resting state in which the subject has his or her eyes closed, but is not limited thereto.

[0071] As used herein, the term "Explainable artificial intelligence (XAI)" refers to an algorithm that provides reasons for decisions made by the AI ​​in a way that humans can understand, thereby providing a basis for its output values.

[0072] More specifically, by applying XAI to a prediction model trained to classify whether a patient has major depressive disorder by inputting EEG data based on multiple electrode channels, it is possible to select key channels that have a high contribution to classifying major depressive disorder or normal.

[0073] As used herein, the term "predictive model" may mean a model trained to classify major depressive disorder based on neurophysiological feature data extracted from an individual's brain wave data.

[0074] According to a feature of the present invention, the prediction model may be a model configured to input brain wave data and output major depressive disorder or normal based on the brain wave data.

[0075] However, this is not limited to this, and the prediction model can be further configured to output 0 or 1 depending on whether the subject has major depressive disorder, or to output the probability of having major depressive disorder. Furthermore, the model can be configured to output more diverse classes based on the severity of major depressive disorder.

[0076] The prediction model may be based on at least one of a Shallow ConvNet, a Shallow-Deep ConvNet, an EfficientNet, an EEGNet, an EEGNet Fusion, a MI-EEGNet, and a MultiT-S ConvNet model.

[0077] In various embodiments of the present invention, the prediction model may be, but is not limited to, a Shallow ConvNet-based model.

[0078] As used herein, the term "primary channel" can be defined as a channel that contributes significantly to the classification of major depressive disorder, particularly in predictive model-based classification of major depressive disorder. In other words, a primary channel is a channel located in a brain region primarily selected for the diagnosis of major depressive disorder, and can refer to a channel with a large difference in activity (or frequency intensity) in EEG data.

[0079] According to a feature of the present invention, the main channel may be at least one of O1, CB1, PO3, POZ and PO4, but is not limited thereto, and the main channel may be different for each individual.

[0080] According to another feature of the present invention, the primary channel may be a channel located in the occipital lobe region.

[0081] According to a feature of the present invention, the main channel can be selected based on leave-one-out cross validation (LOOCV).

[0082] According to a feature of the present invention, the main channel can be selected based on the LRP (Layer-wise relevance propagation) algorithm.

[0083] For example, based on the LRP algorithm, a relevance score is calculated for EEG data corresponding to multiple channels used in the prediction model to classify major depressive disorder. Then, the electrode channel with a higher relevance score than other channels and a significant difference in the presence or absence of major depressive disorder can be determined as the primary channel.

[0084] Meanwhile, the decision on the main channel is not limited to the above-described method.

[0085]

[0086] Hereinafter, with reference to FIGS. 1 and 2a and 2b, a system for providing information on major depressive disorder, a user device, and a device for providing information on major depressive disorder using a device for providing information on major depressive disorder according to one embodiment of the present invention will be described.

[0087] FIG. 1 illustrates a system for providing information on major depressive disorder using a device for providing information on major depressive disorder according to one embodiment of the present invention. FIG. 2a illustrates an exemplary configuration of a user device for receiving information on major depressive disorder according to one embodiment of the present invention. FIG. 2b illustrates an exemplary configuration of a device for providing information on major depressive disorder according to one embodiment of the present invention.

[0088] First, referring to FIG. 1, the information providing system (1000) may be a system configured to provide information on major depressive disorder. At this time, the information providing system (1000) may be configured with a user device (100) that receives information on major depressive disorder, an EEG measuring device (200) that is configured to measure EEG by being in close contact with the user's scalp, and an information providing server (300) that generates information on major depressive disorder based on the received EEG data.

[0089] First, the user device (100) is an electronic device that provides a user interface for displaying information about major depressive disorder, and may include at least one of a smartphone, a tablet PC (personal computer), a laptop, and / or a PC.

[0090] The user device (100) can receive information on major depressive disorder from the information providing server (300) and display the received results through a display unit.

[0091] The information providing server (300) may include a general-purpose computer, laptop, and / or data server that performs various operations to generate information related to diagnosis from brain wave data provided from the brain wave measuring device (200). In this case, the information providing server (300) may be a device for accessing a web server that provides a web page or a mobile web server that provides a mobile website, but is not limited thereto.

