KR20230130318A - Method for diagnosing parkinson's disease and system thereof - Google Patents

Abstract

A method and system for diagnosing Parkinson's disease are provided. The method for diagnosing Parkinson's disease according to some embodiments of the present disclosure performs optimal scaling regression analysis to determine a main independent variable among a plurality of independent variables constituting multidimensional data for a Parkinson's disease patient; Using the determined key independent variables, a diagnostic model is constructed to diagnose the type of cognitive impairment of Parkinson's disease or the risk of progressing to Parkinson's disease-dementia (PDD), and the diagnosis of Parkinson's disease can be accurately performed using the constructed diagnostic model. You can.

Description

Parkinson's disease diagnosis method and system {METHOD FOR DIAGNOSING PARKINSON'S DISEASE AND SYSTEM THEREOF}

The present disclosure relates to a method and system for diagnosing Parkinson's disease, and more specifically, to determine variables (factors) that can be mainly used in diagnosing Parkinson's disease through data analysis, and to use the determined variables to diagnose Parkinson's disease. It relates to a method of performing diagnosis and a system for performing the method.

Parkinson's disease generally causes movement disorders such as bradykinesia, tremors, muscle spasms, postural instability, and gait disorders, but as the disease worsens, it is reported to be accompanied by additional symptoms such as autonomic dysfunction and cognitive impairment. In particular, according to previous research results, the possibility of developing dementia in Parkinson's disease patients is very high, and as many as one out of three patients with Parkinson's Disease with Mild Cognitive Impairment (PD-MCI) have Parkinson's disease. -It is known to progress to dementia (Parkinson's Disease with Dementia, PDD).

When a Parkinson's disease patient progresses to dementia, the burden on caregivers rapidly increases, so early diagnosis of Parkinson's disease-dementia (PDD) is emerging as one of the major issues in geriatric medicine. However, research on methods for early diagnosis of Parkinson's disease-dementia (PDD) is still quite limited.

Korean Patent Publication No. 10-2021-0017616 (published on 21.02.17)

The technical problem to be solved through some embodiments of the present disclosure is to provide a method for accurately diagnosing Parkinson's disease and a system for performing the method.

Another technical problem to be solved through some embodiments of the present disclosure is a method of determining variables (factors) for accurately diagnosing the type of cognitive impairment of Parkinson's disease (or the risk of progressing to Parkinson's disease-dementia) and performing the method. It is to provide a system that does this.

Another technical problem to be solved through some embodiments of the present disclosure is a method of building a model that can accurately diagnose the type of cognitive impairment of Parkinson's disease (or the risk of progressing to Parkinson's disease-dementia) and how to perform the method. providing a system.

The technical problems of the present disclosure are not limited to the technical problems mentioned above, and other technical problems not mentioned can be clearly understood by those skilled in the art from the description below.

In order to solve the above technical problem, a Parkinson's disease diagnosis method according to some embodiments of the present disclosure is a Parkinson's disease diagnosis method performed by at least one computing device, comprising: acquiring multidimensional data on a Parkinson's disease patient; Performing optimal scaling regression analysis on the obtained multidimensional data to determine a main independent variable among the plurality of independent variables, using the determined main independent variables to determine the type of cognitive impairment of Parkinson's disease or It may include constructing a diagnostic model for diagnosing the risk of developing Parkinson's disease-dementia (PDD) and performing a diagnosis of Parkinson's disease using the constructed diagnostic model. At this time, the plurality of independent variables constituting the multidimensional data include variables related to neuropsychological test results of the Parkinson's disease patient, and the dependent variable of the multidimensional data relates to the type of cognitive impairment in Parkinson's disease, and the cognitive impairment Types may include Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD).

In one embodiment, the plurality of independent variables include Global Deterioration Scale (GDS), Hoehn and Yahr stage (H&Y stage), Clinical Dementia Rating (CDR), Schwab & England Activities of Daily Living (S&E ADL), and K-IADL ( Must include at least one of the following test results: the Korean Instrumental Activities of Daily Living (UPDRS), the Unified Parkinson's Disease Rating Scale (UPDRS), the Korean Mini-Mental State Examination (K-MMSE), and the Korean-Montreal Cognitive Assessment (K-MoCA). You can.

In one embodiment, the main independent variables may include variables related to the Korean Mini-Mental State Examination (K-MMSE) and the Hoehn and Yahr stage (H&Y stage) test results.

In one embodiment, the step of determining the main independent variables includes calculating regression coefficients for the plurality of independent variables by performing the optimization scale regression analysis, and calculating the regression coefficients for the plurality of independent variables based on the calculated regression coefficients. It may include the step of determining the main independent variable among the variables.

In one embodiment, the plurality of independent variables include a first variable having an ordinal scale and an interval scale, and determining the main independent variable includes maintaining the order of the first variable, It may include the step of changing the number of sections, the size of the sections, or the value corresponding to the section using alternating least squares.

In order to solve the above-described technical problem, a Parkinson's disease diagnosis system according to some embodiments of the present disclosure includes one or more processors and a memory that stores one or more instructions, wherein the one or more processors store the one or more instructions. An operation of acquiring multidimensional data on a Parkinson's disease patient by executing an operation of performing an optimal scaling regression analysis on the obtained multidimensional data to determine a main independent variable among the plurality of independent variables. , the operation of constructing a diagnostic model to diagnose the type of cognitive impairment of Parkinson's disease or the risk of progressing to Parkinson's disease-dementia (PDD) using the main independent variables determined above, and the diagnosis of Parkinson's disease using the constructed diagnostic model. You can perform the action you are performing. At this time, the plurality of independent variables constituting the multidimensional data include variables related to neuropsychological test results of the Parkinson's disease patient, and the dependent variable of the multidimensional data relates to the type of cognitive impairment in Parkinson's disease, and the cognitive impairment Types may include Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD).

In order to solve the above-described technical problem, a computer program according to some embodiments of the present disclosure includes the steps of combining with a computing device to obtain multidimensional data for a Parkinson's disease patient, and performing optimization scale regression on the obtained multidimensional data. Performing an optimal scaling regression analysis to determine a key independent variable among the plurality of independent variables, determining the risk of developing a type of cognitive impairment or Parkinson's disease-dementia (PDD) in Parkinson's disease using the determined key independent variables. It may be stored in a computer-readable recording medium to execute the steps of constructing a diagnostic model for diagnosing and performing a diagnosis of Parkinson's disease using the constructed diagnostic model. At this time, the plurality of independent variables constituting the multidimensional data include variables related to neuropsychological test results of the Parkinson's disease patient, and the dependent variable of the multidimensional data relates to the type of cognitive impairment in Parkinson's disease, and the cognitive impairment Types may include Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD).

