KR20230097332A - System for early diagnosis of dementia and Method for early diagnosis of dementia using the same - Google Patents

Abstract

According to one aspect of the present invention, according to one aspect of the present invention, the VR task is displayed by the virtual reality providing device (step S1); Collecting biometric data, behavioral data, and cognitive data while the patient is performing the VR task by a data collection device (step S2); Determining whether or not there is dementia by inputting the biometric data, behavioral data, and cognitive data to the multimodal deep learning model learned by the dementia early diagnosis server (step S3); and providing an explanatory power for the existence of the dementia by an early dementia diagnosis server (Step S4).

Description

System for early diagnosis of dementia and method for early diagnosis of dementia using the same {System for early diagnosis of dementia and Method for early diagnosis of dementia using the same}

The present invention relates to an early diagnosis system for dementia and a method for early diagnosis of dementia using the same.

Dementia is an irreversible disease, and prevention through early diagnosis is key. According to data from the National Statistical Office in 2020, the prevalence of dementia in the elderly population aged 65 or older in Korea is 10.29%, and the annual national dementia management cost is about 16.5 trillion won. These costs are estimated to increase to about 63.1 trillion won in 2040.

Advanced countries such as the United States and the European Union are accelerating research and development of artificial intelligence-based VR technology for early diagnosis of dementia, MRI analysis technology, and tone analysis technology. Compared to existing questionnaire-based cognitive ability evaluation (e.g., SNSB) and invasive amyloid beta test methods that take more than 2 hours, AI-based medical technology significantly reduces the time required for early diagnosis of dementia and improves accuracy, thereby improving not only medical staff but also It is expected to increase patient satisfaction.

Instrumental activities of daily living are key indicators for early diagnosis of dementia. To this end, questionnaire-based cognitive ability evaluation (eg SNSB) is being performed.

However, questionnaire-based early diagnosis of dementia takes a lot of time, is affected by demographic factors such as the subject's age and years of education, and has limitations that the results may be distorted by the subject's subjective bias.

As a result, questionnaire-based evaluation technology is used only as a screening index, and additional tests such as EEG and MRI are required for actual early diagnosis of dementia.

Recently, research is being conducted on a methodology that implements virtual daily life using VR technology and directly evaluates the instrumental daily living performance ability. In the VR environment, daily life tasks such as driving, shopping, using the bus, and money management are implemented, and the collected features such as hand movements, head movements, and performance (execution time, number of errors) are used for machine learning. ) to diagnose dementia early.

However, this methodology cannot analyze the subject's complex behavioral patterns and does not provide sufficient explanatory power for the relationship between daily life performance and dementia.

Patent Document: Korean Registered Patent No. 10-2282101 B1 (2021.07.27)

Embodiments of the present invention have been proposed to solve the above problems, and are intended to provide an early diagnosis system for dementia (1) capable of early diagnosis of Alzheimer's dementia and a method for early diagnosis of dementia using the same.

In addition, it is intended to provide an early diagnosis system for dementia and a method for early diagnosis of dementia using the same, which can provide explanatory power for what reason it was determined to be dementia.

In addition, it is intended to provide an early diagnosis system for dementia and a method for early diagnosis of dementia using the same, which can perform examinations on a large number of subjects at the same time, shorten the examination time, and derive highly accurate test results.

In addition, it is intended to provide an early diagnosis system for dementia that can diagnose Alzheimer's dementia without visiting a hospital and an early diagnosis method for dementia using the same.

According to one embodiment of the present invention, according to one aspect of the present invention, the VR task is displayed by the virtual reality providing apparatus 20 (step S1); Collecting biometric data, behavioral data, and cognitive data by the data collection device 30 while the subject to be diagnosed performs the VR task (step S2); Determining whether or not there is dementia by inputting the biometric data, behavioral data, and cognitive data to the multimodal deep learning model learned by the dementia early diagnosis server 10 (step S3); and providing explanatory power on the existence of dementia by the early dementia diagnosis server 10 (Step S4).