[0092] More specifically, the information providing server (300) receives brainwave data from the brainwave measuring device (200), and performs an XAI-based evaluation on the received brainwave data to select channels that have a high contribution to classifying whether or not a person has major depressive disorder.

[0093] In various embodiments of the present invention, the information providing server (300) can use an artificial neural network-based prediction model trained to classify normal or major depressive disorder by inputting brain wave data, and evaluate the contribution of the received brain wave data in the process of the prediction model classifying normal or major depressive disorder.

[0094] The information providing server (300) can provide information related to the main channel and further classification results on whether major depression has occurred to the user device (100).

[0095] In this way, information provided from the information provision server (300) may be provided as a web page via a web browser installed on the user device (100), or may be provided in the form of an application or program. In various embodiments, such data may be provided in a form included in the platform in a client-server environment.

[0096] Next, with reference to FIGS. 2a and 2b, the components of the user device (100) and the information providing server (300) of the present invention will be described in detail.

[0097] First, referring to FIG. 2A, a user device (100) may include a memory interface (110), one or more processors (120), and a peripheral interface (130). Various components within the user device (100) may be connected by one or more communication buses or signal lines.

[0098] The memory interface (110) is connected to the memory (150) and can transmit various data to the processor (120). Here, the memory (150) can include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM, SRAM, ROM, EEPROM, PROM, network storage, cloud, and blockchain data.

[0099] In various embodiments, the memory (150) can store at least one of an operating system (151), a communication module (152), a graphical user interface (GUI) module (153), a sensor processing module (154), a telephone module (155), and an application module (156). Specifically, the operating system (151) can include instructions for processing basic system services and instructions for performing hardware operations. The communication module (152) can communicate with at least one of other devices, computers, and servers. The graphical user interface (GUI) module (153) can process a graphical user interface. The sensor processing module (154) can process sensor-related functions (e.g., processing voice input received using one or more microphones (192)). The telephone module (155) can process telephone-related functions. The application module (156) can perform various functions of the user application, such as electronic messaging, web browsing, media processing, navigation, imaging, and other processing functions. In addition, the user device (100) can store one or more software applications (156-1, 156-2) (e.g., an image enhancement application) associated with a type of service in the memory (150).

[0100] In various embodiments, the memory (150) can store a digital assistant client module (157) (hereinafter, DA client module), and accordingly, can store commands for performing client-side functions of the digital assistant and various user data (158).

[0101] Meanwhile, the DA client module (157) can obtain the user's voice input, text input, touch input, and / or gesture input through various user interfaces (e.g., I / O subsystem (140)) provided in the user device (100).

[0102] Additionally, the DA client module (157) can output data in audiovisual and tactile forms. For example, the DA client module (157) can output data consisting of a combination of at least two or more of voice, sound, notification, text message, menu, graphic, video, animation, and vibration. In addition, the DA client module (157) can communicate with a digital assistant server (not shown) using a communication subsystem (180).

[0103] In various embodiments, the DA client module (157) may collect additional information about the surroundings of the user device (100) from various sensors, subsystems, and peripheral devices to construct a context associated with the user input. For example, the DA client module (157) may provide context information along with the user input to a digital assistant server to infer the user's intent. Here, the context information that may accompany the user input may include sensor information, such as lighting, ambient noise, ambient temperature, images of the surroundings, videos, etc. As another example, the context information may include the physical state of the user device (100) (e.g., device orientation, device position, device temperature, power level, speed, acceleration, motion pattern, cellular signal strength, etc.). As another example, context information may include information related to the software state of the user device (100) (e.g., processes running on the user device (100), installed programs, past and current network activity, background services, error logs, resource usage, etc.).

[0104] In various embodiments, the memory (150) may include additional or deleted instructions, and further, the user device (100) may include additional configurations other than those illustrated in FIG. 2A, or may exclude some configurations.

[0105] The processor (120) can control the overall operation of the user device (100) and execute various commands to implement an interface that provides information on major depressive disorder by driving an application or program stored in the memory (150).