According to various embodiments of the present disclosure, by performing regression analysis on multidimensional data on Parkinson's disease patients, key independent variables that are highly correlated with the dependent variable (or have a strong influence on the dependent variable) are selected among the plurality of independent variables. can be decided. In addition, the accuracy of Parkinson's disease diagnosis can be improved by performing a diagnosis on the diagnosed patient using the basic diagnostic information on the determined main independent variables.

Additionally, by performing optimal scaling regression analysis on multidimensional data, key independent variables can be determined more accurately.

In addition, by constructing a diagnostic model to diagnose the type of cognitive impairment of Parkinson's disease (or the risk of progressing to Parkinson's disease-dementia) using key independent variables, and performing a diagnosis using the constructed diagnostic model, Parkinson's disease-dementia ( Early diagnosis of PDD can be performed accurately.

The effects according to the technical idea of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below.

1 is an exemplary diagram for explaining a Parkinson's disease diagnosis system and its service provision environment according to some embodiments of the present disclosure.
2 is an example flowchart schematically showing a method for diagnosing Parkinson's disease according to some embodiments of the present disclosure.
FIG. 3 is an exemplary diagram for further explaining an optimization scale regression analysis that can be used in a method for diagnosing Parkinson's disease according to some embodiments of the present disclosure.
FIG. 4 is an exemplary diagram for explaining the structure and operation of a diagnostic model according to an embodiment of the present disclosure.
5 to 7 are exemplary diagrams for explaining the structure and operation of a diagnostic model according to another embodiment of the present disclosure.
FIG. 8 is an exemplary diagram for explaining the structure and operation of a diagnostic model according to another embodiment of the present disclosure.
FIG. 9 is an exemplary diagram illustrating an autoencoder-based data purification method according to an embodiment of the present disclosure.
10 illustrates an example computing device that can implement a Parkinson's disease diagnostic system according to some embodiments of the present disclosure.

Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the attached drawings. The advantages and features of the present disclosure and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the technical idea of the present disclosure is not limited to the following embodiments and may be implemented in various different forms. The following examples are merely intended to complete the technical idea of the present disclosure and to be used in the technical field to which the present disclosure belongs. It is provided to fully inform those skilled in the art of the scope of the present disclosure, and the technical idea of the present disclosure is only defined by the scope of the claims.

When adding reference numerals to components in each drawing, it should be noted that identical components are given the same reference numerals as much as possible even if they are shown in different drawings. Additionally, in describing the present disclosure, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present disclosure, the detailed description will be omitted.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings that can be commonly understood by those skilled in the art to which this disclosure pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined. The terminology used herein is for the purpose of describing embodiments and is not intended to limit the disclosure. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context.

Additionally, in describing the components of the present disclosure, terms such as first, second, A, B, (a), and (b) may be used. These terms are only used to distinguish the component from other components, and the nature, sequence, or order of the component is not limited by the term. When a component is described as being “connected,” “coupled,” or “connected” to another component, that component may be directly connected or connected to that other component, but there is another component between each component. It will be understood that elements may be “connected,” “combined,” or “connected.”

As used in this disclosure, “comprises” and/or “comprising” refers to a referenced component, step, operation and/or element that includes one or more other components, steps, operations and/or elements. Does not exclude presence or addition.

Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the attached drawings.

FIG. 1 is an exemplary diagram illustrating a Parkinson's disease diagnosis system 10 and its service provision environment according to some embodiments of the present disclosure.

As shown in FIG. 1, the Parkinson's disease diagnosis system 10 may be a system that performs diagnosis or provides diagnostic services for Parkinson's disease. For example, the Parkinson's disease diagnosis system 10 receives basic diagnostic information from the diagnosed terminal 11, performs a diagnosis of Parkinson's disease based on the received basic diagnostic information, and sends the diagnosis results to the diagnosed terminal 11. can be provided. Hereinafter, for convenience of explanation, the Parkinson's disease diagnosis system 10 will be abbreviated as 'diagnosis system 10'.

Basic diagnostic information may include, but is not limited to, information about the Parkinson's disease patient's demographic information, disease history, and various neuropsychological test results related to Parkinson's disease. For specific examples of basic diagnostic information, please refer to the explanation of 'multidimensional data' that will be described later.

In addition, the diagnostic results may include information on, for example, the condition (progress state) of Parkinson's disease, the type of cognitive impairment of Parkinson's disease, the degree of cognitive impairment, the risk of developing Parkinson's disease-dementia (PDD), treatment methods, etc. , but is not limited to this. Additionally, the cognitive impairment types of Parkinson's disease can be divided into, for example, Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD), but are not limited thereto.

In various embodiments of the present disclosure, the diagnostic system 10 may determine variables (factors) that can be mainly used in diagnosing Parkinson's disease by analyzing multidimensional data on a Parkinson's disease patient. For example, the diagnosis system 10 diagnoses the type of cognitive impairment of Parkinson's disease (or the risk of progressing to Parkinson's disease-dementia) among a plurality of independent variables constituting multidimensional data through optimal scaling regression analysis. Key independent variables can be determined. In addition, the diagnosis system 10 builds a model to diagnose the type of cognitive impairment of Parkinson's disease (or the risk of progression to Parkinson's disease-dementia) using the determined main independent variables, and uses the constructed diagnostic model to provide information about the diagnosed person. Diagnosis can be performed. By doing so, early diagnosis of Parkinson's disease-dementia (PDD) can be accurately performed. This embodiment will be described in detail later with reference to the drawings in FIG. 2 and below.

Diagnostic system 10 may be implemented with at least one computing device. For example, diagnostic system 10 may be implemented as a single computing device. As another example, diagnostic system 10 may be implemented with a plurality of computing devices, wherein a first function of diagnostic system 10 may be implemented in a first computing device and a second function may be implemented in a second computing device. Alternatively, certain functions of diagnostic system 10 may be implemented on multiple computing devices.

The computing device may be any device equipped with a computing (processing) function regardless of type, and an example of such a device will be described later with reference to FIG. 10.

The diagnosis terminal 11 is a terminal owned by a person receiving diagnostic services (e.g. a Parkinson's disease patient), and may be implemented with any computing device.

As shown, the diagnosed terminal 11 and the diagnostic system 10 can communicate through a network. Here, the network is implemented as all types of wired/wireless networks such as Local Area Network (LAN), Wide Area Network (WAN), mobile radio communication network, Wibro (Wireless Broadband Internet), etc. It can be.

