KR102407585B1 - Test method for negative determination of polymerase chain reaction using artificial intelligence - Google Patents

Abstract

A PCR test method performed by an electronic device is provided. The test method may include: collecting, by the electronic device, a plurality of pre-stored PCR test data; learning, by the electronic device, the artificial intelligence model by inputting the pre-stored inspection data into an artificial intelligence model; inputting, by the electronic device, test data for each cycle into the AI model when a new PCR test is performed; determining, by the electronic device, whether a voice is negative based on the input test data for each cycle; includes

Description

Test method for negative determination of polymerase chain reaction using artificial intelligence {Test method for negative determination of polymerase chain reaction using artificial intelligence}

The present invention relates to a test method for negative determination of a polymerase chain reaction using artificial intelligence.

Polymerase chain reaction (PCR) test is a test method for judging positive reactions for various diseases. It is decided whether the result is negative or not.

This polymerase chain reaction can be used to replicate not only DNA but also RNA, and it is determined whether the disease is negative based on the fluorescence generated from the fluorescent marker included in a specific helix.

In particular, as a type of PCR, real-time RT-PCR reverse transcribes RNA to generate DNA, and then replicates the DNA to determine whether the disease is negative. It is being used as a method for the detection of RNA viruses such as COVID-19.

However, in the case of such a PCR test, a lot of time is required for DNA replication for negative determination. For example, if it takes 5 minutes to make one copy (1 cycle), in the case of DNA or RNA that needs to be judged positive after 40 copies (40 cycles of replication), the negative result is returned 200 minutes after the test be able to judge

Therefore, since it takes several hours for one PCR device to test one person, it is impossible to perform a quick test in a situation where a large number of tests are required in a quick time.

Laid-open Patent Publication No. 10-2003-0073255, 2003.09.19

The problem to be solved by the present invention is to provide a test method for negative determination of the polymerase chain reaction using artificial intelligence.

The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

A PCR test method performed by an electronic device according to an aspect of the present invention for solving the above-described problems, the electronic device collects a plurality of pre-stored PCR test data; learning, by the electronic device, the artificial intelligence model by inputting the pre-stored inspection data into an artificial intelligence model; inputting, by the electronic device, test data for each cycle into the AI model when a new PCR test is performed; determining, by the electronic device, whether a voice is negative based on the input test data for each cycle; includes

At this time, the plurality of pre-stored PCR test data is data divided by the first cycle to the n-th cycle of each PCR test data, and the collecting of the plurality of pre-stored PCR test data includes: obtaining first data obtained by classifying PCR test data for each test cycle; acquiring a change amount in the m-th cycle as second data based on the inspection data in the m-1 th cycle, the m th cycle, and the m+1 th cycle; and obtaining the differential coefficient data obtained based on the first data and the second data as third data; Including, m may be 2 or more and n-1 or less.

In this case, the artificial intelligence model includes a first artificial intelligence model for analyzing the first data, a second artificial intelligence model for analyzing the second data, and a third artificial intelligence model for analyzing the third data and the determining of whether the voice is voiced may include checking each input cycle based on a result of ensemble of the output values of the first artificial intelligence model, the second artificial intelligence model, and the third artificial intelligence model. It may include; determining whether or not a voice is based on the data.

In this case, the determining whether the voice is negative may include: acquiring test data in the mth cycle; predicting a result value in an m+1th cycle to an nth cycle based on the test data obtained in the first cycle to the mth cycle; determining that the test result is negative when the predicted result value in the m+1th cycle to the nth cycle is less than a preset first value; may include

In this case, the determining whether the voice is negative may include: acquiring test data in the mth cycle; acquiring a first predicted result value in an m+1th cycle to an nth cycle based on the test data acquired in the first cycle to the mth cycle; acquiring test data in the m+1th cycle; obtaining a second prediction result value in an m+2 th cycle to an n th cycle based on the inspection data obtained in the first cycle to the m+1 th cycle; obtaining fourth data on an amount of change in a prediction result obtained based on the first prediction result value and the second prediction result value; and determining whether a voice is present based on the fourth data. may include

At this time, the artificial intelligence model, the artificial intelligence model further comprises a fourth artificial intelligence model for analyzing the fourth data, and the determining whether the voice is, the first artificial intelligence model, the second Determining whether or not negative based on the input test data for each cycle based on the result of ensemble of the output values of the artificial intelligence model, the third artificial intelligence model, and the fourth artificial intelligence model; can

In this case, the step of determining whether a voice is negative based on the fourth data may include: determining that the test result is negative when the amount of change in the first prediction result value and the second prediction result value is within a preset range; may include

In this case, the test method may include determining that the test result is positive when the predicted result value in the m+1th cycle to the nth cycle is equal to or greater than a preset first value; may include

Other specific details of the invention are included in the detailed description and drawings.

According to the above-described various embodiments of the present invention, there is a new effect of more rapidly obtaining a PCR test.

Effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a system diagram illustrating a system according to an embodiment of the present invention.
2 is a flowchart for explaining the present invention according to an embodiment of the present invention.
3 and 4 are exemplary diagrams illustrating graphs for determining a voice response according to an embodiment of the present invention.
5 is an exemplary diagram for explaining an ensemble model according to an embodiment of the present invention.
6 is a block diagram of an apparatus according to an embodiment of the present invention.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only the present embodiments allow the disclosure of the present invention to be complete, and those of ordinary skill in the art to which the present invention pertains. It is provided to fully understand the scope of the present invention to those skilled in the art, and the present invention is only defined by the scope of the claims.

The terminology used herein is for the purpose of describing the embodiments and is not intended to limit the present invention. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other components in addition to the stated components. Like reference numerals refer to like elements throughout, and "and/or" includes each and every combination of one or more of the recited elements. Although "first", "second", etc. are used to describe various elements, these elements are not limited by these terms, of course. These terms are only used to distinguish one component from another. Accordingly, it goes without saying that the first component mentioned below may be the second component within the spirit of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used herein will have the meaning commonly understood by those of ordinary skill in the art to which this invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless specifically defined explicitly.

As used herein, the term “unit” or “module” refers to a hardware component such as software, FPGA, or ASIC, and “unit” or “module” performs certain roles. However, “part” or “module” is not meant to be limited to software or hardware. A “unit” or “module” may be configured to reside on an addressable storage medium or to reproduce one or more processors. Thus, as an example, “part” or “module” refers to components such as software components, object-oriented software components, class components and task components, processes, functions, properties, Includes procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. Components and functionality provided within “parts” or “modules” may be combined into a smaller number of components and “parts” or “modules” or as additional components and “parts” or “modules”. can be further separated.

Spatially relative terms "below", "beneath", "lower", "above", "upper", etc. It can be used to easily describe the correlation between a component and other components. Spatially relative terms should be understood as terms including different directions of components during use or operation in addition to the directions shown in the drawings. For example, when a component shown in the drawing is turned over, a component described as “beneath” or “beneath” of another component may be placed “above” of the other component. can Accordingly, the exemplary term “below” may include both directions below and above. Components may also be oriented in other orientations, and thus spatially relative terms may be interpreted according to orientation.

In this specification, a computer means all types of hardware devices including at least one processor, and may be understood as encompassing software configurations operating in the corresponding hardware device according to embodiments. For example, a computer may be understood to include, but is not limited to, smart phones, tablet PCs, desktops, notebooks, and user clients and applications running on each device.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Each step described in this specification is described as being performed by a computer, but the subject of each step is not limited thereto, and at least a portion of each step may be performed in different devices according to embodiments.

1 is a system diagram illustrating a system according to an embodiment of the present invention.

As shown in FIG. 1 , the system according to the present invention may include an electronic device 100 and a PCR device 200 .

The electronic device 100 according to the present invention is configured to receive and analyze a PCR test result from the PCR device 200 . The electronic device 100 may determine whether the test result is positive or negative based on PCR data performed so far. According to an embodiment, the electronic device 100 may include at least one of a communication unit, a display, a memory, and a processor.

The communication unit may include at least one of a WiFi chip, a Bluetooth chip, a wireless communication chip, and an NFC chip. In particular, each of the WiFi chip and the Bluetooth chip may perform communication using a WiFi method or a Bluetooth method. In the case of using a WiFi chip or a Bluetooth chip, various types of connection information such as an SSID and a session key are first transmitted and received, and various types of information can be transmitted and received after communication connection using the same. The wireless communication chip refers to a chip that performs communication according to various communication standards such as IEEE, Zigbee, 3rd Generation (3G), 3rd Generation Partnership Project (3GPP), and Long Term Evolution (LTE). The NFC chip refers to a chip operating in an NFC (Near Field Communication) method using a 13.56 MHz band among various RF-ID frequency bands such as 135 kHz, 13.56 MHz, 433 MHz, 860 to 960 MHz, and 2.45 GHz.

The display is implemented as a liquid crystal display (LCD), an organic light emitting diode (OLED), or a plasma display panel (PDP), etc., so that various screens that can be provided can be displayed. .

The memory may store an operating system (O/S) for driving the electronic device 100 . In addition, various software programs or applications for operating the electronic device 100 may be stored in the memory according to various embodiments of the present disclosure. The memory may store various information such as various data input, set, or generated during execution of a program or application.

The processor includes a RAM, a ROM, a main CPU, first to n interfaces, and a bus. In this case, the RAM, ROM, main CPU, first to n interfaces, etc. may be connected to each other through a bus.

On the other hand, the electronic device 100 may include an artificial intelligence model for PCR analysis, and the artificial intelligence model may be constructed in consideration of the field of application of the recognition model, the purpose of learning, or the computer performance of the device. The artificial intelligence model may be, for example, a model based on a neural network. AI models can be designed to simulate the human brain structure on a computer. The artificial intelligence model may include a plurality of network nodes having weights that simulate neurons of a human neural network. A plurality of network nodes may each form a connection relationship to simulate a synaptic activity in which a neuron sends and receives a signal through a synapse. The artificial intelligence model, for example, may include a neural network model or a deep learning model developed from a neural network model. In a deep learning model, a plurality of network nodes may exchange data according to a convolutional connection relationship while being located at different depths (or layers).

delete

delete

delete

delete

For example, a model such as a Deep Neural Network (DNN), a Recurrent Neural Network (RNN), or a Bidirectional Recurrent Deep Neural Network (BRDNN) may be used as the AI model, but is not limited thereto.