[0106] The processor (120) may correspond to a computing device such as a CPU (Central Processing Unit) or an AP (Application Processor). In addition, the processor (120) may be implemented in the form of an integrated chip (Integrated Chip (IC)) such as a SoC (System on Chip) in which various computing devices such as an NPU (Neural Processing Unit) are integrated.

[0107] The peripheral interface (130) can be connected to various sensors, subsystems, and peripheral devices to provide data so that the user device (100) can perform various functions. Here, it can be understood that the user device (100) performs a certain function as being performed by the processor (120).

[0108] The peripheral interface (130) can receive data from a motion sensor (160), a light sensor (light sensor) (161), and a proximity sensor (162), through which the user device (100) can perform orientation, light, and proximity detection functions, etc. For another example, the peripheral interface (130) can receive data from other sensors (163) (positioning system - GPS receiver, temperature sensor, biometric sensor), through which the user device (100) can perform functions related to the other sensors (163).

[0109] In various embodiments, the user device (100) may include a camera subsystem (170) connected to a peripheral interface (130) and an optical sensor (171) connected thereto, which may enable the user device (100) to perform various photographing functions, such as taking pictures and recording video clips.

[0110] In various embodiments, the user device (100) may include a communication subsystem (180) connected to a peripheral interface (130). The communication subsystem (180) may be comprised of one or more wired / wireless networks and may include various communication ports, radio frequency transceivers, and optical transceivers.

[0111] In various embodiments, the user device (100) includes an audio subsystem (190) coupled to a peripheral interface (130), the audio subsystem (190) including one or more speakers (191) and one or more microphones (192), such that the user device (100) can perform voice-activated functions, such as voice recognition, voice replication, digital recording, and telephony functions.

[0112] In various embodiments, the user device (100) may include an I / O subsystem (140) coupled to a peripheral interface (130). For example, the I / O subsystem (140) may control a touch screen (143) included in the user device (100) via a touch screen controller (141). As an example, the touch screen controller (141) may detect a user's contact and movement or cessation of contact and movement using any one of a plurality of touch sensing technologies, such as capacitive, resistive, infrared, surface acoustic wave technology, proximity sensor array, etc. As another example, the I / O subsystem (140) may control other input / control devices (144) included in the user device (100) via other input controller(s) (142). As an example, the other input controller(s) (142) may control one or more buttons, rocker switches, thumb-wheels, infrared ports, USB ports, and pointer devices such as a stylus.

[0113] Next, referring to FIG. 2b, the information providing server (300) may include a communication interface (310), a memory (320), an I / O interface (330), and a processor (340), each component of which may communicate with each other via one or more communication buses or signal lines.

[0114] The communication interface (310) can be connected to the user device (100) and the brain wave measurement device (200) via a wired / wireless communication network to exchange data. For example, the communication interface (310) can receive brain wave data from the brain wave measurement device (200) and transmit information about major depressive disorder (e.g., the type of major channel) to the user device (100).

[0115] Meanwhile, the communication interface (310) that enables transmission and reception of such data includes a communication port (311) and a wireless circuit (312), wherein the wired communication port (311) may include one or more wired interfaces, for example, Ethernet, Universal Serial Bus (USB), FireWire, etc. In addition, the wireless circuit (312) may transmit and receive data with an external device via an RF signal or an optical signal. In addition, the wireless communication may use at least one of a plurality of communication standards, protocols, and technologies, for example, GSM, EDGE, CDMA, TDMA, Bluetooth, Wi-Fi, VoIP, Wi-MAX, or any other suitable communication protocol.

[0116] The memory (320) can store various data used in the information providing server (300). For example, the memory (320) can store brainwave data received from the brainwave measuring device (200), or store major channels determined from the evaluation of brainwave data, an artificial neural network-based prediction model trained to predict clinical characteristics based on brainwave data, and a prediction result of whether or not major depression occurs.

[0117] In various embodiments, the memory (320) may include a volatile or non-volatile storage medium capable of storing various data, commands, and information. For example, the memory (320) may include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM, SRAM, ROM, EEPROM, PROM, network storage storage, cloud, and blockchain data.

[0118] In various embodiments, the memory (320) may store a configuration of at least one of an operating system (321), a communication module (322), a user interface module (323), and one or more applications (324).