So far, the diagnostic system 10 and its exemplary service provision environment according to some embodiments of the present disclosure have been described with reference to FIG. 1 . Hereinafter, a Parkinson's disease diagnosis method that can be performed in the diagnosis system 10 illustrated in FIG. 1 will be described in detail.

The method for diagnosing Parkinson's disease, which will be described below, may be performed by at least one computing device. In other words, the method to be described below may be implemented as one or more instructions that can be executed by at least one computing device (or processor). Hereinafter, for convenience of understanding, the description will be continued assuming that the method to be described later is performed in the diagnostic system 10 in the environment illustrated in FIG. 1. Accordingly, when the subject of a specific step (action) is omitted, it can be understood as being performed by the diagnostic system 10. Of course, in a real environment, some steps of the method described below may be performed on other computing devices.

2 is an example flowchart schematically showing a method for diagnosing Parkinson's disease according to some embodiments of the present disclosure. However, this is only a preferred embodiment for achieving the purpose of the present disclosure, and of course, some steps may be added or deleted as needed.

As shown in FIG. 2, this embodiment may begin at step S21 of acquiring multidimensional data for a Parkinson's disease patient. Here, multidimensional data may mean data consisting of a plurality of independent variables and one or more dependent variables, and may include a plurality of data samples. For example, the diagnostic system 10 can acquire multidimensional data on a large number of Parkinson's disease patients with different types of cognitive impairment, and the cognitive impairment type of Parkinson's disease is, for example, Parkinson's disease-mild cognitive impairment (PD-MCI). and Parkinson's disease-dementia (PDD). The diagnostic system 10 may use any method to acquire multidimensional data.

Multiple independent variables include, for example, demographic information of the Parkinson's disease patient (e.g. gender, age, race, etc.), family history of Parkinson's disease, family history of Alzheimer's disease, education, dominant hand (e.g. left-handed, right-handed), and disease history. Variables related to (e.g. history of traumatic brain injury, high blood pressure, carbon monoxide poisoning, diabetes, hyperlipidemia, stroke, etc.) and variables related to the results of various neuropsychological tests related to Parkinson's disease can be included. However, it is not limited to this. Additionally, examples of neuropsychological tests related to Parkinson's disease include GDS (Global Deterioration Scale), H&Y stage (Hoehn and Yahr stage), CDR (Clinical Dementia Rating), S&E ADL (Schwab & England Activities of Daily Living), and K-IADL. (the Korean Instrumental Activities of Daily Living), UPDRS (Unified Parkinson's Disease Rating Scale), K-MMSE (the Korean Mini-Mental State Examination), and K-MoCA (the Korean-Montreal Cognitive Assessment). It is not limited to this.

The dependent variable may represent the type of cognitive impairment in Parkinson's disease (or the risk of progression to Parkinson's disease-dementia) or the degree of cognitive impairment. Alternatively, the dependent variable may represent the progression stage of Parkinson's disease, etc.

Meanwhile, in one embodiment, the acquired multidimensional data may be data on which data cleaning has been performed. For example, the diagnostic system 10 acquires original multidimensional data on Parkinson's disease patients and performs a process of refining (e.g. removing data samples with outliers, etc.) on the original multidimensional data using an autoencoder-based deep learning model. can do. By doing so, the noise present in the original multidimensional data can be removed and the accuracy of data analysis can be improved. This embodiment will be described later with reference to FIG. 9.

In step S22, optimal scaling regression analysis may be performed on the acquired multidimensional data, and a main independent variable may be determined from a plurality of independent variables using the analysis results. For example, the diagnostic system 10 performs an optimization scale regression analysis based on a linear regression model to calculate a regression coefficient (e.g. standardized regression coefficient, unstandardized regression coefficient) for each of a plurality of independent variables, and uses the calculated regression coefficient Based on this, the main independent variables can be determined. For example, the diagnostic system 10 may determine an independent variable whose size (e.g. absolute value) of the regression coefficient is greater than or equal to the reference value as the main independent variable. This is because a large size of the regression coefficient means that the correlation (or influence on the dependent variable) with the dependent variable (e.g. cognitive impairment type of Parkinson's disease) is large.

Optimized scale regression analysis converts the scale of the variables (e.g. independent variables, dependent variables) that make up multidimensional data into a scale suitable for the regression model (i.e., optimal scale), and performs regression analysis on the multidimensional data with the converted scale. It refers to the technique used.

The reasons for finding the optimal scale of variables are as follows. Firstly, in order to perform regression analysis, the values of variables with a nominal or ordinal scale (i.e., categorical variables) must be converted into appropriate numeric values, and secondly, This is because the size and/or distribution of the value of a specific variable may not match the influence of the variable or the assumptions of the model, which may reduce the accuracy of the regression analysis. For example, although a specific independent variable has an equal interval scale, the influence of that independent variable on the dependent variable (or the relationship between the independent variable and the dependent variable) may not conform to the equal interval scale. In this case, if regression analysis is performed without scale conversion, the accuracy of the analysis will inevitably decrease. As a more specific example, if the K-MMSE score is above the first score, there is almost no possibility of progressing to Parkinson's disease-dementia (PDD), but if the K-MMSE score is below the second score, the possibility of progressing to Parkinson's disease-dementia (PDD) increases exponentially. Let's assume it increases. Even in this case, if linear regression analysis is performed using the K-MMSE score as is, the accuracy of the analysis will inevitably decrease. Therefore, it can be understood that scale conversion is performed, such as subdividing the interval where the KMMSE score is lower than the second score and converting the K-MMSE score to a value of a different size.

To find the optimal scale of variables, alternating least squares can be used. The alternating least squares method refers to a technique of repeatedly updating the scale of variables to minimize the squared error. If you are working in the relevant technical field, you will already be familiar with the alternating least squares method, so its description will be omitted. .

Meanwhile, the method of obtaining the optimal scale may vary depending on the scale characteristics of the variables.

For example, if a specific variable is a variable with an ordinal scale, the diagnostic system 10 can repeatedly transform the value (or scale) of the variable using the alternating least squares method while maintaining the order of the variable. there is. Alternatively, the end-stage system 10 may ignore the order of the variable and repeatedly change the value (or scale) of the variable using the alternating least squares method.

As another example, when a specific variable is a variable having an ordinal scale and an equal interval scale, the diagnosis system 10 maintains the order of the variable and determines the size and number of intervals of the variable and the values corresponding to the intervals. It can be changed repeatedly using the alternating least squares method. For example, as shown in FIG. 3, the diagnostic system 10 can merge and/or subdivide a plurality of sections of the variable (e.g. The optimal measure for (e.g. X) can be obtained.

