Device, method, and computer program for performing clinical trial on basis of electrocardiogram data

Abstract

The present disclosure provides a device, a method, and a computer program for performing a clinical trial. The method according to one embodiment of the present disclosure comprises the steps of: acquiring electrocardiogram data of a clinical trial applicant; and determining whether the applicant is suitable for the clinical trial on the basis of the acquired electrocardiogram data through a pre-trained neural network model.

Description

Device, method and computer program for performing clinical trials based on electrocardiogram data

[0001] The present disclosure relates to deep learning technology in the medical field, and more particularly, to a device, method, and program for performing clinical trials based on electrocardiogram data.

[0002] With recent advancements in information and communication technology and artificial intelligence algorithms, the medical field is actively researching technologies that analyze or predict diseases based on a patient's biosignals. Among these, electrocardiograms (ECGs) are widely used for the diagnosis and monitoring of heart disease, as they allow for noninvasive measurement of the heart's electrophysiological activity. In particular, the six-lead ECG, offering portability and convenient measurement while providing key cardiac function information, is being utilized in non-hospital settings as well.

[0003] Meanwhile, ensuring the cardiac safety of subjects in clinical trials for new drug development is crucial. Failure to accurately assess the effects of a specific drug or treatment on cardiac function can lead to unexpected cardiac abnormalities during the trial, potentially leading to clinical failure or serious adverse events. Despite this, current clinical trials primarily rely on post-hoc responses based on blood tests or imaging diagnostics, making early prediction and real-time response difficult.

[0004] Accordingly, there is a growing demand for technological means that can precisely assess the cardiac condition of clinical trial subjects and proactively identify the risk of adverse reactions by analyzing electrocardiogram data using artificial intelligence.

[0005] The present disclosure, conceived in response to the aforementioned background technology, aims to provide a device, method, and computer program for performing clinical trials based on electrocardiogram data. However, the problems addressed by the present disclosure are not limited to those mentioned above, and other problems not mentioned will be clearly understood based on the description below.

[0006] A method for performing a clinical trial based on electrocardiogram data, the method comprising at least one processor, the method comprising the steps of: obtaining electrocardiogram data of a clinical trial applicant; and identifying a risk group for the clinical trial of the applicant based on the obtained electrocardiogram data through a pre-trained neural network model.

[0007] Alternatively, the step of determining eligibility for the clinical trial includes the step of predicting whether the applicant develops cardiotoxicity during the clinical trial based on the acquired electrocardiogram data through the learned neural network model, and the step of identifying the applicant's risk group for the clinical trial based on the result of the prediction of the development of cardiotoxicity.

[0008] Alternatively, the pre-trained neural network model includes a first neural network model trained to determine whether the applicant has impaired left ventricular systolic function based on the electrocardiogram data, and a second neural network model trained to determine whether the applicant has impaired left ventricular diastolic function based on the electrocardiogram data.

[0009] Alternatively, the step of predicting whether cardiac toxicity occurs during the course of the clinical trial includes the step of inputting the acquired electrocardiogram data into the first neural network model to obtain a first score for the possibility of left ventricular systolic dysfunction of the applicant, and the step of predicting whether cardiac toxicity occurs during the course of the clinical trial based on the obtained first score.

[0010] Alternatively, the step of predicting whether cardiac toxicity occurs during the course of the clinical trial includes the step of inputting the acquired electrocardiogram data into the second neural network model to obtain a second score for the possibility of left ventricular diastolic dysfunction of the applicant, and the step of predicting whether cardiac toxicity occurs during the course of the clinical trial based on the obtained second score.

[0011] Alternatively, the step of predicting whether cardiac toxicity occurs during the course of the clinical trial includes the step of identifying a risk group for cardiac toxicity of the applicant based on a first score for the possibility of left ventricular systolic dysfunction of the applicant obtained from the first neural network model and a second score for the possibility of left ventricular diastolic dysfunction of the applicant obtained from the second neural network model, thereby predicting whether cardiac toxicity occurs during the course of the clinical trial of the applicant.

[0012] Alternatively, the step of predicting whether cardiotoxicity occurs during the course of the clinical trial includes a step of predicting that the applicant falls into a high-risk group for cardiotoxicity if the first score is equal to or greater than a first reference value and the second score is equal to or greater than a second reference value, and the step of determining whether the applicant is suitable for the clinical trial includes a step of determining that the applicant is unsuitable for the clinical trial if the applicant is predicted to fall into a high-risk group for cardiotoxicity.

[0013] Alternatively, the step of predicting whether cardiac toxicity occurs during the course of the clinical trial includes a step of identifying the applicant as belonging to a high-risk group for cardiac toxicity if the first score is equal to or greater than a first reference value and the second score is equal to or greater than a second reference value, and if the applicant is identified as belonging to a high-risk group for cardiac toxicity, a step of setting the dosage of the drug administered to the applicant to be lower than that of a applicant belonging to a lower-risk group than the high-risk group.

[0014] Alternatively, if the applicant is determined to be suitable for the clinical trial based on the first score and the second score, the step of selecting the applicant as a clinical trial subject is included, and the step of obtaining electrocardiogram data from the applicant during the course of the clinical trial is included, and determining whether or not to discontinue the clinical trial for the clinical subject based on the obtained electrocardiogram data through the pre-trained neural network model is included.

[0015] Alternatively, the method includes a step of adjusting the dosage of a drug administered to the clinical subject based on the acquired electrocardiogram data through the learned neural network model.

[0016] A computing device for performing a clinical trial based on electrocardiogram data for solving the aforementioned task includes a processor including at least one core and a memory including program codes executable by the processor; wherein the processor acquires electrocardiogram data of a clinical trial applicant and determines whether the applicant is suitable for the clinical trial based on the acquired electrocardiogram data through a pre-learned neural network model.