[0119] An operating system (321) (e.g., an embedded operating system such as LINUX, UNIX, MAC OS, WINDOWS, VxWorks, etc.) may include various software components and drivers to control and manage general system operations (e.g., memory management, storage device control, power management, etc.) and may support communication between various hardware, firmware, and software components.

[0120] The communication module (322) can support communication with other devices through the communication interface (310). The communication module (322) can include various software components for processing data received by the wired communication port (311) or wireless circuit (312) of the communication interface (310).

[0121] The user interface module (323) can receive a user's request or input from a keyboard, touch screen, microphone, etc. through an I / O interface (330) and provide a user interface on the display.

[0122] The application (324) may include a program or module configured to be executed by one or more processors (330). Here, the application for providing information about major depressive disorder may be implemented on a server farm.

[0123] The I / O interface (330) can connect at least one of an input / output device (not shown) of the information providing server (300), such as a display, a keyboard, a touch screen, and a microphone, to the user interface module (323). The I / O interface (330) can receive user input (e.g., voice input, keyboard input, touch input, etc.) together with the user interface module (323) and process a command according to the received input.

[0124] The processor (340) is connected to a communication interface (310), a memory (320), and an I / O interface (330) to control the overall operation of the information providing server (300), and can perform various commands for providing information through an application or program stored in the memory (320).

[0125] The processor (340) may correspond to a computing device such as a CPU (Central Processing Unit) or an AP (Application Processor). In addition, the processor (340) may be implemented in the form of an integrated chip (Integrated Chip (IC)) such as a SoC (System on Chip) in which various computing devices are integrated. Alternatively, the processor (340) may include a module for calculating an artificial neural network model, such as an NPU (Neural Processing Unit).

[0126] In various embodiments, the processor (340) may be configured to provide information about major depressive disorder by calculating whether or not a person has major depressive disorder, or a probability of major depressive disorder or not major depressive disorder, using a predictive model learned to classify whether or not a person has major depressive disorder based on brain wave data corresponding to selected major channels.

[0127]

[0128] Hereinafter, information providing methods according to various embodiments of the present invention will be described with reference to FIGS. 3, 4a, 4b, and 5.

[0129] First, referring to FIG. 3, brain wave data of an individual is received according to a method for providing information on major depressive disorder according to an embodiment of the present invention (S310). Next, an evaluation is performed on multiple channels corresponding to the brain wave data based on XAI (S320). Then, based on the evaluation results, a major channel is selected (S330), and whether or not a patient has major depressive disorder is determined based on the major channel (S340).

[0130] According to a feature of the present invention, in the step (S310) of receiving brain wave data of an individual, brain wave data obtained in a resting state can be obtained. At this time, the individual may be a male individual with no history of drug use, but is not limited thereto.

[0131] According to another feature of the present invention, in the step (S310) of receiving brain wave data of an object, brain wave data measured from at least two electrode channels selected from among FP1, FPZ, FP2, AF3, AF4, F7, F5, F3, F1, FZ, F2, F4, F6, F8, FT7, FC5, FC3, FC1, FCZ, FC2, FC4, FC6, FT8, T7, C5, C3, C1, CZ, C2, C4, C6, T8, TP7, CP5, CP3, CP1, CPZ, CP2, CP4, CP6, TP8, P7, P5, P3, P1, PZ, P2, P4, P6, P8, PO7, PO5, PO3, POZ, PO4, PO6, PO8, CB1, O1, OZ, O2 and CB2 of the brain wave measuring device (200) It is received.

[0132] According to another feature of the present invention, preprocessing of the brain wave data can be performed after the step (S310) in which the brain wave data of the object is received.

[0133] For example, in the stage where preprocessing is performed, eye movement, noise due to movement, and noise generated from the external environment can be removed, and epoch processing can be performed for a predetermined length (e.g., a length of about 3 minutes).

[0134] Next, in the step (S320) where evaluations are performed on multiple channels, an evaluation is performed on the contribution to predicting the onset of major depressive disorder.