Meanwhile, in the above-described various embodiments, the operation of the artificial intelligence model has been described, but in general, there is a disadvantage that a large storage space and a high-performance GPU are required in order for the artificial intelligence model to operate smoothly.

Accordingly, the artificial intelligence model operated according to another embodiment of the present disclosure may be compressed and used as necessary. For example, the AI model may be compressed using at least one of Pruning, Quantization, Decomposition, and Knowledge Distillation. Pruning is one of the compression techniques that deletes meaningless weights or weights that do not significantly affect the output value among the weights of the AI model. Quantization is one of compression techniques that quantizes each weight into a preset bit. Decomposition is one of compression techniques for reducing the size of a weight by linearly decomposing a weight matrix or tensor, which is a set of weights. Knowledge Distillation is one of the compression techniques that creates and trains a Student model smaller than the original model by using the original model as the Teacher model. In this case, the Student model may be generated through the aforementioned pruning, decomposition, or quantization.

Hereinafter, specific embodiments of the present invention will be described with reference to FIGS. 2 to 4 .

2 is a flowchart for explaining the present invention according to an embodiment of the present invention.

In step S110, the electronic device 100 may collect a plurality of pre-stored PCR test data.

In an embodiment, the inspection data may be data including at least one of numerical data as shown in Tables 1 and 2 to be described later, and graph data obtained based on the numerical data.

In this case, the plurality of pre-stored PCR test data may be data obtained by dividing each PCR test data by first cycle to nth cycle. According to an embodiment, n may be 40, but is not limited thereto.

Meanwhile, the electronic device 100 may acquire first data obtained by classifying a plurality of pre-stored PCR test data for each test cycle.

For example, as shown in Table 1 below, the electronic device 100 may obtain first data obtained by classifying pre-stored PCR test data for each test cycle.

Cycle A1 A2 A3 A4 A5 A6 A7 One 2.44 1.09 1.56 -37.32 58.45 23.01 39.00 2 -4.06 -5.02 13.06 6.59 38.14 1.40 10.33 3 -3.19 6.29 6.35 0.51 24.99 2.54 4.39 4 3.95 0.97 10.31 2.42 12.26 -2.97 6.64 5 -2.33 -8.83 4.80 15.88 7.51 2.48 -0.97 6 0.09 -9.41 -7.07 0.19 -7.07 -7.91 -5.37 7 7.41 -0.33 1.92 11.40 -8.65 -1.81 -0.26 8 13.59 0.95 -6.15 2.89 -9.44 -2.39 -1.72 9 8.91 6.88 -11.32 -4.29 -6.31 -2.51 3.38 10 7.17 9.16 -4.75 1.99 -5.38 -2.14 5.84 11 2.66 2.99 2.73 -5.04 -0.51 4.14 -1.32 12 4.40 -4.81 4.04 -4.67 9.22 -2.20 -3.33 13 6.56 -6.06 12.66 13.38 1.94 -1.91 -2.22 14 -4.83 2.03 -1.06 10.50 13.41 -3.28 6.97 15 -1.30 -1.22 4.53 3.50 16.21 0.83 3.01 16 -7.64 0.44 15.49 -3.19 27.98 0.76 5.74 17 -10.83 0.02 3.09 2.03 26.46 1.24 0.66 18 -8.88 1.11 1.90 -13.86 34.60 3.18 3.50 19 -19.81 0.63 -7.06 -15.18 28.64 5.97 -1.06 20 -17.10 -3.07 -2.48 -19.19 39.11 -3.53 5.34 21 -28.66 -7.34 -3.55 -17.78 36.48 -1.94 -5.29 22 -35.00 -2.64 -8.81 -32.87 47.76 -7.16 -4.12 23 -45.00 -5.05 -17.50 -34.93 53.39 -4.13 -2.38 24 -44.33 -14.66 -10.55 -33.53 54.03 -12.12 -1.14 25 -46.87 -12.38 -11.88 -38.65 61.41 -13.09 3.14 26 -54.89 -14.32 -2.15 -46.60 74.97 -6.33 -8.97 27 -54.76 -8.06 -16.79 -51.41 88.41 -3.34 -3.76 28 -60.19 -2.49 -16.38 -55.39 97.68 -7.64 2.21 29 -71.27 -7.51 -11.68 -59.27 102.79 -4.61 -5.12 30 -58.17 0.55 -6.37 -63.14 111.83 -3.54 -4.54 31 -73.20 -7.10 -5.68 -68.45 126.52 -12.53 3.05 32 -77.61 -7.47 -12.77 -69.25 129.79 2.71 -4.99 33 -80.55 1.20 -8.07 -72.82 137.96 -9.19 0.19 34 -85.03 -2.51 -6.91 -76.21 152.27 4.82 0.00 35 -90.47 -5.01 -6.54 -89.30 168.40 -5.87 -2.08 36 -97.63 -2.55 -12.70 -89.98 177.92 0.67 1.59 37 -86.34 3.78 -11.72 -98.25 183.62 0.93 -0.08 38 -96.61 -0.29 -7.47 -98.10 199.35 4.30 -2.06 39 -99.70 2.72 -3.41 -94.67 212.86 3.19 3.14 40 -95.41 8.28 -0.88 -114.12 222.87 4.46 2.72

A change amount in the m-th cycle may be acquired as the second data based on the inspection data in the m-1 th cycle, the m th cycle, and the m+1 th cycle. In this case, of course, m may be 2 or more and n-1 or less. For example, as shown in Table 2 below, the electronic device 100 may obtain second data on the amount of change in each cycle. have.