As described above, when the optimization scale regression analysis is completed, the diagnostic system 10 may determine a main independent variable among a plurality of independent variables based on the regression coefficient. Alternatively, the diagnostic system 10 may determine key independent variables by further considering standard error, significance probability (P-value), etc. in addition to the regression coefficient.

In one embodiment, the key independent variables determined may be the test results of the K-MMSE and H&Y stages. According to the experimental results of the inventors of the present disclosure, the regression coefficients of the K-MMSE and H&Y stages were found to be higher than those of other independent variables. The results of the experiments performed by the present inventors will be described later.

This will be described again with reference to FIG. 2 .

In step S23, a model for diagnosing the type of cognitive impairment of Parkinson's disease (or the risk of progression to Parkinson's disease-dementia) can be constructed using key independent variables. For example, the diagnostic system 10 can build a diagnostic model with key independent variables as factors. Alternatively, the diagnostic system 10 may construct a diagnostic model that has the main independent variable and other independent variables as factors. At this time, the main independent variable may be given a higher weight than other independent variables.

The specific form of the diagnostic model may vary depending on the embodiment.

In one embodiment, the diagnostic model may be a model that diagnoses the type of cognitive impairment in Parkinson's disease by comparing the values of key independent variables with reference values. For example, the diagnostic model may be a model that diagnoses Parkinson's disease-dementia (PDD) or that there is a risk of developing Parkinson's disease-dementia (PDD) when the K-MMSE score (test result) is below the standard value. At this time, the reference value may be any one of 23 to 25, and it has been confirmed that diagnostic accuracy is excellent when a reference value within this numerical range is used. As another example, the diagnostic model may be a model that diagnoses Parkinson's disease-dementia (PDD) or that there is a risk of Parkinson's disease-dementia (PDD) when the H&Y stage test result is above the reference value. At this time, the reference value may be any one of 2.0 to 3.0, and it has been confirmed that diagnostic accuracy is excellent when the reference value within this numerical range is used.

In another embodiment, the diagnostic model may be a regression model (e.g. linear regression model, logistic regression model) with key independent variables as factors. At this time, the dependent variable of the regression model may be related to the type of cognitive impairment of Parkinson's disease (or the risk of progression to Parkinson's disease-dementia).

In another embodiment, the diagnostic model may be an artificial neural network-based deep learning model that predicts the type of cognitive impairment of Parkinson's disease by receiving input values of key independent variables. This embodiment will be described later with reference to FIG. 4.

In another embodiment, the diagnostic model may be an autoencoder-based deep learning model. This embodiment will be described later with reference to FIGS. 5 to 8.

In another embodiment, the diagnostic model may include a first sub-diagnostic model built using a first main independent variable and a second sub-diagnostic model built using a second main independent variable. In this case, the diagnosis system 10 may perform a diagnosis of Parkinson's disease by comprehensively considering the diagnosis results of the first sub-diagnosis model and the second sub-diagnosis model.

In another embodiment, the diagnostic model may be a model based on various combinations of the above-described embodiments.

In step S24, a diagnosis of Parkinson's disease may be performed using the constructed diagnostic model. For example, the diagnosis system 10 receives basic diagnostic information from the terminal 11 of the diagnosed terminal, performs diagnosis on the diagnosed terminal through a diagnostic model based on the basic diagnostic information, and sends the diagnosis result to the terminal 11 of the diagnosed terminal. can be provided to. Basic diagnostic information may include, for example, data (values) on the key independent variables described above.

As a more specific example, the diagnosis system 10 can diagnose the type of Parkinson's disease cognitive impairment of the diagnosed person and the risk of developing Parkinson's disease-dementia (PDD) from basic diagnosis information through a diagnosis model. In addition, the diagnosis system 10 performs repetitive diagnosis periodically or aperiodically through the diagnosis model, and the degree of worsening of the diagnosed person's Parkinson's disease symptoms based on changes in diagnosis results (e.g. change in risk, change in type of cognitive impairment). can also be predicted.

So far, a method for diagnosing Parkinson's disease according to some embodiments of the present disclosure has been described with reference to FIGS. 2 and 3 . According to the above-described method, by performing regression analysis on multidimensional data on Parkinson's disease patients, key independent variables that are highly correlated with the dependent variable (or have a strong influence on the dependent variable) can be determined among a plurality of independent variables. . In addition, the accuracy of Parkinson's disease diagnosis can be improved by performing a diagnosis on the diagnosed patient using the basic diagnostic information on the determined main independent variables. Additionally, by performing optimal scale regression analysis on multidimensional data, key independent variables can be determined more accurately. In addition, by constructing a diagnostic model to diagnose the type of cognitive impairment of Parkinson's disease (or the risk of progressing to Parkinson's disease-dementia) using key independent variables, and performing a diagnosis using the constructed diagnostic model, Parkinson's disease-dementia ( Early diagnosis of PDD can be performed accurately.

Hereinafter, the structure and operation of the diagnostic model according to some embodiments of the present disclosure will be described with reference to FIGS. 4 to 8.

FIG. 4 is an exemplary diagram for explaining the structure and operation of the diagnostic model 40 according to an embodiment of the present disclosure.

As shown in FIG. 4, the diagnostic model 40 according to this embodiment may be a deep learning model based on an artificial neural network. Specifically, the diagnostic model 40 receives the values of the main independent variables and calculates a confidence score for the type of cognitive impairment in Parkinson's disease (e.g. Parkinson's disease-mild cognitive impairment, Parkinson's disease-dementia) (or Parkinson's disease-dementia). It may be an artificial neural network-based model that outputs (predicts) the risk of developing dementia. Of course, the diagnostic model 40 may receive additional input values of other independent variables in addition to the main independent variables. For reference, the risk of developing Parkinson's disease-dementia (PDD) may in some cases be a value calculated based on the confidence score for Parkinson's disease-dementia (PDD).

The diagnostic model 40 can be learned using a training set consisting of data (values) on key independent variables and data (values) on cognitive impairment types in Parkinson's disease. For example, the diagnostic system 10 extracts data on key independent variables and cognitive impairment types from multidimensional data on Parkinson's disease patients to create a training set, and uses the generated training set to learn the diagnostic model 40. You can. At this time, the values of the main independent variables constituting the training set (or each training sample) may be values converted to have an optimal scale.