[0135] According to a feature of the present invention, in a step (S320) where an evaluation is performed on a plurality of channels, an evaluation is performed on an artificial neural network-based prediction model that is trained to classify normal or major depressive disorder using brain wave data as input, and an evaluation of the contribution of brain wave data received in the process of the prediction model classifying normal or major depressive disorder is performed.

[0136] At this time, referring to FIG. 4a, the prediction model may be a model trained to input EEG data for a major depression subject and a normal control subject (410), and then pass through a temporal convolution layer (420), a spatial filter layer (430), and a mean pooling layer (440), and finally output normal or major depression disorder (450).

[0137] According to another feature of the present invention, the step (S320) in which evaluation is performed on multiple channels can be performed based on an LRP algorithm that calculates the importance of each input feature by propagating the importance in a backward direction from the output of the model to the input.

[0138] More specifically, referring to Fig. 4b, in the forward pass, the input EEG data passes through the input node (Xi) and the activation layer, and moves from the output node to the output node. Then, in the backward propagation (Relevance Propagation), starting from the output node, the contribution score is calculated, and the weight (Z ij ) is assigned, the importance is propagated backwards. This process allows us to interpret how much each input node contributes to the prediction of major depressive disorder.

[0139] That is, through the step (S320) in which evaluation is performed on multiple channels, it can be confirmed which physiological characteristics are most important in predicting whether a prediction model will develop a major depressive disorder and how the importance is distributed.

[0140] Referring again to FIG. 3, in the step (S330) where the main channel is selected, the main channel, which is a channel with a high contribution to the classification of major depressive disorder based on the prediction model, can be determined.

[0141] According to a feature of the present invention, in the step (S330) where a main channel is selected, the received brainwave data is sorted based on the contribution score, and the main channel can be selected from the received brainwave data within a predetermined ranking.

[0142] According to various embodiments, in the step (S330) of selecting a primary channel, channels are sequentially selected in order of highest to lowest calculated contribution values, thereby retraining the prediction model. Then, an evaluation of diagnostic performance is performed, thereby selecting a primary channel.

[0143] Accordingly, in the step of selecting the main channel (S330), the main channel associated with the diagnostic performance of the prediction model can be selected, so that the prediction model can predict whether or not there is a major depressive disorder while maintaining the diagnostic performance without using the EEG data corresponding to all 62 electrode channels.

[0144] That is, at the step (S340) where it is determined whether a person has major depressive disorder based on a major channel, the onset of major depressive disorder can be determined using a prediction model in which major channel-based learning is performed.

[0145] More specifically, referring to FIG. 5, a major channel based on an LRP score is selected (S510), and learning is performed on a prediction model to output whether major depressive disorder has occurred by inputting EEG data corresponding to the major channel (S520), and prediction of whether major depressive disorder has occurred using new EEG data can be performed (S530).

[0146] For example, a predictive model might output 1 if the individual is at high risk for developing major depressive disorder, and 0 if the individual is at high risk for developing major depressive disorder and has a high probability of being normal.

[0147] According to the method for providing information on major depressive disorder based on XAI according to various embodiments of the present invention, it is possible to extract neurophysiological features of patients and confirm which features were used for diagnosis, so that reliable information can be provided when assisting diagnosis.

[0148] In addition, the present invention can provide an information provision system integrated with a prediction algorithm based on XAI and an artificial neural network, so that it is possible to select major channels that have a great influence on the diagnosis of each patient with major depressive disorder, thereby enabling customized diagnosis.

[0149] Furthermore, the present invention can increase practicality and reliability by providing an information provision system to which a channel selection algorithm is applied so as to enable the use of a minimum number of electrodes.

[0150] Evaluation: Screening of key channels and assessment of brain activation areas

[0151] Hereinafter, with reference to FIGS. 6a and 6b, FIGS. 7a to 7e, and FIGS. 8a to 8c, the results of selecting major channels of a device for providing information on major depressive disorder according to one embodiment of the present invention will be described. FIGS. 6a to 6b illustrate learning data and an evaluation procedure of a prediction model used in various embodiments of the present invention. FIGS. 7a to 7e illustrate evaluation results for selecting major channels based on contribution scores. FIGS. 8a to 8c illustrate topographic maps based on contribution scores for a normal control group and major depressive disorder subjects.