Cycle A1 A2 A3 A4 A5 A6 A7 One 2.44 1.09 1.56 -37.32 58.45 23.01 39.00 2 -6.50 -6.11 11.50 43.90 -20.32 -21.61 -28.67 3 0.87 11.31 -6.70 -6.08 -13.15 1.14 -5.94 4 7.14 -5.32 3.95 1.91 -12.72 -5.50 2.25 5 -6.27 -9.80 -5.51 13.46 -4.76 5.45 -7.60 6 2.42 -0.58 -11.87 -15.69 -14.58 -10.39 -4.41 7 7.32 9.08 8.99 11.21 -1.58 6.10 5.11 8 6.18 1.28 -8.08 -8.51 -0.79 -0.58 -1.46 9 -4.68 5.93 -5.17 -7.18 3.13 -0.12 5.10 10 -1.75 2.28 6.57 6.28 0.93 0.37 2.46 11 -4.51 -6.17 7.48 -7.03 4.87 6.27 -7.16 12 1.74 -7.79 1.31 0.38 9.73 -6.33 -2.01 13 2.16 -1.25 8.62 18.04 -7.28 0.29 1.11 14 -11.38 8.08 -13.73 -2.88 11.48 -1.37 9.19 15 3.53 -3.25 5.59 -6.99 2.80 4.12 -3.96 16 -6.34 1.66 10.96 -6.69 11.77 -0.08 2.73 17 -3.20 -0.42 -12.40 5.22 -1.52 0.48 -5.08 18 1.96 1.09 -1.20 -15.90 8.14 1.94 2.84 19 -10.94 -0.49 -8.96 -1.31 -5.96 2.79 -4.57 20 2.72 -3.70 4.58 -4.01 10.46 -9.50 6.41 21 -11.57 -4.27 -1.07 1.41 -2.63 1.59 -10.63 22 -6.34 4.70 -5.25 -15.08 11.29 -5.22 1.17 23 -10.00 -2.41 -8.69 -2.06 5.62 3.03 1.75 24 0.67 -9.61 6.95 1.39 0.64 -7.99 1.24 25 -2.54 2.28 -1.32 -5.12 7.38 -0.97 4.28 26 -8.02 -1.95 9.73 -7.94 13.56 6.76 -12.11 27 0.13 6.26 -14.64 -4.82 13.44 2.99 5.21 28 -5.43 5.57 0.41 -3.97 9.27 -4.30 5.97 29 -11.07 -5.02 4.69 -3.88 5.11 3.03 -7.32 30 13.09 8.06 5.31 -3.87 9.04 1.07 0.58 31 -15.03 -7.65 0.70 -5.32 14.69 -8.99 7.59 32 -4.41 -0.37 -7.09 -0.79 3.27 15.24 -8.05 33 -2.94 8.68 4.70 -3.57 8.18 -11.91 5.18 34 -4.47 -3.71 1.16 -3.40 14.30 14.01 -0.19 35 -5.44 -2.49 0.37 -13.09 16.13 -10.69 -2.08 36 -7.16 2.46 -6.17 -0.67 9.52 6.54 3.67 37 11.28 6.33 0.98 -8.27 5.70 0.26 -1.67 38 -10.26 -4.07 4.25 0.15 15.73 3.37 -1.98 39 -3.09 3.02 4.07 3.43 13.51 -1.11 5.19 40 4.29 5.56 2.53 -19.44 10.01 1.27 -0.41

Specifically, the value of the first cycle of the A1 test data of Table 2 is the same as the value of the first cycle of the A1 test data of Table 1. The value of the second cycle of the A1 test data of Table 2 is the difference value (-4.06-2.44) between the value of the second cycle of the A1 test data of Table 1 and the value of the first cycle. That is, in the mth cycle, The second data may be derived as in Equation 1 below.

[Equation 1]

In this case, in D _{(a, b)} , a may mean a table b, and a may mean a value in the a-th cycle. B is one of the values of 1 or 2.

As another embodiment, the electronic device 100 may obtain the differential coefficient data obtained based on the first data and the second data as the third data.

The electronic device 100 may learn the artificial intelligence model by using the above-described first to third data as training data.

Specifically, in step S120 , the electronic device 100 may learn the artificial intelligence model by inputting pre-stored inspection data into the artificial intelligence model.

As described above, the electronic device 100 may learn the artificial intelligence model based on the first data to the third data.

In this case, the artificial intelligence model may include a first artificial intelligence model for analyzing the first data, a second artificial intelligence model for analyzing the second data, and a third artificial intelligence model for analyzing the third data.