Once learning is completed, the diagnostic system 10 can perform a diagnosis of Parkinson's disease using the diagnostic model 40. For example, the diagnosis system 10 can obtain a confidence score for the type of cognitive impairment of Parkinson's disease by inputting the diagnosis basic information (e.g. diagnosis basic information corresponding to the main independent variable) of the diagnosed person into the diagnosis model 40. . In addition, the diagnosis system 10 can diagnose the type of cognitive impairment of Parkinson's disease of the diagnosed person based on the obtained confidence score. For example, if the confidence score for Parkinson's disease-dementia (PDD) is above the standard value, the diagnosis system 10 can diagnose the diagnosed person's type of Parkinson's disease cognitive impairment as Parkinson's disease-dementia (PDD).

Below are exemplary diagrams for explaining the structure and operation of a diagnostic model according to another embodiment of the present disclosure, with reference to FIGS. 5 to 7 .

As shown in FIGS. 5 to 7, the diagnosis model according to this embodiment may be an auto-encoder-based deep learning model. The autoencoder-based deep learning model may encompass the basic autoencoder and its various variations (e.g. variant autoencoder). Hereinafter, for convenience of understanding, the autoencoder-based deep learning model is referred to as the 'basic autoencoder'. Let us continue the explanation assuming that it consists of a 'structure'.

First, with reference to FIG. 5, the autoencoder-based deep learning model will be briefly described. Figure 5 illustrates a basic autoencoder structure in which the dimensionality (d) of the latent vector (53; or bottleneck layer) is 3.

As shown in FIG. 5, the autoencoder-based deep learning model 50 may be composed of an encoder 52 and a decoder 54, and the restoration loss between the input data 51 and the output data 55 It can be learned based on (reconstruction loss). At this time, inside the deep learning model 50, the process of encoding the data 51 input through the encoder 51 into a latent vector 53 and decoding the latent vector 53 through the decoder 54 is performed. may be, and as the restoration loss is back-propagated, the weight parameters of the encoder 52 and the decoder 54 may be updated. As this learning process is repeated for a number of training samples constituting the training set, the deep learning model 50 has the ability to compress (encode) and restore (decode) the input data 51. You can do it. Anyone working in the relevant technology field will already be familiar with the operating principles and learning methods of autoencoder-based deep learning models, so further explanation will be omitted.

As shown in FIG. 6, the diagnostic model 60 according to this embodiment is configured to include an encoder 61 and a decoder 62, and is a training set for Parkinson's disease-mild cognitive impairment (PD-MCI) patients. It can be constructed by learning (63). For example, the diagnostic system 10 may learn the diagnostic model 60 based on the restoration loss between the training sample 64 input to the diagnostic model 60 and the data 65 restored through the diagnostic model 60. there is. At this time, each training sample (e.g. 64) constituting the training set 63 may include values for the main independent variables determined through the above-described optimization scale regression analysis, and the values of the main independent variables are adjusted to have an optimization scale. It may be a converted value.

Once learning is completed, the diagnostic system 10 can perform a diagnosis of Parkinson's disease using the diagnostic model 60. Specifically, as shown in FIG. 7 , the diagnosis system 10 can calculate the restoration loss for the basic diagnosis information 71 of the diagnosed patient through the diagnosis model 60. As described above, restoration loss can be calculated based on the difference between the input information 71 and the information 72 restored through the diagnostic model 60. And, if the calculated restoration loss is greater than the standard value, the diagnosis system 10 diagnoses the diagnosed person's Parkinson's disease cognitive impairment type as Parkinson's disease-dementia (PDD), or there is a risk of progression to Parkinson's disease-dementia (PDD). (or high). A large restoration loss means that the diagnosed person's basic diagnostic information (71) is not restored well, which means that the diagnosed person's basic diagnostic information (71) does not correspond to the Parkinson's disease-mild cognitive impairment (PD-MCI) type. This is because it means that there is a high possibility that it will not happen. For reference, the risk of developing Parkinson's disease-dementia (PDD) may in some cases be a value calculated based on restoration loss. For example, the risk may be calculated as a value proportional to the restoration loss.

In some cases, the diagnostic model 60 may be learned as a training set for Parkinson's disease-dementia (PDD) patients. In this case, when the restoration loss calculated through the diagnostic model 60 is below the standard value, the diagnostic system 10 diagnoses the type of Parkinson's disease cognitive impairment of the diagnosed person as Parkinson's disease-dementia (PDD) or Parkinson's disease-dementia ( It can be diagnosed that the risk of developing PDD exists (or is high).

Alternatively, the diagnostic model 60 is a first sub-diagnostic model learned with a training set for Parkinson's disease-mild cognitive impairment (PD-MCI) patients and a second sub-diagnostic model learned with a training set for Parkinson's disease-dementia (PDD) patients. It may also be configured to include a sub-diagnosis model. In this case, the diagnosis system 10 may diagnose the patient's Parkinson's disease by comprehensively considering the restoration loss calculated through the first sub-diagnosis model and the second sub-diagnosis model.

Hereinafter, the structure and operation of the diagnostic model 80 according to another embodiment of the present disclosure will be described with reference to FIG. 8. However, for clarity of the present disclosure, description of content that overlaps with the previous embodiments will be omitted.

As shown in FIG. 8, the diagnostic model 80 according to this embodiment may be configured to further include a classifier 83 in addition to the encoder 81 and decoder 82. At this time, the classifier 83 may be configured to predict the type of cognitive impairment of Parkinson's disease (or the risk of progressing to Parkinson's disease-dementia).

The diagnostic model 80 can be learned using a training set 85 that includes training samples of Parkinson's disease-mild cognitive impairment (PD-MCI) patients and Parkinson's disease-dementia (PDD) patients. In some cases, the training set 85 may further include training samples for normal people. At this time, each training sample (e.g. 86) constituting the training set 85 may include values for the main independent variables determined through the above-described optimization scale regression analysis, and the values of the main independent variables are adjusted to have an optimization scale. It may be a converted value.

Like the above-described diagnostic model 60, the diagnostic model 80 may also be learned based on the restoration loss. The restoration loss may be calculated based on the difference between the input training sample 86 and the data 87 restored through the diagnostic model 80.

Additionally, the diagnostic model 80 may be further learned based on the classification loss calculated through the classifier 83. For example, the diagnosis system 10 extracts a latent vector 84 for the training sample 86 through the encoder 81 and calculates a classification loss for the latent vector 84 extracted through the classifier 83. You can. Classification loss can be calculated based on the difference between the confidence score (or risk of progressing to Parkinson's disease-dementia) for the type of cognitive impairment predicted through the classifier 83 and the correct answer. Additionally, the diagnosis system 10 may update the classifier 83 and the encoder 81 using the classification loss calculated while the decoder 82 is frozen. By doing so, the encoder 81 can be trained to extract a latent vector 84 suitable for classification of cognitive impairment type (or risk of progression to Parkinson's disease-dementia).