In addition, the VR task in the step S1 is controlled by the VR task setting module 310, and the VR task controlled by the VR task setting module 310 includes a task of ordering a menu through a kiosk and dementia Early diagnosis methods can be provided.

In addition, the task of ordering a menu through the kiosk is a first task of selecting a dining place, a second task of selecting a type of main meal, a third task of selecting a side menu, a fourth task of selecting a beverage, A method for early diagnosis of dementia including a fifth task of selecting a payment method and a sixth task of inputting a password may be provided.

In addition, the step S3 includes building a multimodal deep learning model learned by the dementia diagnosis module 320 of the dementia early diagnosis server 10, and the learned multimodal deep learning model is classified into different taxa. A method for early diagnosis of dementia may be provided, which includes constructing time-series classification of biological data, behavioral data, and cognitive data of a subject to be diagnosed, and training the convolutional neural network (CNN) model.

In addition, in the step S3, the heart rate and skin conductance constituting the biometric data, and the 3-axis hand motion, 3-axis head motion, and 3-axis eye movement constituting the behavioral data, and the number of errors and execution time constituting the cognitive data After this is input to the multimodal deep learning model, after learning through a convolution layer, batch normalization, and an action layer, the diagnosis subject is evaluated through global average pooling (GAP). A method for early diagnosis of dementia including determining whether a person has dementia may be provided.

In addition, the step S4 includes a step of mapping the behavior of the diagnosis subject to a general cognitive function (GCF) through a class activation map (CAM) by the explanatory power providing module 330, and the general cognitive function (GCF) Attention, which means the ability to pay attention to and focus on stimuli and information in the surrounding environment, memory, which means self-regulation and memory abilities used in daily life, and mentally Psychomotor speed, which means the speed at which you perform, problem solving, which is a systematic and rational problem-solving ability, and language, which is a necessary language ability for communication with others, can be classified. A method for early diagnosis of dementia may be provided.

According to one aspect of the present invention, a virtual reality providing device capable of displaying a VR task 20; a data collection device 30 capable of collecting biometric data, behavioral data, and cognitive data while the diagnosis subject performs the VR task; and an early dementia diagnosis server 10 capable of inputting the biometric data, behavioral data, and cognitive data into the learned multimodal deep learning model to determine whether or not there is dementia, and to provide an explanatory power for the presence or absence of dementia. system can be provided.

In addition, the dementia early diagnosis server 10 includes a VR task setting module 310 that controls the VR task displayed by the virtual reality providing device 20; Dementia diagnosis module 320 capable of building a learned multimodal deep learning model; And an explanatory power providing module 330 capable of mapping the behavior of the subject to diagnosis while performing the VR task to a general cognitive function (GCF) through a class activation map (CAM). An early diagnosis system for dementia may be provided. there is.

The early diagnosis system for dementia and the method for early diagnosis of dementia using the system according to embodiments of the present invention can diagnose Alzheimer's dementia at an early stage.

In addition, it is possible to provide an explanatory power for what reason was determined to be dementia.

In addition, it is possible to simultaneously perform tests on a plurality of subjects to be diagnosed, shorten the test time, and derive highly accurate test results.

In addition, it is possible to diagnose Alzheimer's dementia without visiting a hospital.

1 is a diagram schematically illustrating an early dementia diagnosis system 1 according to an embodiment of the present invention.
FIG. 2 is a diagram schematically illustrating the configuration of the early dementia diagnosis server 10 of FIG. 1 .
FIG. 3 is a flowchart schematically illustrating a method for early diagnosis of dementia executed by the early diagnosis system for dementia 1 of FIG. 1 .
4 is a diagram showing biometric data (heart rate, skin conductance), behavioral data (3-axis hand movements, 3-axis head movements), and cognitive data (number of errors, execution time) collected while a subject under diagnosis performed a VR task. .
5 shows biometric data, behavioral data, and cognitive data being input into a multimodal deep learning model and then going through global average pooling (GAP) to determine whether a diagnosis subject is within a normal range or a diagnosis subject with mild cognitive impairment. It is a diagram that conceptually shows what is to be.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. In addition, in describing the present invention, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present invention, the detailed description will be omitted.