That is, the electronic device 100 may input the acquired first data to the third data into different AI models to derive test results, and determine whether or not the test result is negative based on the plurality of derived test results.

In step S130 , when a new PCR test is performed, the electronic device 100 may input test data for each cycle into the AI model.

That is, whenever one cycle ends, the electronic device 100 acquires inspection data in real time, predicts inspection data in subsequent cycles based on the acquired data, and based on the predicted value of the inspection data It may be determined whether the test result is negative.

In operation S140 , the electronic device 100 may determine whether a voice is negative based on the input test data for each cycle.

According to an embodiment, the electronic device 100 receives the inspection data for each cycle input based on the result of ensemble of the output values of the first artificial intelligence model, the second artificial intelligence model, and the third artificial intelligence model. Based on this, it is possible to determine whether a voice is present or not.

Meanwhile, according to various embodiments of the present disclosure, the ensemble model may be used in various ways. In one embodiment, numerical data of the inspection data is used as training data of an RNN artificial intelligence model to which MLP (Multilayer Perception) is applied, and graph data is used as training data of a CNN artificial intelligence model, artificial intelligence learned by numerical data It goes without saying that the artificial intelligence model learned by the model and graph data can be ensembled to derive the result.

In the present invention, when a plurality of artificial intelligence models are ensembled, the ensembled artificial intelligence model may be referred to as an ensemble model. That is, the ensemble model may refer to a model that calculates and uses the outputs of at least two artificial intelligence models. Each AI model constituting the ensemble model can be called an individual model or an AI model.

Meanwhile, the ensemble model may be generated by ensembles an output value of a specific (or arbitrary) layer among a plurality of layers included in the first artificial intelligence model to the third artificial intelligence model. In this case, a layer may mean one layer in a multilayer neural network structure. That is, in a plurality of ensemble models, one layer may be defined as a layer. In general, a layer may be defined in the form of y=f(Wx+b) with respect to an input vector x and an output vector y, and W and b may be referred to as layer parameters.

Meanwhile, FIG. 5 is an exemplary diagram illustrating an embodiment of a method for generating an ensemble model. Specifically, the first to third artificial intelligence models may include a conv layer, a ReLU layer, and a Softmax layer. The ensemble model shown in FIG. 5 may be a basic ensemble model (Simple Ensemble) that ensembles final output values of the first artificial intelligence model to the third artificial intelligence model.

However, according to various embodiments of the present invention, the ensemble model does not transfer the output value of the ReLU layer to the next layer, but instead transfers the ensemble value of the output value of each ReLU layer to the next layer. Of course, it can be an ensemble) model. In this case, there is an effect of saving runtime memory.

As another embodiment, it is similar to the Stepwise Ensemble model, but if it is not appropriate to ensemble all layers as needed, a Stepwise Partial Ensemble model may be used.

For example, the electronic device 100 may step-wise ensemble the first artificial intelligence model and the second artificial intelligence model, and output the result value of the third artificial intelligence model as it is.

As another embodiment, when the performance of the electronic device 100 is limited, the electronic device 100 may apply a cascade partial ensemble model. For example, the electronic device 100 performs step-wise ensemble of the first artificial intelligence model and the second artificial intelligence model, and the ensembled output value is an input value of one artificial intelligence model (eg, the fourth artificial intelligence model). and the final output values of the fourth artificial intelligence model and the third artificial intelligence model can be ensembled. Such a cascade ensemble model can save memory as the number of artificial intelligence models decreases as the steps are repeated. That is, when there are a plurality of artificial intelligence models predicted to have similar results, the cascade ensemble model may be used.

As another embodiment, of course, the electronic device 100 may apply a stochastic partial ensemble model for deriving a stable result value. The probabilistic ensemble model ensembles at least two artificial intelligence models among a plurality of artificial intelligence models, but one output value may be ensembled with a plurality of artificial intelligence models, and this ensemble is determined probabilistically. For example, in the probabilistic ensemble model, the first artificial intelligence model and the second artificial intelligence model are ensembled, the first artificial intelligence model and the third artificial intelligence model are ensembled, and the second artificial intelligence model and the third artificial intelligence model are ensembled. and the three ensembled output values can be used as input values for two AI models.

Hereinafter, the result value derivation process using the artificial intelligence model will be described in detail.

The electronic device 100 may obtain the test data in the mth cycle and input it into the artificial intelligence model. Specifically, the electronic device 100 may predict the result value in the m+1th cycle to the nth cycle by inputting the inspection data obtained in the first cycle to the mth cycle into the AI model.

In this case, when the predicted result value in the m+1th cycle to the nth cycle is less than a preset first value, the electronic device 100 may determine the test result as negative.

In this case, the preset first value may be a value corresponding to at least one of the first line 310 illustrated in FIG. 3 or the second line 410 illustrated in FIG. 4 .

Meanwhile, according to another embodiment, the electronic device 100 may further include fourth data for voice determination and a fourth artificial intelligence model for learning the fourth data.