Once learning is completed, the diagnostic system 10 can perform a diagnosis of Parkinson's disease using the diagnostic model 80. For example, the diagnosis system 10 extracts a latent vector (e.g. 84) for the basic diagnosis information of the patient through the encoder 81 of the diagnosis model 80, and inputs the extracted latent vector into the classifier 83 You can obtain a confidence score for the type of cognitive impairment. In addition, the diagnosis system 10 can diagnose the type of cognitive impairment of Parkinson's disease of the diagnosed person based on the obtained confidence score. For example, if the confidence score for Parkinson's disease-dementia (PDD) is above the standard value, the diagnosis system 10 can diagnose the diagnosed person's type of Parkinson's disease cognitive impairment as Parkinson's disease-dementia (PDD).

So far, the structure and operation of the diagnostic models 40, 60, and 80 according to some embodiments of the present disclosure have been described with reference to FIGS. 4 to 8. According to the above, by constructing a diagnostic model to diagnose the type of cognitive impairment (or risk of progressing to Parkinson's disease-dementia) of Parkinson's disease using key independent variables and performing a diagnosis using the constructed diagnostic model, Parkinson's disease Diagnosis can be performed accurately. In particular, early diagnosis of Parkinson's disease-dementia (PDD) can be performed accurately.

Hereinafter, a data purification method according to an embodiment of the present disclosure will be described with reference to FIG. 9.

FIG. 9 is an exemplary diagram for explaining a data purification method according to an embodiment of the present disclosure.

As shown in FIG. 9, this embodiment relates to a method of refining data using an autoencoder-based deep learning model 90. For example, the diagnostic system 10 may refine multidimensional data about Parkinson's disease patients using the deep learning model 90 before performing regression analysis.

Specifically, the deep learning model 90 may be configured to include an encoder 91 and a decoder 92, and may be learned using a training set 93 for a normal person (see normal). For example, the diagnosis system 10 creates a deep learning model 90 based on the restoration loss between each training sample 94 constituting the training set 93 and the data 95 restored through the deep learning model 90. can be learned.

Once learning is completed, the diagnostic system 10 can use the deep learning model 90 to refine the original multidimensional data on Parkinson's disease patients. Specifically, the diagnosis system 10 may calculate restoration loss for each data sample constituting the original multidimensional data. Additionally, the diagnosis system 10 may remove data samples in which the restoration loss calculated from the original multidimensional data is below the standard value. A small restoration loss means that the data sample can be restored well through the deep learning model (90), which means that there is a high possibility that the data sample is data for a normal person (or patient data with outliers). Because it means.

Meanwhile, the diagnosis system 10 may perform missing value correction for data samples using an autoencoder-based deep learning model. Specifically, the diagnostic system 10 uses a deep learning model learned with the training set of Parkinson's disease-mild cognitive impairment (PD-MCI) patients on data samples of Parkinson's disease-mild cognitive impairment (PD-MCI) patients. Missing value correction can be performed on data samples of Parkinson's disease-dementia (PDD) patients using a deep learning model learned with the training set of Parkinson's disease-dementia (PDD) patients. For example, the diagnostic system 10 assigns a specific value to a missing value in a data sample of a patient with Parkinson's disease-mild cognitive impairment (PD-MCI), changes the assigned value, and repeatedly calculates the restoration loss through a deep learning model. You can. Additionally, the diagnosis system 10 may correct the missing value with an assigned value when the restoration loss is less than or equal to the reference value. A small restoration loss means that the data sample can be well restored through a deep learning model, which means that the data sample filled with missing values has similar characteristics to other data samples of the same cognitive impairment type. .

So far, a data purification method according to an embodiment of the present disclosure has been described with reference to FIG. 9. According to the above-described method, original multidimensional data can be accurately refined using an autoencoder-based deep learning model.

Hereinafter, an exemplary computing device 100 capable of implementing the diagnostic system 10 according to some embodiments of the present disclosure will be described with reference to FIG. 10.

FIG. 10 is an exemplary hardware configuration diagram showing the computing device 100.

As shown in FIG. 10, the computing device 100 includes one or more processors 101, a bus 103, a communication interface 104, and a memory (loading) a computer program executed by the processor 101. 102) and a storage 105 that stores a computer program 106. However, only components related to the embodiment of the present disclosure are shown in FIG. 10 . Accordingly, a person skilled in the art to which this disclosure pertains can recognize that other general-purpose components may be included in addition to the components shown in FIG. 10 . That is, the computing device 100 may further include various components in addition to those shown in FIG. 10 . Additionally, in some cases, the computing device 100 may be configured with some of the components shown in FIG. 10 omitted. Hereinafter, each component of the computing device 100 will be described.

The processor 101 may control the overall operation of each component of the computing device 100. The processor 101 includes at least one of a Central Processing Unit (CPU), Micro Processor Unit (MPU), Micro Controller Unit (MCU), Graphic Processing Unit (GPU), or any type of processor well known in the art of the present disclosure. It can be configured to include. Additionally, the processor 101 may perform operations on at least one application or program to execute operations/methods according to embodiments of the present disclosure. Computing device 100 may include one or more processors.

Next, memory 102 may store various data, instructions and/or information. Memory 102 may load a computer program 106 from storage 105 to execute operations/methods according to embodiments of the present disclosure. The memory 102 may be implemented as a volatile memory such as RAM, but the technical scope of the present disclosure is not limited thereto.

Next, the bus 103 may provide communication functionality between components of the computing device 100. The bus 103 may be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

Next, the communication interface 104 may support wired and wireless Internet communication of the computing device 100. Additionally, the communication interface 104 may support various communication methods other than Internet communication. To this end, the communication interface 104 may be configured to include a communication module well known in the technical field of the present disclosure.

Next, storage 105 may non-transitory store one or more computer programs 106. The storage 105 may be a non-volatile memory such as Read Only Memory (ROM), Erasable Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), flash memory, a hard disk, a removable disk, or a device well known in the art to which this disclosure pertains. It may be configured to include any known type of computer-readable recording medium.

Next, the computer program 106 may include one or more instructions that, when loaded into the memory 102, cause the processor 101 to perform operations/methods according to various embodiments of the present disclosure. That is, the processor 101 can perform operations/methods according to various embodiments of the present disclosure by executing the one or more instructions.