[0152] First, referring to Figure 6a, EEG data of 40 individuals with major depressive disorder (MDD) and 41 healthy controls (HC) were used in this evaluation. All subjects were male, and individuals with MDD could be male individuals with MDD who had no history of medication use.

[0153] At this time, a Shallow ConvNet model based on a convolutional neural network was used to classify the two groups, and evaluation based on leave-one-out cross-validation was performed.

[0154] Furthermore, 80 brain wave data, excluding one brain wave data used as evaluation data (test data), were used as learning and testing data, and the ratio of learning data and testing data was 4:1.

[0155] Referring to Figure 6b, the contribution scores for each channel of the test data used by the Shallow ConvNet prediction model to classify the two groups were calculated using the LRP algorithm. Channels with calculated contribution values ​​were sequentially selected from the highest to the lowest and applied to the training of the prediction model. The diagnostic performance was then confirmed using the evaluation data. This allows for the determination of which brain regions are the primary channels when classifying EEG data from subjects with major depressive disorder and healthy control subjects.

[0156] Referring to Figures 7a and 7b, the diagnostic performance was verified by reducing the number of electrode channels. When 5 channels were used, the diagnostic accuracy increased sharply to 92.59% compared to when fewer channels were used. Furthermore, it was shown that a high diagnostic performance of over 90% was maintained even when 5 or more electrode channels were used.

[0157] Next, referring to Fig. 7c, the results of counting the number of overlapping channels among the five channels selected as the main channels according to the contribution score for 81 individuals are shown. More specifically, when the five channels were selected, the five channels with the largest number of overlapping selections were 'O1', 'CB1', 'PO3', 'POZ', and 'PO4', and these channels are occipital lobe channels and may be the main channels that contribute highly to the classification of major depressive disorder in the prediction model.

[0158] Referring further to FIGS. 7d ​​and 7e, the frequency of channels selected as primary channels according to the number of channels is shown.

[0159] For example, for the FPZ channel, the frequency value is “3”, which may mean that when the number of channels is 1, FPZ was selected as the main channel with a high contribution score in 3 out of 81 individual data.

[0160] That is, when five channels were used, 'PO3', 'POZ', 'PO4', 'CB1', and 'O1', which showed high selection frequencies in all objects, could be determined as the main channels.

[0161] These results may imply that the prediction model can predict major depressive disorder while maintaining high diagnostic performance using only the major channels of 'PO3', 'POZ', 'PO4', 'CB1', and 'O1' among EEG data.

[0162] Meanwhile, the types of key channels are not limited to this and may vary from individual to individual. That is, based on the information provision methods according to various embodiments of the present invention, it is possible to select key channels that influence the diagnosis of each individual with major depressive disorder, thereby enabling customized diagnosis.

[0163] Next, referring to Figures 8a to 8c, the contribution score-based topography for normal controls and major depressive disorder subjects is depicted.

[0164] First, referring to Figure 8a, the average contribution scores of subjects with major depressive disorder and normal controls using the evaluation data were checked, and the areas where the contribution scores of each group were largely distributed were found to be contrasting. In particular, subjects with major depressive disorder had a large distribution of contribution scores in the right hemisphere frontal and occipital regions (or corresponding channels), while in contrast, normal subjects had a large distribution of contribution scores in the left hemisphere frontal and occipital lobes.

[0165] Next, referring to Figure 8b, the results of the neurophysiological analysis of male subjects with major depressive disorder (MDD) using contribution scores calculated using the LRP method are depicted. An independent t-test was applied to examine differences in contribution scores between subjects with MDD and the normal group. Subsequently, a correlation analysis was performed between contribution scores and symptom scores (HAM-D (depression scale), HAM-A (anxiety scale)) for channels with significant differences.

[0166] At this time, in (a) of 8b, red indicates an area where the contribution scores of major depressive disorder subjects are significantly greater than the normal area, and blue indicates a significantly smaller area. That is, the major depressive disorder group showed a significantly higher contribution score in the FPz (t-value = 2.0072, p-value = 0.0481) channel, but showed a significantly reduced contribution score than the normal area in the FC1 and FCz (FC1 - t-value = -2.0262, p-value = 0.0461; FCz - t-value = -2.3037, p-value = 0.0239) channels. At this time, when examining the relationship between the contribution score and symptom score (depression and anxiety) of major depressive disorder subjects in (b) of 8b, in the case of FC1, there appears to be a significant correlation between the contribution score and symptom score (HAM-D depression scale).