[0017] A computer program stored in a computer-readable storage medium for solving the aforementioned problem, wherein the computer program, when executed on one or more processors, causes the computer program to perform an operation of performing a clinical trial based on electrocardiogram data, the operation including an operation of acquiring electrocardiogram data of a clinical trial applicant and an operation of determining whether the applicant is suitable for the clinical trial based on the acquired electrocardiogram data through a pre-learned neural network model.

[0018] According to one embodiment of the present disclosure, the risk of cardiac toxicity in clinical trial applicants can be predicted in advance based on electrocardiogram (ECG) data, and thus, the suitability for the clinical trial can be quantitatively determined. Furthermore, by repeatedly monitoring ECG data during the clinical trial, risk signals resulting from changes in the subject's cardiac condition can be detected in real time, and based on these results, the decision to discontinue the clinical trial or automatically adjust drug administration conditions can be made. This can enhance the safety and reliability of clinical trials and minimize the risk of adverse cardiac events.

[0019] FIG. 1 is a configuration diagram of a computing device according to an embodiment of the present disclosure.

[0020] FIG. 2 is a flowchart of a method for performing a clinical trial based on electrocardiogram data according to one embodiment of the present disclosure.

[0021] FIG. 3 is an exemplary diagram of a method for determining whether a clinical trial applicant is suitable for a clinical trial based on electrocardiogram data according to one embodiment of the present disclosure.

[0022] FIG. 4 is a flowchart of a method for predicting whether a volunteer develops cardiac toxicity and determining whether the volunteer is suitable for a clinical trial according to one embodiment of the present disclosure.

[0023] FIG. 5 is an exemplary diagram of a method for predicting whether a volunteer will develop cardiac toxicity by determining left ventricular systolic dysfunction and left ventricular diastolic dysfunction according to one embodiment of the present disclosure.

[0024] FIG. 6 is an exemplary diagram of a method for determining the status of a clinical trial applicant using a plurality of sub-neural network models included in a first neural network model trained to produce a score corresponding to the possibility of left ventricular systolic dysfunction according to one embodiment of the present disclosure.

[0025] FIG. 7 is a block diagram of a computing device according to another embodiment of the present disclosure.

[0026] Below, embodiments of the present disclosure are described in detail with reference to the attached drawings so that those skilled in the art can easily implement the present disclosure. The embodiments presented in this disclosure are provided to enable those skilled in the art to utilize or implement the contents of the present disclosure. Accordingly, various modifications to the embodiments of the present disclosure will be apparent to those skilled in the art. That is, the present disclosure may be implemented in various different forms and is not limited to the embodiments described below.

[0027] Throughout the specification of this disclosure, identical or similar drawing numbers refer to identical or similar components. Furthermore, for the purpose of clearly describing the disclosure, drawing numbers for parts in the drawings that are not relevant to the description of the disclosure may be omitted.

[0028] The term "or" as used herein is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified herein or clear from context, "X employs A or B" should be understood to mean either of its natural inclusive permutations. For example, unless otherwise specified herein or clear from context, "X employs A or B" can be interpreted to mean either X employs A, X employs B, or X employs both A and B.

[0029] The term "and / or" as used herein should be understood to refer to and include all possible combinations of one or more of the related concepts listed.

[0030] The terms "comprises" and / or "comprising" as used herein should be understood to mean the presence of certain features and / or components. However, it should be understood that the terms "comprises" and / or "comprising" do not exclude the presence or addition of one or more other features, other components, and / or combinations thereof.

[0031] Unless otherwise specified in this disclosure or unless the context makes it clear that the singular form is being referred to, the singular should generally be construed to include “one or more.”

[0032] The term "Nth (N is a natural number)" used in this disclosure can be understood as an expression used to mutually distinguish components of this disclosure based on a predetermined standard such as a functional perspective, a structural perspective, or convenience of explanation. For example, components performing different functional roles in this disclosure can be distinguished as a first component or a second component. However, components that are substantially the same within the technical spirit of this disclosure but must be distinguished for convenience of explanation may also be distinguished as a first component or a second component.

[0033] The term "acquisition" as used in this disclosure may be understood to mean not only receiving data through a wired or wireless communication network with an external device or system, but also generating data in an on-device form.

[0034] Meanwhile, the term "module" or "unit" used in the present disclosure can be understood as a term referring to an independent functional unit that processes computing resources, such as a computer-related entity, firmware, software or a part thereof, hardware or a part thereof, or a combination of software and hardware. At this time, the "module" or "unit" may be a unit composed of a single element, or a unit expressed as a combination or set of multiple elements. For example, as a narrow concept, a "module" or "unit" may refer to a hardware element of a computing device or a set thereof, an application program that performs a specific function of software, a processing process implemented through software execution, or a set of instructions for program execution, etc. In addition, as a broad concept, a "module" or "unit" may refer to the computing device itself that constitutes the system, or an application running on the computing device, etc. However, since the above-described concept is only an example, the concept of “module” or “part” may be defined in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0035] The term "model" as used herein may be understood as a system implemented using mathematical concepts and language to solve a specific problem, a set of software units to solve a specific problem, or an abstract model of a processing process to solve a specific problem. For example, a neural network "model" may refer to the entire system implemented as a neural network that has problem-solving capabilities through learning. In this case, the neural network can have problem-solving capabilities by optimizing the parameters connecting nodes or neurons through learning. A neural network "model" may include a single neural network or a set of neural networks that are a combination of multiple neural networks.

[0036] The term "data" used in this disclosure may include "images," signals, and the like. The term "image" used in this disclosure may refer to multidimensional data composed of discrete image elements. In other words, "image" may be understood as a term referring to a digital representation of an object visible to the human eye. For example, "image" may refer to multidimensional data composed of elements corresponding to pixels in a two-dimensional image. "Image" may refer to multidimensional data composed of elements corresponding to voxels in a three-dimensional image.

[0037] The explanation of the above terms is intended to aid understanding of the present disclosure. Therefore, unless explicitly stated as limiting the contents of the present disclosure, it should be noted that the above terms are not intended to limit the technical ideas of the contents of the present disclosure.