1 is a diagram schematically illustrating an early dementia diagnosis system 1 according to an embodiment of the present invention.

Referring to FIG. 1 , the early diagnosis system for dementia 1 may include an early dementia diagnosis server 10, a virtual reality providing device 20, a data collection device 30, and a communication network 40.

The early diagnosis system for dementia 1 may be named a VR digital myomarker system for early diagnosis of Alzheimer's dementia, and the early diagnosis server for dementia 10, the virtual reality providing device 20, and the data collection device 30 are connected to each other. It may be provided as an independent device for data communication through the communication network 40, or composed of one physical device and provided for direct data communication.

In this embodiment, an example in which the dementia early diagnosis server 10, the virtual reality providing device 20, and the data collection device 30 are provided as separate and independent devices will be described.

In the early diagnosis system for dementia 1 of the present embodiment, a VR task is presented through the virtual reality providing device 20, and biometric data (eg, heart rate, skin conductance) and behavioral data are presented while the diagnosis subject performs the VR task. (For example, hand 3-axis movement, head 3-axis movement, eyeball 3-axis movement) and cognitive data (number of errors, execution time) are collected, and the collected data are constructed by the dementia early diagnosis server 10 Dementia can be determined by inputting it into the multimodal deep learning model.

In addition, the early dementia diagnosis system (1) can implement complex behavioral pattern feature extraction and model learning through deep learning-based time-series classification, and utilize CAM (class activation map) to It is possible to map the behavioral pattern in GCF (general cognitive function), and based on this, explainability of why the subject was diagnosed with dementia early. A detailed description of this will be given later.

The dementia early diagnosis server 10 displays the VR task on the virtual reality providing device 20, collects data (biological data, behavioral data, cognitive data) while the diagnosis subject performs the VR task, and based on this It can be understood as a server that can diagnose dementia early and provide explanatory power for it. A detailed description of this will be given later.

The virtual reality providing device 20 may include a head-mounted display (HDM) that displays a VR task while the diagnosis subject wears it on their head.

The data collection device 30 may include a biometric data collection device that measures the heart rate and skin conductance of the patient to be diagnosed.

In addition, the data collection device 30 may include a behavior data collection device that measures the 3-axis motion of the hand, the 3-axis head, and the 3-axis eye movement of the subject to be diagnosed. For example, the device for collecting behavioral data is an apparatus mounted on a head-mounted display (HDM) for measuring 3-axis head and 3-axis eye movements, or a device that can be grasped by a user to measure 3-axis hand motion. can include

In addition, the data collection device 30 may include a cognitive data collection unit that measures the number of errors and execution time of the subject to be diagnosed.

Here, the biometric data collection device for measuring heart rate and skin conductance, and the behavioral data collection device for measuring 3-axis hand movement, 3-axis head movement, and 3-axis eye movement may be hardware, and the number of errors and execution time of the subject of diagnosis The cognitive data collection unit for measuring may be software embedded in the virtual reality providing device 20 .

However, the biometric data collection device, the behavioral data collection device, and the cognitive data collection unit are illustrative, and may include other devices for collecting biometric data, behavioral data, and cognitive data.

FIG. 2 is a diagram schematically illustrating the configuration of the early dementia diagnosis server 10 of FIG. 1 .

Referring to FIG. 2 , the dementia early diagnosis server 10 may include a memory 200, a processor 300, and a communication module 400.