In this case, the electronic device 100 inspects each input cycle based on the result of ensemble of the output values of the first artificial intelligence model, the second artificial intelligence model, the third artificial intelligence model, and the fourth artificial intelligence model. Based on the data, it is possible to determine whether a voice is present.

Specifically, the electronic device 100 acquires the test data in the mth cycle, and the first prediction in the m+1th cycle to the nth cycle based on the test data acquired in the first cycle to the mth cycle result can be obtained.

Thereafter, the electronic device 100 acquires the inspection data in the m+1th cycle, and based on the inspection data obtained in the first cycle to the m+1th cycle, in the m+2th cycle to the nth cycle. 2 A prediction result can be obtained.

The electronic device 100 may obtain fourth data on the amount of change in the prediction result obtained based on the first prediction result value and the second prediction result value, and determine whether a voice is present based on the fourth data. .

Specifically, when the amount of change of the first prediction result value and the second prediction result value is within a preset range, the electronic device 100 may determine that the test result is negative. That is, as PCR test data is collected in real time, the artificial intelligence model can predict future data based on the data collected so far. In this case, if the difference between the future data predicted at the m-th and the future data predicted at the m+1 is not large, it can be interpreted as meaning that the AI model has accurately determined. Accordingly, when the amount of change of the first prediction result value and the second prediction result value is within a preset range, the electronic device 100 may determine that the test result is negative.

Furthermore, due to the characteristics of the artificial intelligence model, when little real-time data is collected, the deviation of the prediction data may be large, but when the real-time data is sufficiently collected, the deviation of the prediction data will be small. That is, after the amount of input data becomes greater than or equal to a preset value, the deviation of the prediction data becomes smaller.

In this case, in the above-described embodiment, a method of acquiring fourth data based on data collected up to the mth cycle and data collected up to the m+1th cycle has been described. However, by using an arbitrary value greater than 1, Of course, the fourth data may be acquired based on the data collected up to the cycle and the data collected up to the m+i-th cycle.

Table 3 below shows data output through an artificial intelligence model using actual data up to 5 cycles, 10 cycles, 15 cycles, 20 cycles, 25 cycles, and 30 cycles, respectively, based on the actually collected PCR data. to be.

Cycle Actual data (40 cycles) 5 cycles 10 cycles 15 cycles 20 cycles 25 cycles 30 cycles One 2.44 - - - - - - 2 -4.06 - - - - - - 3 -3.19 - - - - - - 4 3.95 - - - - - - 5 -2.33 - - - - - - 6 0.09 0.71 - - - - - 7 7.41 10.54 - - - - - 8 13.59 17.24 - - - - - 9 8.91 10.25 - - - - - 10 7.17 10.52 - - - - - 11 2.66 8.69 2.11 - - - - 12 4.40 6.54 5.51 - - - - 13 6.56 15.45 13.20 - - - - 14 -4.83 1.84 0.98 - - - - 15 -1.30 0.04 0.15 - - - - 16 -7.64 -10.68 -9.52 -8.15 - - - 17 -10.83 -20.47 -18.12 -15.48 - - - 18 -8.88 -15.14 -12.25 -10.52 - - - 19 -19.81 -29.12 -27.21 -25.48 - - - 20 -17.10 -30.57 -25.21 -20.46 - - - 21 -28.66 -34.15 -33.58 -31.15 -29.95 - - 22 -35.00 -42.52 -38.15 -32.75 -34.11 - - 23 -45.00 -44.41 -42.25 -47.83 -46.01 - - 24 -44.33 -42.28 -42.12 -43.25 -44.85 - - 25 -46.87 -49.55 -48.45 -47.58 -47.01 - - 26 -54.89 -42.14 -49.21 -50.06 -52.89 -53.89 - 27 -54.76 -40.77 -47.24 -52.90 -54.00 -54.41 - 28 -60.19 -52.32 -58.25 -59.45 -59.99 -60.01 - 29 -71.27 -54.95 -65.15 -66.80 -69.63 -70.02 - 30 -58.17 -58.23 -60.74 -59.31 -58.99 -58.42 - 31 -73.20 -59.65 -66.14 -69.69 -72.82 -72.99 -73.20 32 -77.61 -61.74 -68.85 -71.11 -76.17 -77.51 -77.59 33 -80.55 -66.22 -75.92 -77.28 -79.91 -80.01 -80.51 34 -85.03 -69.17 -76.17 -77.98 -83.15 -84.02 -84.99 35 -90.47 -77.32 -85.55 -89.17 -88.02 -89.92 -90.44 36 -97.63 -75.67 -89.68 -92.55 -95.11 -97.12 -97.23 37 -86.34 -79.21 -84.15 -84.40 -85.85 -85.99 -86.11 38 -96.61 -80.11 -88.52 -92.09 -95.18 -97.58 -96.99 39 -99.70 -82.21 -85.30 -90.71 -96.50 -98.08 -99.00 40 -95.41 -84.44 -98.25 -99.02 -94.02 -94.99 -95.02

Looking at Table 3, it can be seen that the more the actual data is collected, the more the predicted data is similar to the actual data, and further, when 20 or more cycles of data are collected, it can be confirmed that the predicted data and the actual data are almost identical. That is, according to the results of Table 3, the electronic device 100 can determine the test result even if only data up to 20 cycles are collected, so that it has an effect of reducing the time by about half compared to the existing test. .