[0038] FIG. 1 is a configuration diagram of a computing device according to an embodiment of the present disclosure.

[0039] Referring to FIG. 1, a computing device (100) according to an embodiment of the present disclosure may include a processor (110), a memory (120), and a communication interface (130). However, FIG. 1 is merely an example, and the computing device (100) may include other components for implementing a computing environment. In addition, only some of the components disclosed in FIG. 1 may be included in the computing device (100).

[0040] A processor (110) according to an embodiment of the present disclosure may be understood as a configuration unit including hardware and / or software for performing computing operations. For example, the processor (110) may read a computer program to perform data processing for machine learning. The processor (110) may process computational processes such as processing input data for machine learning, feature extraction for machine learning, and error calculation based on backpropagation. The processor (110) for performing such data processing may include a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), or a field programmable gate array (FPGA). The above-described type of processor (110) is only one example, and thus, the type of processor (110) may be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0041] The processor (110) is connected to other components of the computing device (100) (i.e., memory (120) and communication interface (130)) and controls the overall operation of the computing device (100).

[0042] The memory (120) according to one embodiment of the present disclosure may be understood as a configuration unit including hardware and / or software for storing and managing data processed in the computing device (100). That is, the memory (120) may store any type of data generated or determined by the processor (110) and any type of data received by the communication interface (130). For example, the memory (120) may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory, a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. In addition, the memory (120) may also include a database system that controls and manages data in a predetermined system. The type of memory (120) described above is only one example, and thus the type of memory (120) can be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0043] The memory (120) can structure and organize and manage data, combinations of data, and program codes executable by the processor (110) required for the processor (110) to perform operations. For example, the memory (120) can store a neural network model trained to output data that can determine a risk group for a clinical trial or suitability for a clinical trial of an applicant based on electrocardiogram data (20) received through a communication interface (130) to be described later, and can store program codes that operate to perform learning of the neural network model, program codes that operate the neural network model to receive electrocardiogram data (20) and perform inference according to the purpose of use of the computing device (100) (for example, the purpose of calculating a score for a specific disease), and output or processed data generated as the program code is executed. In addition, the memory (120) can also store learning data for training the neural network model.

[0044] A communication interface (130) according to an embodiment of the present disclosure may be understood as a component that transmits and receives data through any known wired or wireless communication system. For example, the communication interface (130) may perform data transmission and reception using a wired or wireless communication system such as a local area network (LAN), wideband code division multiple access (WCDMA), long term evolution (LTE), wireless broadband internet (WiBro), fifth generation mobile communication (5G), ultrawide-band, ZigBee, radio frequency (RF) communication, wireless LAN, wireless fidelity, near field communication (NFC), or Bluetooth. Since the above-described communication systems are only examples, the wired and wireless communication system for data transmission and reception of the communication interface (130) may be applied in various ways other than the above-described examples.

[0045] The communication interface (130) can receive data necessary for the processor (110) to perform calculations through wired or wireless communication with any system or any client, etc. In addition, the communication interface (130) can transmit data generated through calculations of the processor (110) through wired or wireless communication with any system or any client, etc. For example, the communication interface (130) can receive the user's electrocardiogram data (20) through communication with a biosignal measuring device (200). The communication interface (130) can transmit output data of the neural network model, intermediate data derived from the calculation process of the processor (110), processed data, etc. through communication with the aforementioned database, server, or computing device, etc.

[0046] FIG. 2 is a flowchart illustrating a method for conducting a clinical trial based on electrocardiogram data according to an embodiment of the present disclosure. FIG. 3 is an exemplary diagram illustrating a method for determining whether a clinical trial applicant is eligible for a clinical trial based on electrocardiogram data according to an embodiment of the present disclosure.

[0047] Referring to FIG. 2, according to one embodiment of the present disclosure, a processor (110) obtains electrocardiogram data (20) of a clinical trial applicant (S210). Here, a clinical trial applicant refers to a clinical trial applicant who has applied to be a participant in a specific clinical trial, but whose participation in the clinical trial has not been determined. The processor (110) can obtain electrocardiogram data (20) of the clinical trial applicant from a biosignal measuring device connected through a communication interface. At this time, the electrocardiogram data (20) may include 12-lead electrocardiogram data, 6-lead electrocardiogram data, 1-lead electrocardiogram data, etc. The electrocardiogram data (20) may be data obtained by sampling each electrocardiogram signal measured through the biosignal measuring device at 500 points per second (Sampling rate = 500 Hz) and dividing the data into 8-second-long sections.

[0048] A biosignal measuring device may be a device that measures an electrocardiogram of a clinical trial volunteer to obtain electrocardiogram data (20). The biosignal measuring device may include a plurality of electrodes, and may obtain electrocardiogram data (20) by attaching the plurality of electrodes to the body of the clinical trial volunteer. For example, the biosignal measuring device may include three electrodes, and may be a device that simultaneously attaches the three electrodes to the body of the clinical trial volunteer, thereby obtaining electrocardiogram data (20) of Lead ° and Lead ± based on the potential difference between the three electrodes, and then calculates electrocardiogram data (20) of the remaining Leads using the obtained electrocardiogram data (20) (i.e., the electrocardiogram data (20) of Lead ° and Lead ±). Alternatively, the biosignal measuring device may be a device (e.g., a smart watch, etc.) that includes two electrodes, and obtains electrocardiogram data (20) corresponding to the attachment positions included in the electrocardiogram data (20) by attaching the two electrodes to specific positions on the body of the clinical trial volunteer.

[0049] However, the present invention is not limited thereto, and the processor (110) may also directly acquire electrocardiogram data (20) for a clinical trial applicant using a sensing unit including a plurality of electrodes included in the computing device. The above-described description applies equally to the method of directly acquiring electrocardiogram data (20), so a detailed description thereof will be omitted.