The processor 300 may be configured to process commands of a computer program by performing basic arithmetic, logic, and input/output operations. Instructions may be provided to the processor 300 from the memory 200 or the communication module 400 . In addition, commands may be provided to the processor 300 through a communication channel between respective components constituting the dementia early diagnosis server 10 .

The processor 300 sets a VR task, builds a multimodal deep learning model, inputs the biometric data, behavioral data, and cognitive data of the subject to be diagnosed into the multimodal deep learning model to determine whether or not there is dementia, and for some reason. It can provide expainability whether the diagnosed subject was diagnosed with dementia at an early stage. Specific components of the processor 300 for executing this will be described later.

In addition, components of the processor 300 may include an artificial neural network pre-learned through deep learning. For example, at least one of the components of the processor 300 may include a convolutional neural network (CNN), and the convolutional neural network (CNN) is a behavior of a diagnosis subject through time-series classification. It can be constructed by pattern feature extraction.

A multimodal deep learning model can be built through this time-series classification convolutional neural network (CNN), and the collected biometric data, behavioral data, and cognitive data are input to the multimodal deep learning model and raw It is analyzed in the form of data, and complex patterns in it can be analyzed.

The memory 200 is a computer-readable recording medium and may include a random access memory (RAM), a read only memory (ROM), and a permanent mass storage device such as a disk drive. In addition, the processor 300 may load program codes stored in the memory 200 to determine whether or not there is dementia or provide explanatory power. These program codes may be loaded from a recording medium (for example, DVD, memory card, etc.) readable by a separate computer, or transferred from another device through the communication module 400 and stored in the memory 200.

In addition, in the memory 200, a VR task is set, a multi-modal deep learning model is built, time-series data is input into a multi-modal deep learning model to determine whether or not there is dementia, or for some reason, an explanatory function is used to determine whether a person diagnosed with dementia is diagnosed early. A database 210 capable of storing necessary data to provide expainability may be provided.

The communication module 400 may provide a function for communicating with the virtual reality providing device 20 and the data collecting device 30 through the communication network 40 .

The processor 300 of the early dementia diagnosis server 10 may include a VR task setting module 310, a dementia diagnosis module 320, and an explanatory power providing module 330, which are physically configured, which will be described in detail later. do.

3 is a flowchart schematically illustrating a method for early diagnosis of dementia executed by the early diagnosis system for dementia 1 of FIG. 1, and FIG. 4 is a biometric data (heart rate, skin conductance) collected while a diagnosis subject performs a VR task. , It is a diagram showing behavioral data (3-axis movement of the hand, 3-axis movement of the head) and cognitive data (number of errors, execution time), and FIG. Through global average pooling (GAP), it is a diagram conceptually showing that it is determined whether a diagnosis subject is within the normal range or a diagnosis subject with mild cognitive impairment.

Referring to FIGS. 3 to 5 , the method for early diagnosis of dementia includes displaying a VR task by the virtual reality providing device 20 (step S1); Collecting biometric data, behavioral data, and cognitive data by the data collection device 30 while the subject to be diagnosed performs the VR task (step S2); Determining whether or not there is dementia by inputting the biometric data, behavioral data, and cognitive data from the dementia early diagnosis server 10 to the learned multimodal deep learning model (step S3); and providing explanatory power on whether or not the dementia exists by the early dementia diagnosis server 10 (step S4).

First, the step (S1 step) in which the VR task is displayed by the virtual reality providing device 20 will be described in detail.

The VR task displayed on the virtual reality providing device 20 (eg, head-mounted display (HDM)) may be controlled by the VR task setting module 310 .

For example, the VR task setting module 310 may set the VR task as a task of ordering a menu through a kiosk. Specifically, the task of ordering a menu through a kiosk may include performing a plurality of tasks, and the plurality of tasks include a first task of selecting a restaurant and a type of main meal (eg, hamburger). It may include a second task to select a side menu, a third task to select a side menu, a fourth task to select a beverage, a fifth task to select a payment method, and a sixth task to input a password.