For example, the computer program 106 may include an operation of acquiring multidimensional data on a Parkinson's disease patient, an operation of performing an optimization scale regression analysis on the obtained multidimensional data, and an operation of determining a main independent variable among a plurality of independent variables, and determining a main independent variable among a plurality of independent variables. An operation to build a diagnostic model that diagnoses the type of cognitive impairment of Parkinson's disease or the risk of progressing to Parkinson's disease-dementia (PDD) using key independent variables, and one or more to perform a diagnosis of Parkinson's disease using the constructed diagnostic model. May contain instructions. In this case, the diagnostic system 10 according to some embodiments of the present disclosure may be implemented through the computing device 100.

So far, an exemplary computing device 100 capable of implementing the diagnostic system 10 according to some embodiments of the present disclosure has been described with reference to FIG. 10 . Below, we will briefly introduce the results of experiments performed by the present inventors.

After receiving approval from both the Research Ethics Review Committee and the Distribution Committee of the National Biobank of Korea under the Korea Centers for Disease Control and Prevention, the present inventors obtained data samples of 289 Parkinson's disease patients aged 65 years or older. A total of 289 Parkinson's disease patients consisted of 179 Parkinson's disease-mild cognitive impairment (PD-MCI) patients and 110 Parkinson's disease-dementia (PDD) patients.

Next, the present inventors constructed multidimensional data for regression analysis based on the obtained data. Specifically, the present inventors constructed multidimensional data by setting the dependent variable as the cognitive impairment type of Parkinson's disease (i.e., PDD-MCI and PDD) and the results of various neuropsychological tests related to Parkinson's disease as independent variables. Please refer to Table 1 below for the set independent variables.

Next, the present inventors performed an optimization scale regression analysis based on the linear regression model as described above. The alternating least squares method was used to find the optimal scale for each variable, and for independent variables with ordinal scales such as K-MMSE and H&Y stages, calculations were performed to find the optimal scale under the constraint of maintaining order. The results of the regression analysis are shown in Table 1 below.

In Table 1 below, 'b' means the regression coefficient, 'SE' means the standard error, 'df' means the degree of freedom, and 'F' means the F-statistic. Additionally, 'P' refers to the probability of significance (P-value). Anyone working in the relevant technical field will already be familiar with the meaning and calculation method of each statistical value obtained as a result of regression analysis, so description thereof will be omitted.

independent variable b S.E. df F P K-MMSE -.522 .168 2 9.684 <0.001 KMoCA -.206 .238 3 .750 .527 CDR-GS .127 .269 One .222 .639 CDR-SB -.271 .412 3 .431 .732 K-IADL .237 .224 2 1.119 .334 UPDRS total .433 .444 3 .949 .423 UPDRS motor -.338 .330 3 1.045 .380 H&Y stage .440 .197 3 5.008 .004 S&E ADL .353 .333 2 1.123 .333

Referring to Table 1, among the plurality of independent variables, the size (e.g. absolute value) of the regression coefficient (b) for 'K-MMSE' and 'H&Y Stage' was found to be relatively large, and the standard error (SE) was found to be relatively small. This means that 'K-MMSE' and 'H&Y stage' have a greater correlation with the dependent variable than other independent variables, and can be understood as indicating that they are effective variables for early diagnosis of Parkinson's disease-dementia (PDD).

Tables 2 and 3 below show the optimal scale for K-MMSE and H&Y stage test results among the independent variables. The quantification index below refers to the value converted through the optimal scaling method, and the larger the negative value, the greater the impact on the dependent variable (i.e. Parkinson's disease-dementia).

K-MMSE Quantification Index 3-14 -1.260 15-18 -1.198 19-20 -1.013 21-22 -.706 23-24 -.320 25 .135 26 .656 28 1.508 29-30 1.616

Referring to Table 2, the K-MMSE test result is a variable with an equal interval scale and a perfect score of 30 (i.e., the total number of score intervals is 30), but when the optimal scale is obtained, the number of intervals changes to about 10. , it can be seen that the degree of change is quite large. This means that if regression analysis is performed using data from variables with equal interval scales, the accuracy of the analysis can be greatly reduced. Conversely, it means that the accuracy of regression analysis can be greatly improved by finding the optimal scale. .

In addition, it can be seen that the sign of the quantification index changes when the K-MMSE test result is about 23 to 25 points, which means that if the K-MMSE test result is about 25 points or more, the risk of developing Parkinson's disease-dementia (PDD) is higher. is very low, and a score of about 23 or less means that the risk of developing Parkinson's disease-dementia (PDD) increases significantly. Therefore, when diagnosing the cognitive impairment type of Parkinson's disease or the risk of progressing to Parkinson's disease-dementia (PDD) using the K-MMSE test results, it is desirable to use any one of about 23 to 25 points as the reference value. You can see that

H&Y Stage Quantification Index 1.0 -2.787 1.5 -.609 2.0 -.187 2.5 -.081 3.0 .151 4.0 1.179 5.0 2.167

Referring to Table 3, the test results of the H&Y stage are also variables with an equal interval scale, but when the optimal scale is obtained, the number of sections changes to about 7. In addition, it can be seen that the sign of the quantification index changes when the H&Y stage test result is between about 2.0 and 3.0, which means that if the H&Y stage test result is about 3.0 or higher, the risk of developing Parkinson's disease-dementia (PDD) is very low. , below about 2.0 means that the risk of developing Parkinson's disease-dementia (PDD) increases significantly. Therefore, when diagnosing the cognitive impairment type of Parkinson's disease or the risk of progressing to Parkinson's disease-dementia (PDD) using the H&Y stage test results, it can be seen that it is desirable to use any one of about 2.0 to 3.0 as the reference value. there is.

The results of experiments conducted by the present inventors so far have been briefly introduced.

The technical idea of the present disclosure described so far with reference to FIGS. 1 to 10 may be implemented as computer-readable code on a computer-readable medium. The computer-readable recording medium may be, for example, a removable recording medium (CD, DVD, Blu-ray disk, USB storage device, removable hard disk) or a fixed recording medium (ROM, RAM, computer-equipped hard disk). You can. The computer program recorded on the computer-readable recording medium can be transmitted to another computing device through a network such as the Internet, installed on the other computing device, and thus used on the other computing device.

In the above, even though all the components constituting the embodiments of the present disclosure have been described as being combined or operated in combination, the technical idea of the present disclosure is not necessarily limited to these embodiments. That is, within the scope of the purpose of the present disclosure, all of the components may be operated by selectively combining one or more of them.