[0167] According to the above results, differences in contribution scores in the fronto-central area were observed between the major depressive disorder group and the normal control group. Furthermore, the contribution score of FC1 showed a significant negative correlation with the patients' depression score (HAM-D), which may indicate that contribution scores decreased as the patients' symptoms worsened. In other words, the above results may suggest that features related to the neural characteristics of major depressive disorder were extracted from the deep learning model.

[0168] Moreover, contribution scores should be investigated to explain the neurophysiological characteristics of major depressive disorder individuals, and these values ​​may serve as promising biomarkers for understanding neuropathology.

[0169] Next, referring to Figure 8c, the channel locations according to the number of channels containing the main channel exhibit a similar pattern to the contribution score distribution described above. Specifically, the EEG data from subjects with major depressive disorder exhibited stronger frequency power in the occipital lobe region than in normal control subjects. In this regard, most of the main channels with high contribution scores appear to be located in the occipital lobe region.

[0170] These results may imply that EEG data acquired from the main channels in the occipital lobe region contribute significantly to determining the onset of major depressive disorder in the prediction model.

[0171] This confirmed the correlation between neurophysiological information in brain regions with significant differences between the two groups and the diagnostic performance of the prediction model.

[0172] That is, in various embodiments of the present invention, the XAI-based information providing device can be applied to a computer-assisted diagnosis system, and practicality and reliability can be increased as the minimum number of electrodes in an area directly affecting the diagnosis of a major depressive disorder subject can be used through a channel selection algorithm.

[0173]

[0174] Although the embodiments of the present invention have been described in more detail with reference to the attached drawings, the present invention is not necessarily limited to these embodiments, and various modifications may be implemented without departing from the technical spirit of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical spirit of the present invention, but to explain it, and the scope of the technical spirit of the present invention is not limited by these embodiments. Therefore, it should be understood that the embodiments described above are illustrative in all aspects and not restrictive. The protection scope of the present invention should be interpreted by the claims below, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the present invention.

[0175] [National Research and Development Project Supporting This Invention]

[0176] [Project ID] 1425173283

[0177] [Assignment Number] S3197996

[0178] [Ministry Name] Ministry of SMEs and Startups

[0179] [Name of Project Management (Specialist) Agency] Small and Medium Business Technology Information Promotion Agency

[0180] [Research Project Name] Startup Growth Technology Development

[0181] [Research Project Title] Development of a Deep Learning Model for Biomarker Derivation Based on EEG Data for the Precise Diagnosis of Major Depressive Disorder

[0182] [Name of the project performing organization] Bwave Co., Ltd.

[0183] Research Period: October 1, 2021 - September 30, 2023

Claims

1. A method for providing information on major depressive disorder implemented by a processor, A step of receiving brain wave data corresponding to multiple electrode channels of an object; A step of performing an evaluation based on an explainable artificial intelligence (XAI) algorithm for the above brainwave data, and A method for providing information on major depressive disorder, comprising a step of selecting a major channel among the plurality of electrode channels based on an evaluation result, wherein the major channel is defined as a channel with a high contribution to classifying whether or not a patient has major depressive disorder.

2. In paragraph 1, The steps for performing the above evaluation are: A method for providing information on major depressive disorder, comprising a step of evaluating the contribution of the received brainwave data to the process of classifying normal or major depressive disorder by using an artificial neural network-based predictive model trained to classify normal or major depressive disorder using brainwave data as input.

3. In paragraph 2, The above receiving step is, A step of receiving brain wave data corresponding to the plurality of electrode channels from a subject with major depressive disorder and a normal control subject, The steps for performing the above evaluation are: Including a step of calculating a contribution score for each of the brain wave data received in the process of classifying the major depressive disorder subject or the control subject using the above prediction model, The above selection step is, A method for providing information on major depressive disorder, comprising a step of selecting the main channel based on the contribution score.