[0050] Meanwhile, the processor (110) can identify the risk group of the applicant for a clinical trial based on the electrocardiogram data (20) acquired through the pre-trained neural network model (S220). Here, the clinical trial is a procedure performed to verify the safety and efficacy of a specific drug or treatment, and may include a series of medical research processes that precisely observe the physiological responses of clinical trial participants selected from among the applicants and evaluate the possibility of adverse reactions in advance. The risk group for the clinical trial may be a risk grade for the clinical trial classified based on the status of the applicant identified using the pre-trained neural network model. In particular, it may be a risk level or risk grade of the applicant predicted based on the status of the applicant identified when the applicant participated in the clinical trial.

[0051] By inputting electrocardiogram data (20) into a pre-trained neural network model, data indicating the condition of the applicant can be obtained as an indicator for determining the risk group for the clinical trial of the applicant (or for determining whether the applicant is suitable for the clinical trial).

[0052] To this end, a neural network model can be trained based on training data including labeled data indicating the status of a clinical trial applicant, with electrocardiogram data (20) as input data. In particular, the label may relate to the presence or absence of cardiotoxicity or heart disease in the subject whose electrocardiogram data was measured. Heart disease may include left ventricular systolic dysfunction (LVSD) and left ventricular diastolic dysfunction (LVDD).

[0053] The processor (110) may input a plurality of electrocardiogram data (20) included in the input data into a neural network model, and may calculate a loss function based on the difference between the output value (e.g., a score on cardiotoxicity or a score on heart disease) for each electrocardiogram data (20) of the neural network model and the label (e.g., presence or absence of cardiotoxicity, presence or absence of left ventricular systolic dysfunction, presence or absence of left ventricular diastolic dysfunction) assigned to the label data. The loss function may be defined as a cross entropy loss or an objective function that optimizes the balance between precision and recall. Based on the calculated loss function, the processor (110) may adjust the weights of the model through backpropagation. By repeating this process, the processor (110) may improve the classification performance of the neural network model with respect to the condition of the subject, and ultimately obtain a neural network model trained to determine the possibility of cardiotoxicity or heart disease of a clinical trial applicant based on the electrocardiogram data (20).

[0054] Neural network models can be implemented with multi-layer perceptron (MLP), convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks, or residual neural networks (ResNet) structures.

[0055] Referring to FIG. 3, the processor (110) inputs the acquired electrocardiogram data (20) into a pre-trained neural network model (600) to obtain output information regarding the status of the clinical trial applicant (10), and can determine the status of the clinical trial applicant (10) based on the obtained output information. In addition, the processor (110) can determine whether the applicant (10) is suitable for the clinical trial based on the status of the applicant (10).

[0056] If the processor (110) determines that the applicant (10) falls into a risk group (e.g., normal group, low risk group) that is set as suitable for the clinical trial among multiple risk groups set for the clinical trial, the processor (110) may determine that the applicant (10) is suitable for the clinical trial. Here, the risk group of the applicant (10) may be determined according to output information regarding the status of the clinical trial applicant (10) obtained from the pre-learned neural network model (600). In addition, if the processor (110) determines that the clinical trial applicant (10) is suitable for the clinical trial, the processor (110) may select the applicant (10) as a clinical trial subject.

[0057] FIG. 4 is a flowchart of a method for predicting the occurrence of cardiac toxicity in an applicant (10) according to one embodiment of the present disclosure and determining suitability for a clinical trial. S410 illustrated in FIG. 4 may correspond to S210 illustrated in FIG. 2 . Therefore, a detailed description thereof will be omitted.

[0058] According to one embodiment of the present disclosure, the processor (110) can predict whether cardiac toxicity occurs during the clinical trial of an applicant (10) based on electrocardiogram data acquired through a pre-learned neural network model (600) (S420), and can identify a risk group for the clinical trial of the applicant based on the predicted result of cardiac toxicity occurrence (S430).

[0059] Specifically, the processor (110) can predict the cardiotoxicity of the applicant (10) and determine the risk group of the applicant (10) for a clinical trial. Cardiotoxicity can be a physiological response that negatively affects cardiac tissue or cardiac function due to the administration of a drug or treatment. At this time, cardiotoxicity can appear in the form of cardiac diseases such as decreased systolic function of the left ventricle, decreased diastolic function, arrhythmia, and heart failure. If cardiotoxicity is not predicted in advance, it can lead to serious adverse reactions or treatment failure in the clinical trial subject. Accordingly, the processor (110) can analyze the characteristic information related to cardiotoxicity inherent in the electrocardiogram data (20) through the neural network model (600), thereby preliminarily assessing whether the applicant (10) currently has cardiotoxicity or the possibility of developing cardiotoxicity in the applicant (10), and based on the results, can identify the risk group of the applicant (10) for a clinical trial.

[0060] If the processor (110) predicts that the applicant (10) has cardiotoxicity or is likely to develop cardiotoxicity when conducting a clinical trial, the applicant (10) may be judged as a high-risk group for the clinical trial. In addition, the processor (110) may determine that the applicant (10) is unsuitable for the clinical trial.

[0061] The processor can determine whether or not cardiac toxicity occurs by analyzing the electrocardiogram data of the applicant (10) using the pre-trained neural network model (600). At this time, the processor can predict whether or not cardiac toxicity occurs in the applicant (10) by taking into account the characteristics of the clinical trial, the purpose of the clinical trial, and the type of the clinical trial. In this regard, according to one embodiment of the present disclosure, the processor (110) can predict whether or not cardiac toxicity occurs in the course of the clinical trial of the applicant (10) based on the electrocardiogram data acquired through the pre-trained neural network model (600), and can determine whether or not the applicant (10) is suitable for the clinical trial based on the predicted result of the cardiac toxicity occurrence. For example, the processor can input electrocardiogram data into the pre-trained neural network model (600) based on learning data including label data regarding the presence or absence of cardiac toxicity of the subject, and obtain a score corresponding to the cardiac toxicity of the applicant (10) from the neural network model (600).