However, the type of VR task is exemplary, and the VR task can be customized according to the subject of diagnosis.

Biometric data, behavioral data, and cognitive data may be collected (step S2) by the data collection device 30 while the patient to be diagnosed performs the task of ordering a menu through the kiosk through the virtual reality providing device 20.

Here, biometric data may include heart rate and skin conductance, behavioral data may include hand 3-axis movement, head 3-axis movement, and eyeball 3-axis movement, and cognitive data may include the number of errors and execution time. there is.

4 is a diagram showing biometric data (heart rate, skin conductance), behavioral data (3-axis hand movements, 3-axis head movements), and cognitive data (number of errors, execution time) collected while a subject under diagnosis performed a VR task. .

In this way, the biometric data, behavioral data, and cognitive data collected by the data collection device 30 may be transmitted to the dementia early diagnosis server 10 .

After that, the biometric data, behavioral data, and cognitive data are input to the multimodal deep learning model learned by the dementia early diagnosis server 10, and a step (S3 step) of determining whether or not there is dementia can be executed. The explanation is as follows.

First, the dementia diagnosis module 320 of the dementia early diagnosis server 10 may include a learned multimodal deep learning model.

At this time, the learned multimodal deep learning model is applied to diagnosis subjects belonging to different taxa (e.g., N subjects in the normal range, N subjects with mild cognitive impairment, and N subjects suffering from dementia). First-order time-series classification is performed by training each of biometric data, behavioral data, and cognitive data with a convolutional neural network (CNN) model, and overlapping time-series data classification is performed by combining them to perform second-order time-series data classification. (Nested time-series classification) is performed.

More specifically, the heart rate, skin conductance, 3-axis movement of the hand, 3-axis movement of the head, 3-axis movement of the eye, number of errors, and execution time constituting biometric data, behavioral data, and cognitive data are time-series classification. After being input into the multimodal deep learning model, temporal characteristics in the data are extracted again through a convolution layer, and bias in the time series data through batch normalization and an action layer bias), and based on the final result of calculating the overall contextual feature of the time series data through global average pooling (GAP), the diagnosis subject is in which taxon (for example, within the normal range) It can be predicted with a probability whether a person belongs to a diagnosis subject, a diagnosis subject with mild cognitive impairment, or a diagnosis subject suffering from dementia.

Here, a convolution layer, batch normalization, action layer, and global average pooling (GAP) may be executed by the dementia diagnosis module 320 .

5 is a diagram showing whether the above-described biometric data, behavioral data, and cognitive data are entered into a multimodal deep learning model and then subjected to global average pooling (GAP) to determine whether the diagnosis subject is within the normal range or diagnosed with mild cognitive impairment. It is a diagram conceptually showing that recognition is determined.

Thereafter, a step (step S4) of providing explanatory power on the presence or absence of dementia may be executed by the early dementia diagnosis server 10, and a detailed description thereof is as follows.

In step S4, the behavior of the subject to be diagnosed while performing the VR task by the explanatory power providing module 330 of the early dementia diagnosis server 10 can be mapped to general cognitive function (GCF) through a class activation map (CAM). there is. Through this, it is possible to provide explanatory power for why the diagnosis subject was diagnosed with dementia early.

Specifically, GCF (general cognitive function) is attention, which refers to the ability to pay attention to stimuli and information in the surrounding environment, and memory, which refers to self-regulation and memory abilities used in daily life. , psychomotor speed, which means the speed at which a given piece of information or task is mentally performed, problem solving, which is the ability to systematically and rationally solve problems, and language skills, which are necessary for daily life and communication with others. May include language.

Here, information on attention, memory, psychomotor speed, problem solving, and language, which are specific information on GCF (general cognitive function), is stored in a database (210). ) may be stored in

As described above, the dementia early diagnosis server 10 may include a VR task setting module 310, a dementia diagnosis module 320, and an explanatory power providing module 330, which are physical components. As follows.