4. In paragraph 3, The above selection step is, A step of aligning the received brainwave data based on the contribution score, and A method for providing information on major depressive disorder, comprising the step of selecting the main channel from the received brain wave data within a predetermined rank.

5. In paragraph 2, After the step of selecting the above main channels, A step of training the prediction model to classify normal or depressive disorder by inputting brain wave data corresponding to the main channel; A step of receiving new brain wave data for the above object, and A method for providing information on major depressive disorder, further comprising a step of determining whether the subject has major depressive disorder based on the new brain wave data using the learned prediction model.

6. In paragraph 2, A method for providing information on major depressive disorder, wherein the above prediction model is a model based on one of the algorithms of Shallow ConvNet, EEGNet, EEGNet Fusion, and MI-EEGNet.

7. In paragraph 1, The above algorithm is a method for providing information on major depressive disorder, which is a layer-wise relevance propagation (LRP) algorithm.

8. In paragraph 1, A method for providing information on major depressive disorder, wherein the subject is a male with no history of drug use.

9. In paragraph 1, A method for providing information on major depressive disorder, wherein the major channel is at least one of O1, CB1, PO3, POZ and PO4.

10. In paragraph 1, The above main channels are: A method for providing information on major depressive disorder, the electrode channels corresponding to the frontal lobe region and the occipital lobe region.

11. A communication unit configured to receive brain wave data corresponding to multiple electrode channels of an object, and including a processor functionally connected to the above communication unit, The above processor, Perform an evaluation based on an artificial intelligence algorithm that can explain the above brainwave data, A device for providing information on major depressive disorder, configured to select a major channel among the plurality of electrode channels based on the evaluation results, wherein the major channel is defined as a channel with a high contribution to classifying whether or not a patient has major depressive disorder.

12. In paragraph 11, The above processor, A device for providing information on major depressive disorder, further configured to evaluate the contribution of the received brainwave data in the process of classifying normal or major depressive disorder using an artificial neural network-based predictive model trained to classify normal or major depressive disorder using brainwave data as input.

13. In paragraph 12, The above communication department, configured to receive brain wave data corresponding to the plurality of electrode channels from a subject with major depressive disorder and a normal control subject, The above processor, In the process of classifying the major depressive disorder subject or the control subject using the above prediction model, a contribution score is calculated for each of the received brain wave data, and A device for providing information on major depressive disorder, further configured to select the main channel based on the contribution score.

14. In paragraph 13, The above processor, Based on the above contribution score, the received brainwave data is sorted, A device for providing information on major depressive disorder, further configured to select the main channel from the received brain wave data within a predetermined rank.

15. In paragraph 12, The above communication department, Further configured to receive new brainwave data for the above object, The above processor, The prediction model is trained to classify normal or depressive disorder by inputting brain wave data corresponding to the above main channels, A device for providing information on major depressive disorder, further configured to determine whether the subject has major depressive disorder based on the new brain wave data using the learned prediction model.

16. In paragraph 12, A device for providing information on major depressive disorder, wherein the above prediction model is a model based on one of the algorithms of Shallow ConvNet, EEGNet, EEGNet Fusion, and MI-EEGNet.

17. In paragraph 11, The above algorithm is a device for providing information on major depressive disorder, which is an LRP algorithm.

18. In paragraph 11, The above entity is a device for providing information on major depressive disorder, and is a male with no history of drug use.

19. In paragraph 11, A device for providing information on major depressive disorder, wherein the above major channel is at least one of O1, CB1, PO3, POZ and PO4.

20. In paragraph 11, The above main channels are: A device for providing information on major depressive disorder, the channels corresponding to the frontal lobe region and the occipital lobe region.

21. EEG data corresponding to multiple electrode channels of the object, and An internal memory configured to store an artificial intelligence algorithm that can explain the above brain wave data, and It is configured to access the above internal memory, A system for providing information on major depressive disorder, comprising a processing unit configured to perform an algorithm-based evaluation on the brainwave data and select a major channel among the plurality of electrode channels based on the evaluation result, wherein the major channel is defined as a channel with a high contribution to classifying whether or not a patient has major depressive disorder.

Images