[0062] The processor (110) can determine whether the applicant (10) has cardiotoxicity based on a score corresponding to cardiotoxicity obtained from the neural network model (600). For example, if the obtained score is equal to or greater than a preset first value, the processor (110) can determine that the applicant (10) has cardiotoxicity.

[0063] Alternatively, the processor (110) may determine the possibility of developing cardiotoxicity in the applicant (10) based on the score corresponding to cardiotoxicity obtained from the neural network model (600). Specifically, the processor (110) may determine the possibility of developing cardiotoxicity in the applicant (10) during the clinical trial based on the score obtained from the neural network model (600). In particular, if the score is equal to or greater than a preset second value, the processor (110) may determine that the applicant (10) is a high-risk group for the clinical trial, if the score is equal to or greater than a preset third value and less than the second value, the processor (110) may determine that the applicant (10) is a low-risk group for the clinical trial, and if the score is less than the preset third value, the processor (110) may determine that the applicant (10) is a normal group for the clinical trial. In this case, if the applicant (10) is a high-risk group, the processor (110) may determine that the applicant (10) is highly likely to develop cardiotoxicity during the clinical trial, and may determine that the applicant (10) is unsuitable for the clinical trial. At this time, the reference values ​​for classifying the risk group (i.e., the second value and the third value) and the number of risk groups can be set based on the type of clinical trial.

[0064] Meanwhile, according to one embodiment of the present disclosure, the processor can determine the possibility of a left ventricular systolic dysfunction and a left ventricular diastolic dysfunction of a clinical trial applicant (10) based on a pre-learned neural network model (600), and can use the determination result to predict whether or not the applicant (10) develops cardiac toxicity.

[0065] FIG. 5 is an exemplary diagram of a method for predicting whether or not a volunteer (10) develops cardiac toxicity by determining left ventricular systolic dysfunction and left ventricular diastolic dysfunction according to one embodiment of the present disclosure.

[0066] According to one embodiment of the present disclosure, a pre-learned neural network model (600) may include a neural network model (600) (hereinafter, a first neural network model (610)) trained to determine whether a left ventricular systolic function of an applicant (10) is impaired based on electrocardiogram data. At this time, the processor inputs the acquired electrocardiogram data into the first neural network model (610) to obtain a score (hereinafter, a first score) for the possibility of left ventricular systolic function impairment of the applicant (10), and based on the acquired first score, it is possible to predict whether cardiac toxicity occurs during the course of a clinical trial.

[0067] Here, the score may be a numerical value that quantifies the possibility of left ventricular systolic dysfunction of the clinical trial applicant (10) determined based on feature information extracted from the electrocardiogram data (20) (e.g., feature points (Q wave, R wave, etc.) in the electrocardiogram signal, intervals between feature points (RR interval, etc.), waveform shape, amplitude, etc.). The score may be a probability value output from the first neural network model (610) (e.g., a softmax layer of the first neural network model (610)), or may be a value calculated by applying a preset weight to the probability value. The description of the neural network model (600) described above may be equally applied to the first neural network model (610).

[0068] The processor (110) can input the acquired electrocardiogram data (20) into the first neural network model (610) to obtain a score corresponding to the possibility of left ventricular systolic dysfunction of the clinical trial volunteer (10). Then, the processor (110) can determine whether or not the volunteer (10) develops cardiotoxicity during the clinical trial based on the acquired score. At this time, the higher the score, the higher the possibility of left ventricular systolic dysfunction of the clinical trial volunteer (10), and the higher the possibility of cardiotoxicity of the volunteer (10). If the processor determines that the possibility of cardiotoxicity is high, the processor can determine that the volunteer (10) is a high-risk group for the clinical trial and thus is unsuitable for the clinical trial.

[0069] In addition, according to one embodiment of the present disclosure, the pre-learned neural network model (600) may include a neural network model (600) (hereinafter, a second neural network model (620)) trained to determine whether the applicant (10) has left ventricular diastolic dysfunction based on electrocardiogram data. At this time, the processor inputs the acquired electrocardiogram data into the second neural network model (620) to obtain a score (hereinafter, a second score) for the possibility of the applicant (10) having left ventricular diastolic dysfunction, and based on the obtained second score, it is possible to predict whether cardiac toxicity occurs during the course of the clinical trial.

[0070] Here, the score may be a numerical value indicating the possibility of left ventricular systolic / diastolic dysfunction of the clinical trial applicant (10) determined based on feature information extracted from the electrocardiogram data (20) (e.g., feature points (Q wave, R wave, etc.) in the electrocardiogram signal, intervals between feature points (RR interval, etc.), waveform shape, amplitude, etc.). The score may be a probability value output from the second neural network model (620) (e.g., a softmax layer of the second neural network model (620)), or may be a value calculated by applying a preset weight to the probability value. The description of the neural network model (600) described above may be equally applied to the second neural network model (620).

[0071] The processor (110) can input the acquired electrocardiogram data (20) into the second neural network model (620) to obtain a score corresponding to the possibility of left ventricular systolic / diastolic dysfunction of the clinical trial applicant (10). Then, the processor (110) can determine whether or not cardiac toxicity occurs in the applicant (10) during the clinical trial based on the acquired score. At this time, the higher the score, the higher the possibility of left ventricular diastolic dysfunction in the clinical trial applicant (10), and if it is determined that the possibility of cardiac toxicity in the applicant (10) is high, the processor can determine that the applicant (10) is a high-risk group for the clinical trial and is therefore unsuitable for the clinical trial.

[0072] In particular, the processor (110) can compare the acquired first and second scores with preset reference values ​​(e.g., the fourth to seventh values) to classify the status of the clinical trial volunteer (10) regarding cardiotoxicity into a preset grade. Here, the preset grade may be a grade that classifies the degree of onset of cardiotoxicity of the clinical trial volunteer (10). The grade classified for the degree of onset of cardiotoxicity of the clinical trial volunteer (10) may correspond to a risk group set for the above-described clinical trial.