Meanwhile, even when the test result using the artificial intelligence model is determined to be negative in the mth cycle, the electronic device 100 may additionally acquire test data up to the nth cycle.

Specifically, the electronic device 100 compares the additionally acquired data and the inspection data for the m+1th cycle to the nth cycle predicted through the artificial intelligence model, and retrains the artificial intelligence model based on the comparison result. can

In an embodiment, when the positive/negative determination derived based on actual examination data obtained from the examination up to the nth cycle is different from the positive/negative determination determined by the artificial intelligence model, the electronic device 100 performs the actual examination Data can be used as training data, and artificial intelligence models can be trained.

In another embodiment, even when the positive/negative determination derived based on the actual examination data obtained from the examination up to the nth cycle and the positive/negative determination determined by the artificial intelligence model are the same, the examination data (actual examination data Alternatively, the shape of the graph generated from the prediction test data) may be slightly different. Accordingly, the electronic device 100 may learn the artificial intelligence model by inputting the difference value between the actual inspection data and the predicted inspection data as learning data into the artificial intelligence model.

According to the above-described method, the electronic device 100 may acquire learning data for improving the accuracy of the artificial intelligence model in real time.

However, in the case of a special situation in which a plurality of people must be tested within a short time, there may be cases in which learning data cannot be acquired in order to reduce the time and cost incurred in performing the PCR test to the end. Accordingly, the electronic device 100 may determine whether to perform the PCR test until the end by comprehensively considering the type of PCR test, the number of waiting people, and the urgency of the test.

3 and 4 are exemplary diagrams illustrating graphs for determining a positive/negative reaction according to an embodiment of the present invention.

Specifically, FIG. 3 is an exemplary diagram for explaining the number of cycles required for a positive determination based on actual test data, and FIG. 4 is an exemplary diagram for explaining the number of cycles required for a positive determination using an artificial intelligence model. .

As shown in FIG. 3 , the cycle in the first line 310 corresponding to the preset second value is a value corresponding to the third line 320 , and the test result is positive for the first time in approximately the 23rd cycle. judgment can be confirmed. However, as shown in FIG. 4 , when the artificial intelligence model is used, it can be predicted that the test result is positive based on test data acquired up to approximately the twelfth cycle. That is, the PCR test method using the artificial intelligence model according to the present invention has an effect of saving about 50% of the time compared to the existing test method.

Meanwhile, it goes without saying that the electronic device 100 may perform a positive determination as well as the aforementioned negative determination by using the artificial intelligence model. As an embodiment, when the predicted result value in the m+1th cycle to the nth cycle is equal to or greater than a preset first value, the electronic device 100 may determine the test result as positive.

As an embodiment, when the result value in the kth cycle is equal to or greater than a preset first value, the electronic device 100 determines whether a difference between the k value and the m value is equal to or greater than a preset second value, and the k value When the difference between the m value and the m value is equal to or greater than a preset second value, the accuracy in the k-th cycle may be determined, and if the accuracy is equal to or greater than the third preset value, the test result may be determined to be positive.

In this case, k may be a value having a result value exceeding a first preset value among values m+1 or more and n or less. That is, the k value may be a value indicating a positive reaction first among predicted test results.

Meanwhile, the difference between the k value and the m value may mean a time difference between the test data actually acquired by the PCR apparatus 200 and the predicted data. That is, the larger the difference between the k and m values, the more distant future results are predicted. Therefore, the greater the difference between the k value and the m value, the lower the accuracy of the predicted test result. Accordingly, when the difference between the k value and the m value is equal to or greater than the second preset value, the electronic device 100 may measure the accuracy of the test result.

When the difference between the k value and the m value is equal to or less than the second preset value, the electronic device 100 may omit the accuracy determination. That is, when the difference between the k value and the m value is equal to or less than the second preset value, high accuracy is expected even if no separate accuracy determination is made, so that only the prediction result can be determined without wasting resources for accuracy determination. Accordingly, when the difference between the k value and the m value is less than the second preset value, the electronic device 100 may determine the test result as positive.

Meanwhile, the preset second value may be determined through various methods. For example, the preset second value may be determined based on the amount of learning data input to the artificial intelligence model. That is, as the amount of learning data input to the artificial intelligence model increases, the preset second value may increase.

At this time, the amount of learning data may mean the amount of data for all PCR tests input to the artificial intelligence model, but is not limited thereto. The amount of learning data is for a specific disease (eg, COVID-19) Of course, it may mean the amount of test data.

Meanwhile, when the above-described accuracy is less than the third preset value, the electronic device 100 may acquire the test data acquired in the m+1th cycle. That is, when positive or negative is not determined in an arbitrary cycle, the electronic device 100 may repeat the above-described positive determination process by acquiring test data of the next cycle.

6 is a block diagram of an apparatus according to an embodiment of the present invention.

The processor 102 may include one or more cores (not shown) and a graphic processing unit (not shown) and/or a connection path (eg, a bus, etc.) for transmitting and receiving signals to and from other components. .