Although operations are shown in the drawings in a specific order, it should not be understood that the operations must be performed in the specific order shown or sequential order or that all illustrated operations must be performed to obtain the desired results. In certain situations, multitasking and parallel processing may be advantageous. Moreover, the separation of the various components in the embodiments described above should not be construed as necessarily requiring such separation, and the program components and systems described may generally be integrated together into a single software product or packaged into multiple software products. You must understand that it exists.

Although embodiments of the present disclosure have been described above with reference to the attached drawings, those skilled in the art will understand that the present disclosure can be implemented in other specific forms without changing the technical idea or essential features. I can understand that there is. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. The scope of protection of this disclosure should be interpreted in accordance with the claims below, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of rights of the technical ideas defined by this disclosure.

Claims (12)

A method of diagnosing Parkinson's disease performed by at least one computing device, comprising:
Obtaining multidimensional data on a Parkinson's disease patient - A plurality of independent variables constituting the multidimensional data include variables related to neuropsychological test results of the Parkinson's disease patient, and the dependent variable of the multidimensional data is a Parkinson's disease patient. It relates to types of cognitive impairment, and the cognitive impairment types include Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD) -;
performing optimal scaling regression analysis on the obtained multidimensional data to determine a main independent variable among the plurality of independent variables;
Constructing a diagnostic model for diagnosing the type of cognitive impairment of Parkinson's disease or the risk of developing Parkinson's disease-dementia (PDD) using the determined main independent variables; and
Comprising the step of performing a diagnosis of Parkinson's disease using the constructed diagnostic model,
How to diagnose Parkinson's disease. According to paragraph 1,
The plurality of independent variables include GDS (Global Deterioration Scale), H&Y stage (Hoehn and Yahr stage), CDR (Clinical Dementia Rating), S&E ADL (Schwab & England Activities of Daily Living), and K-IADL (the Korean Instrumental Activities of Daily Living). Daily Living), Unified Parkinson's Disease Rating Scale (UPDRS), Korean Mini-Mental State Examination (K-MMSE), and Korean-Montreal Cognitive Assessment (K-MoCA) test results,
How to diagnose Parkinson's disease. According to paragraph 1,
The main independent variables include variables related to the Korean Mini-Mental State Examination (K-MMSE) and the Hoehn and Yahr stage (H&Y stage) test results.
How to diagnose Parkinson's disease. According to paragraph 3,
If the K-MMSE test result is below the standard value, the diagnostic model diagnoses that the diagnosed person's Parkinson's disease cognitive impairment type corresponds to Parkinson's disease-dementia (PDD), or that the diagnosed person's condition corresponds to Parkinson's disease-dementia (PDD). ) is a model that diagnoses that there is a risk of progression to
The reference value is any one of 23 to 25 points,
How to diagnose Parkinson's disease. According to paragraph 3,
If the H&Y stage test result is below the standard value, the diagnostic model diagnoses that the diagnosed person's Parkinson's disease cognitive impairment type corresponds to Parkinson's disease-dementia (PDD), or that the diagnosed person's condition corresponds to Parkinson's disease-dementia (PDD). It is a model that diagnoses that there is a risk of progression to
The reference value is any one of 2.0 to 3.0,
How to diagnose Parkinson's disease. According to paragraph 1,
The optimization scale regression analysis is performed based on a linear regression model,
How to diagnose Parkinson's disease. According to paragraph 1,
The step of determining the main independent variables is,
calculating regression coefficients for the plurality of independent variables by performing the optimization scale regression analysis; and
Comprising the step of determining the main independent variable among the plurality of independent variables based on the calculated regression coefficient,
How to diagnose Parkinson's disease. According to paragraph 1,
The plurality of independent variables include a first variable having an ordinal scale and an interval scale,
The step of determining the main independent variables is,
While maintaining the order of the first variable, changing the number of sections, the size of the section, or the value corresponding to the section by using alternating least squares. ,
How to diagnose Parkinson's disease. According to paragraph 1,
The step of acquiring the multidimensional data is,
Obtaining original multidimensional data including a plurality of data samples for the Parkinson's disease patient;
Obtaining a deep learning model based on an autoencoder learned with a training set for normal people;
calculating a reconstruction loss for the plurality of data samples through the acquired deep learning model; and
Including removing samples in which the calculated restoration loss is less than or equal to a reference value from the original multidimensional data,
How to diagnose Parkinson's disease. According to paragraph 1,
The diagnostic model includes an encoder, decoder and classifier,
The classifier is configured to receive a latent vector extracted through the encoder and predict the type of cognitive impairment or the risk,
The step of building the diagnostic model is,
Generating a training set using data on the determined main independent variables and cognitive impairment type data of the Parkinson's disease patient among the acquired multidimensional data; and
Comprising the step of learning the diagnostic model using the generated training set,
The step of learning the diagnostic model is,
training the encoder and the decoder using a reconstruction loss; and
In a state where the decoder is frozen, further training the classifier and the encoder using the classification loss calculated through the classifier,
How to diagnose Parkinson's disease. One or more processors; and
Includes memory for storing one or more instructions,
The one or more processors:
By executing one or more instructions stored above,
An operation of acquiring multidimensional data on a Parkinson's disease patient - a plurality of independent variables constituting the multidimensional data include variables related to neuropsychological test results of the Parkinson's disease patient, and the dependent variable of the multidimensional data is the Parkinson's disease patient. It relates to types of cognitive impairment, and the cognitive impairment types include Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD) -,
An operation of performing optimal scaling regression analysis on the acquired multidimensional data to determine a main independent variable among the plurality of independent variables;
An operation to construct a diagnostic model to diagnose the type of cognitive impairment of Parkinson's disease or the risk of progressing to Parkinson's disease-dementia (PDD) using the main independent variables determined above, and
Performing an operation to diagnose Parkinson's disease using the constructed diagnostic model,
Parkinson's disease diagnosis system. Combined with a computing device,
Obtaining multidimensional data on a Parkinson's disease patient - A plurality of independent variables constituting the multidimensional data include variables related to neuropsychological test results of the Parkinson's disease patient, and the dependent variable of the multidimensional data is a Parkinson's disease patient. It relates to types of cognitive impairment, and the cognitive impairment types include Parkinson's disease-mild cognitive impairment (PD-MCI) and Parkinson's disease-dementia (PDD) -;
performing optimal scaling regression analysis on the obtained multidimensional data to determine a main independent variable among the plurality of independent variables;
Constructing a diagnostic model for diagnosing the type of cognitive impairment of Parkinson's disease or the risk of developing Parkinson's disease-dementia (PDD) using the determined main independent variables; and
Stored in a computer-readable recording medium to execute the step of performing a diagnosis of Parkinson's disease using the constructed diagnostic model,
computer program.