[0073] For example, if the first score and the second score are less than the fourth value and the fifth value, respectively, the processor (110) determines the clinical trial applicant (10) as having a normal grade (or normal group) for cardiotoxicity; if the first score is equal to or greater than the fourth value and less than the sixth value, and the second score is less than the fifth value, the processor (110) determines the clinical trial applicant (10) as having a caution grade (or low-risk group); if the second score is equal to or greater than the fifth value and less than the seventh value, and the first score is less than the fourth value, the processor (110) determines the clinical trial applicant (10) as having a caution grade (low-risk group); if the first score is equal to or greater than the fourth value and less than the sixth value, and the second score is equal to or greater than the fifth value and less than the seventh value, the processor (110) determines the clinical trial applicant (10) as having a borderline grade (or medium-risk group) for cardiotoxicity; if the first score and the second score are equal to or greater than the sixth value and greater than the seventh value, the processor (110) determines the clinical trial applicant (10) as having a risk grade (or high-risk group) for cardiotoxicity. At this time, the processor may determine that the applicant (10) is unsuitable for the clinical trial if he or she falls into the high-risk group. However, the classification of the risk group and the determination of suitability for the risk group described above are merely examples and may vary depending on the specific embodiment.

[0074] Meanwhile, the fourth to seventh values ​​may be set based on the type of clinical trial, the characteristics of the target drug or treatment of the clinical trial, the biological characteristics of the applicant (10), or the results of past clinical data analysis, respectively. For example, the older the applicant (10), the lower the fourth and seventh values ​​may be set relative to younger applicants (10). Alternatively, the higher the dosage of the target drug of the clinical trial, the lower the fourth and seventh values ​​may be set.

[0075] Meanwhile, according to one embodiment of the present disclosure, the processor (110) may arrange a plurality of electrocardiogram data included in the electrocardiogram data (20) in a preset order to obtain integrated electrocardiogram data in which the plurality of electrocardiogram data (20) are integrated. For example, when the electrocardiogram data (20) is obtained according to a 6-lead method, the processor (110) may arrange six electrocardiogram data (20) included in the electrocardiogram data (20) in the order of Lead I, Lead II, Lead III, aVR, aVL, and aVF to obtain integrated electrocardiogram data in a matrix form.

[0076] In addition, the processor (110) can input the acquired integrated electrocardiogram data into the first neural network model (610) to obtain a score corresponding to the possibility of left ventricular systolic dysfunction of the clinical trial applicant (10). In this way, the first neural network model (610) can analyze the feature information of six electrocardiogram data (20) included in the electrocardiogram data (20) to more precisely determine the possibility of left ventricular systolic dysfunction of the clinical trial applicant (10).

[0077] To this end, the input data included in the learning data used to train the first neural network model (610) may also be in the form of integrated electrocardiogram data in which six electrocardiogram data (20) are arranged in a preset order. At this time, the first neural network model (610) may include a plurality of neural network models (600) according to the arrangement order of the plurality of electrocardiogram data (20) included in the electrocardiogram data (20). That is, the plurality of neural network models (600) included in the first neural network model (610) may be trained or perform inference using a plurality of electrocardiogram data (20) arranged in different orders as input data, respectively. For example, the first neural network model (610) may include a 1-1 neural network model corresponding to a first arrangement order (the order of Lead I, Lead II, Lead III, aVR, aVL, and aVF) of a plurality of electrocardiogram data (20) and a 1-2 neural network model corresponding to a second arrangement order (the order of Lead I, Lead II, Lead III, aVL, aVR, and aVF) of a plurality of electrocardiogram data (20). The 1-1 neural network model and the 1-2 neural network model may each be trained using integrated data having different orders of aVR and aVL data as input data. The processor (110) inputs first integrated electrocardiogram data obtained by arranging a plurality of electrocardiogram data (20) according to a first arrangement order (the order of Lead I, Lead II, Lead III, aVR, aVL, and aVF) into a 1-1 neural network model to obtain a first score corresponding to the possibility of left ventricular systolic dysfunction of a clinical trial applicant (10), and inputs second integrated electrocardiogram data obtained by arranging a plurality of electrocardiogram data (20) according to a second arrangement order (the order of Lead I, Lead II, Lead III, aVL, aVR, and aVF) into a 1-2 neural network model to obtain a second score corresponding to the possibility of left ventricular systolic dysfunction of a clinical trial applicant (10).At this time, the processor (110) can determine the status of the clinical trial volunteer (10) for heart disease based on the acquired first score and second score. Specifically, the processor (110) can determine the status of the clinical trial volunteer (10) by comparing the first score and the second score with preset reference values. The preset reference values ​​may be set differently for the first score and the second score, which may be determined according to the sensitivity characteristics of the 1-1 neural network model and the 1-2 neural network model in the learning process or the distribution characteristics of the output first score. If the judgment results for the status of the clinical trial volunteer (10) do not match, the processor (110) can determine the judgment result according to the higher value score as the final judgment result. Alternatively, the processor (110) can calculate the average value of the first score and the second score, and compare the calculated average value with the preset reference value to determine the status of the clinical trial volunteer (10) for the possibility of left ventricular systolic dysfunction.

[0078] The description of the present disclosure described above can be equally applied to the second neural network model (620).

[0079] FIG. 6 is an exemplary diagram of a method for determining the status of a clinical trial applicant (10) using a plurality of sub-neural network models (611 to 616) included in a first neural network model (610) trained to produce a score corresponding to the possibility of left ventricular systolic dysfunction according to one embodiment of the present disclosure.