The processor 102 according to an embodiment performs the method described with reference to FIGS. 1 to 8 by executing one or more instructions stored in the memory 104 .

For example, the processor 102 acquires new training data by executing one or more instructions stored in the memory, and performs a test on the acquired new training data using the learned model, and the test result, labeling Extracting the first learning data in which the obtained information is obtained with an accuracy greater than or equal to a predetermined first reference value, deleting the extracted first learning data from the new learning data, and removing the new learning data from which the extracted learning data is deleted It is possible to re-train the learned model using

On the other hand, the processor 102 is a RAM (Random Access Memory, not shown) and ROM (Read-Only Memory: ROM) for temporarily and / or permanently storing a signal (or, data) processed inside the processor 102. , not shown) may be further included. In addition, the processor 102 may be implemented in the form of a system on chip (SoC) including at least one of a graphic processing unit, a RAM, and a ROM.

The memory 104 may store programs (one or more instructions) for processing and controlling the processor 102 . Programs stored in the memory 104 may be divided into a plurality of modules according to functions.

The steps of a method or algorithm described in relation to an embodiment of the present invention may be implemented directly in hardware, as a software module executed by hardware, or by a combination thereof. A software module may contain random access memory (RAM), read only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, hard disk, removable disk, CD-ROM, or It may reside in any type of computer-readable recording medium well known in the art to which the present invention pertains.

The components of the present invention may be implemented as a program (or application) to be executed in combination with a computer, which is hardware, and stored in a medium. Components of the present invention may be implemented as software programming or software components, and similarly, embodiments may include various algorithms implemented as data structures, processes, routines, or combinations of other programming constructs, including C, C++ , may be implemented in a programming or scripting language such as Java, assembler, or the like. Functional aspects may be implemented in an algorithm running on one or more processors.

In the above, embodiments of the present invention have been described with reference to the accompanying drawings, but those of ordinary skill in the art to which the present invention pertains can realize that the present invention can be embodied in other specific forms without changing the technical spirit or essential features thereof. you will be able to understand Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

100: electronic device
200: PCR device

Claims (10)

In the PCR test method performed by an electronic device,
collecting, by the electronic device, a plurality of pre-stored PCR test data;
learning, by the electronic device, the artificial intelligence model by inputting the pre-stored inspection data into an artificial intelligence model;
inputting, by the electronic device, test data for each cycle into the AI model when a new PCR test is performed;
determining, by the electronic device, whether a voice is negative based on the input test data for each cycle; including,
The plurality of pre-stored PCR test data is,
Each PCR test data is data divided by the first cycle to the nth cycle,
The step of collecting the pre-stored plurality of PCR test data comprises:
obtaining first data obtained by classifying the plurality of pre-stored PCR test data for each test cycle;
acquiring a change amount in the m-th cycle as second data based on the inspection data in the m-1 th cycle, the m th cycle, and the m+1 th cycle; and
obtaining the differential coefficient data obtained based on the first data and the second data as third data; including,
Wherein m is 2 or more and n-1 or less,
The artificial intelligence model includes a first artificial intelligence model for analyzing the first data, a second artificial intelligence model for analyzing the second data, and a third artificial intelligence model for analyzing the third data, ,
The step of determining whether the voice is
Determining whether negative based on a plurality of PCR test results derived from an ensemble model, including at least two or more artificial intelligence models among the first artificial intelligence model, the second artificial intelligence model, and the third artificial intelligence model step; including;
The step of determining whether the voice is
acquiring test data in the mth cycle;
obtaining a first predicted result value in an m+1th cycle to an nth cycle based on the test data obtained in the first cycle to the mth cycle;
acquiring test data in the m+1th cycle;
obtaining a second prediction result value in an m+2th cycle to an nth cycle based on the test data obtained in the first cycle to the m+1th cycle;
obtaining fourth data on a change amount of a prediction result obtained based on the first prediction result value and the second prediction result value; and
determining whether there is a voice based on the fourth data; including,
The ensemble model is
An inspection method, characterized in that it is a model that calculates and uses the outputs of at least two artificial intelligence models among a plurality of artificial intelligence models. delete delete delete delete According to claim 1,
The artificial intelligence model is
Further comprising a fourth artificial intelligence model for analyzing the fourth data,
The step of determining whether the voice is
A plurality of PCR test results derived from an ensemble model, including at least two or more artificial intelligence models among the first artificial intelligence model, the second artificial intelligence model, the third artificial intelligence model, and the fourth artificial intelligence model A test method comprising a; determining whether or not a negative based on the According to claim 1,
The step of determining whether a voice is voiced based on the fourth data,
determining that the PCR test result is negative when the amount of change of the first prediction result value and the second prediction result value is within a preset range; inspection method comprising
According to claim 1,
The inspection method is
determining that the PCR test result is positive when the predicted result value in the m+1th cycle to the nth cycle is greater than or equal to a preset first value; inspection method comprising
a memory storing one or more instructions; and
a processor executing the one or more instructions stored in the memory;
The processor by executing the one or more instructions,
An apparatus for performing the method of claim 1 . A computer program stored in a computer-readable recording medium in combination with a computer, which is hardware, to perform the method of claim 1.

Images