[0080] Meanwhile, according to an embodiment of the present disclosure, the first neural network model (610) and the second neural network model (620) may include a plurality of sub-neural network models corresponding to each lead type included in the electrocardiogram data. At this time, the processor (110) may input a plurality of electrocardiogram data (20) included in the electrocardiogram data (20), rather than the integrated data of the electrocardiogram data (20), into the sub-neural network models corresponding to each type, thereby obtaining a plurality of first scores and a plurality of second scores of the clinical trial applicant (10) from the plurality of sub-neural network models. Specifically, each sub-neural network model may be trained with a plurality of electrocardiogram data (20) corresponding to different types as input data. For example, referring to FIG. 6, the first sub-neural network model (611) may be trained with a plurality of electrocardiogram data (20) corresponding to Lead I as input data, and the second sub-neural network model (612) may be trained with a plurality of electrocardiogram data (20) corresponding to Lead II as input data. The remaining sub-neural network models (613 to 616) can also be trained with different types of electrocardiogram data (20) included in the electrocardiogram data (20).

[0081] The processor (110) may input a plurality of electrocardiogram data corresponding to different types included in the electrocardiogram data (20) into each sub-neural network model, and obtain a plurality of first scores for the possibility of left ventricular systolic dysfunction and a plurality of second scores for the possibility of left ventricular diastolic dysfunction from each sub-neural network model. Then, the processor (110) may determine the condition of the clinical trial volunteer (10) regarding heart disease based on the obtained plurality of first and second scores. For example, the processor (110) may calculate an average value of the plurality of first scores and compare the calculated average value with preset values ​​(the first value and the third value) to determine the degree of possibility of left ventricular systolic dysfunction of the clinical trial volunteer (10). At this time, the processor (110) may calculate a standard deviation based on the difference between the plurality of scores, remove a score exceeding a preset multiple of the standard deviation, and then determine the average value of the remaining scores as the final score. Meanwhile, the same may be applied to the second score.

[0082] Meanwhile, the processor (110) may select at least one clinical trial subject from among a plurality of clinical trial applicants (10) based on the predicted results regarding cardiac toxicity. To reiterate the above-described example, the processor (110) may select as clinical trial subjects applicants (10) belonging to the normal group, low-risk group, and medium-risk group with respect to cardiac toxicity. However, this is not limited thereto, and applicants (10) belonging to the high-risk group may also be selected as clinical trial subjects, but the clinical trial for such applicants (10) may be conducted differently from applicants (10) belonging to other risk groups.

[0083] In this regard, the processor (110) can set different clinical trial plans for each clinical trial subject. Specifically, the processor (110) can apply a customized dosing strategy based on the individual subject's risk level of cardiac toxicity by varying the dosage of the drug in the clinical trial or adjusting the administration cycle or route of administration. For example, for subjects in the high-risk group, a relatively smaller amount of drug may be administered compared to subjects in the normal, caution, or borderline categories, or the dosing interval may be increased.

[0084] Meanwhile, the processor (110) repeatedly obtains electrocardiogram data (20) from the subject during the clinical trial process, and analyzes the obtained electrocardiogram data (20) using the first and second neural network models (620) to determine whether to discontinue the clinical trial for the subject or to adjust the medication administered to the subject. That is, the processor (110) repeatedly calculates a first score and a second score for each subject, and can monitor the pattern and trend of the first score and the second score.

[0085] At this time, if the subject's cardiac toxicity risk level shows a tendency to increase over time or a rapid change is detected, the processor (110) may stop the clinical trial of the subject or adjust at least one of the dosage, administration cycle, or administration route of the drug administered to the subject.

[0086] In this way, the processor (110) can analyze changes in the subject's cardiac function in real time based on electrocardiogram data (20), and determine whether to discontinue the clinical trial or individualize the drug administration strategy based on the results, thereby improving the safety and precision of the clinical trial.

[0087] FIG. 7 is a block diagram of a computing device (700) according to another embodiment of the present disclosure.

[0088] Referring to FIG. 7, a computing device (700) according to an embodiment of the present disclosure includes a processor (710), a memory (720), a communication interface (730), a sensing unit (740), a display (750), a clinical trial volunteer interface (760), a camera (770), and a speaker (780). Among the configurations illustrated in FIG. 7, the processor (710), the memory (720), and the communication interface (730) correspond to the processor (110) and memory (120) configurations of the computing device (100) illustrated in FIG. 1, and thus a detailed description thereof will be omitted.

[0089] The sensing unit (740) can obtain electrocardiogram data (20) of a clinical trial volunteer. For example, the sensing unit (740) can include a plurality of electrodes. At this time, the processor (110) can obtain electrocardiogram data (20) of the clinical trial volunteer as electrocardiogram data (20) through at least one electrode. Alternatively, the processor can obtain sensing values ​​regarding the posture of the clinical trial volunteer by using a gyro sensor, an IMU sensor, or the like included in the sensing unit (740).

[0090] The display (750) can display various images. Here, the images include both still images and moving images. The display (750) can output the results of the assessment of the status of a clinical trial applicant (e.g., risk group result information) or output information related to an electrocardiogram (e.g., an electrocardiogram graph). The display (750) can be implemented as a display in various forms, such as an LCD (Liquid Crystal Display Panel), an OLED (Organic Light Emitting Diodes), an LCoS (Liquid Crystal on Silicon), a DLP (Digital Light Processing), etc. In addition, the display (750) can also include a driving circuit, a backlight unit, etc., which can be implemented in a form, such as an a-TFT, an LTPS (low temperature poly silicon) TFT, an OTFT (organic TFT), etc.

[0091] Meanwhile, the display (750) may be implemented as a touch screen by being combined with a touch panel, and in this case, the display (750) may perform the function of not only an output interface that outputs an image through the touch screen, but also an input interface that receives a touch input from a clinical trial applicant.

[0092] The clinical trial applicant interface (760) is a component used by the computing device (700) to interact with the clinical trial applicant, and may include, but is not limited to, at least one of a touch sensor, a motion sensor, a button, a jog dial, and a switch. The processor (710) may receive the underlying disease and biological information of the clinical trial applicant through the clinical trial applicant interface (760).

[0093] The camera (770) captures images of objects surrounding the computing device (700). Specifically, the camera (770) can capture images of clinical trial volunteers. At this time, the processor (710) can identify underlying diseases of the clinical trial volunteers based on the captured images. To this end, the camera (770) can be implemented as an imaging device such as a CMOS image sensor (CIS) with a CMOS structure, a charge coupled device (CCD) with a CCD structure, etc. However, the present invention is not limited thereto, and the camera (770) can be implemented as a camera module with various resolutions capable of capturing subjects.

[0094] Meanwhile, the camera (770) may be implemented as a depth camera (e.g., an IR depth camera), a stereo camera, or an RGB camera.

[0095] The speaker (780) is a component that outputs various audio data that have undergone various processing operations, such as decoding, amplification, and noise filtering, by an audio processing unit (not shown). The speaker (780) can output various notification sounds or voice messages. According to one embodiment of the present disclosure, the processor (710) can convert an electrical signal received from an external device into the voice of a clinical trial volunteer and output it through the speaker (780). For example, the speaker (780) can output the judgment result regarding the status of a clinical trial volunteer in the form of a voice message.

[0096] The various embodiments of the present disclosure described above can be combined with additional embodiments and modified within the scope understood by those skilled in the art in light of the detailed description above. It should be understood that the embodiments of the present disclosure are illustrative in all respects and not restrictive. For example, each component described as a single component may be implemented in a distributed manner, and likewise, components described as distributed may be implemented in a combined manner. Accordingly, all changes or modifications derived from the meaning, scope, and equivalent concepts of the claims of the present disclosure should be construed as being included within the scope of the present disclosure.

Claims

1. A method for performing a clinical trial based on electrocardiogram data including at least one processor, A step of obtaining electrocardiogram data of an applicant for a clinical trial; and A step of identifying a risk group for the clinical trial of the applicant based on the acquired electrocardiogram data through a pre-learned neural network model; method.

2. In paragraph 1, The step of identifying the risk group for the above clinical trial of the above applicant is, A step of predicting whether cardiac toxicity occurs in the course of the clinical trial of the applicant based on the acquired electrocardiogram data through the learned neural network model; and A step of identifying a risk group for the clinical trial of the applicant based on the prediction result of the occurrence of the cardiac toxicity; method.

3. In paragraph 2, The above-mentioned pre-trained neural network model is, At least one of a first neural network model trained to determine whether the applicant has impaired left ventricular systolic function based on the electrocardiogram data or a second neural network model trained to determine whether the applicant has impaired left ventricular diastolic function based on the electrocardiogram data, method.

4. In paragraph 3, The step of predicting whether cardiac toxicity occurs during the course of the above clinical trial is: A step of inputting the acquired electrocardiogram data into the first neural network model to obtain a first score for the possibility of left ventricular systolic dysfunction of the applicant; and A step of predicting whether cardiac toxicity occurs during the course of the clinical trial based on the first score obtained above; method.

5. In paragraph 3, The step of predicting whether cardiac toxicity occurs during the course of the above clinical trial is: A step of inputting the acquired electrocardiogram data into the second neural network model to obtain a second score for the possibility of left ventricular diastolic dysfunction of the applicant; and A step of predicting whether cardiac toxicity occurs during the course of the clinical trial based on the second score obtained above; including; method.

6. In paragraph 3, The step of predicting whether cardiac toxicity occurs during the course of the above clinical trial is: A step of identifying a risk group for cardiotoxicity of the applicant based on a first score for the possibility of left ventricular systolic dysfunction of the applicant obtained from the first neural network model and a second score for the possibility of left ventricular diastolic dysfunction of the applicant obtained from the second neural network model, thereby predicting whether the applicant will develop cardiotoxicity during the course of the clinical trial; including; method.

7. In paragraph 6, The step of predicting whether cardiac toxicity occurs during the course of the above clinical trial is: A step of identifying the applicant as being at high risk of cardiac toxicity if the first score is greater than or equal to a first reference value and the second score is greater than or equal to a second reference value; If the applicant is identified as being at high risk for cardiotoxicity, a step of determining that the applicant is unsuitable for the clinical trial is included; method.

8. In paragraph 6, The step of predicting whether cardiac toxicity occurs during the course of the above clinical trial is: A step of identifying the applicant as being at high risk of cardiac toxicity if the first score is greater than or equal to a first reference value and the second score is greater than or equal to a second reference value; If the applicant is identified as belonging to the high risk group for the cardiotoxicity, a step of setting the dosage of the drug administered to the applicant to be lower than that of the applicant belonging to the lower risk group than the high risk group; method.

9. In paragraph 6, A step of selecting the applicant as a clinical trial subject if the applicant is determined to be suitable for the clinical trial based on the first score and the second score; and Including a step of obtaining electrocardiogram data from the applicant during the course of the clinical trial, and determining whether to discontinue the clinical trial for the clinical subject based on the obtained electrocardiogram data through the pre-learned neural network model; method.

10. In paragraph 9, A step of adjusting the dosage of a drug administered to the clinical subject based on the acquired electrocardiogram data through the learned neural network model; method.

11. In a computing device that performs clinical trials based on electrocardiogram data, a processor comprising at least one core; and a memory including program codes executable by the processor; The above processor, Obtaining electrocardiogram data of a clinical trial applicant, and determining whether the applicant is suitable for the clinical trial based on the obtained electrocardiogram data through a pre-trained neural network model. method. Computing device.

12. A computer program stored in a computer-readable storage medium, wherein the computer program, when executed on one or more processors, causes an operation to perform a clinical trial based on electrocardiogram data. The above action is, The act of obtaining electrocardiogram data of a clinical trial applicant; and An operation of determining whether the applicant is suitable for the clinical trial based on the acquired electrocardiogram data through a pre-learned neural network model; Computer program.

Images