The VR task setting module 310 may control the VR task displayed by the virtual reality providing device 20 .

For example, the VR task setting module 310 may set the VR task as a task of ordering a menu through a kiosk. Specifically, the task of ordering a menu through a kiosk may include performing a plurality of tasks, and the plurality of tasks include a first task of selecting a restaurant and a type of main meal (eg, hamburger). It may include a second task to select a side menu, a third task to select a side menu, a fourth task to select a beverage, a fifth task to select a payment method, and a sixth task to input a password.

The dementia diagnosis module 320 can build a learned multimodal deep learning model, and the learned multimodal deep learning model is a diagnosis subject belonging to different taxa (for example, N persons diagnosed in the normal range, N Time-series classification of biological data, behavioral data, and cognition data for 100 people diagnosed with mild cognitive impairment and N people diagnosed with dementia), and by training the convolutional neural network (CNN) model can be built

The explanatory power providing module 330 can map the behavior of the subject to diagnosis while performing the VR task to the general cognitive function (GCF) through a class activation map (CAM), through which the subject to be diagnosed can be diagnosed with dementia early for some reason. Can provide an explanation of what happened.

The early dementia diagnosis system 1 of this embodiment analyzes the bio-behavior-cognitive time-series data of the subject in the VR environment, extracts features through deep learning-based time-series classification, model learning, and utilizes CAM It is original in terms of providing explanatory power through subject movement and GCF (general cognitive function) mapping in a VR task.

In addition, the early diagnosis system for dementia 1 and the method for early diagnosis of dementia using the same can be applied to the continuously growing field of early diagnosis of dementia and VR medical field.

Compared to conventional questionnaire-based examinations, MRI examinations, etc., VR examination can be applied to the actual field because it can be performed by several people at the same time, reduce time with quantified data, and obtain highly accurate examination results. In addition, VR can be used regardless of time and place, and has a high sense of immersion based on virtual reality. Therefore, it is easy to apply and evaluate instrumental daily abilities even in situations where you feel reluctance to visit a hospital due to COVID-19 or in a situation where you cannot visit a hospital.

In addition, as VR, which was consumed as cultural and leisure content, was brought into the medical field, VR played many roles such as diagnosis and treatment of diseases. By diagnosing dementia, which is emerging as a social problem, at an early stage, it solves social and economic problems by reducing the high prevalence of dementia and the enormous dementia management cost.

In addition, as the use of artificial intelligence expands in the medical field, it is possible to provide better medical services by overcoming the disconnected structure and data silos by department that hinder smooth communication in the medical environment and by increasing the efficiency of hospital operation and patient management and treatment. .

This technology allows the subject to get rid of the objection of the dementia test and take the test as a play. This can help the elderly to apply the use learned in the virtual environment to the real environment by indirectly experiencing situations that the elderly avoid in daily life, such as using a kiosk, through VR.

We will develop this technology in conjunction with EEG and MRI so that it can be applied in clinical practice, secure quality, safety, and effectiveness, and develop it into a safe and reliable diagnosis with permission from the Ministry of Food and Drug Safety. Also, taking advantage of the portability of VR technology, not only VR can be installed in public health centers, but also home care is possible, and rehabilitation programs such as dementia daily care service can be operated regularly.

Although the early diagnosis system for dementia and the method for early diagnosis of dementia using the same according to an embodiment of the present invention have been described as specific embodiments, this is only an example, and the present invention is not limited thereto. should be construed as having the widest scope. A person skilled in the art may implement a pattern of a shape not indicated by combining or substituting the disclosed embodiments, but this also does not deviate from the scope of the present invention. In addition, those skilled in the art can easily change or modify the disclosed embodiments based on this specification, and it is clear that such changes or modifications also fall within the scope of the present invention.

1: Early diagnosis system for dementia
10: Dementia Early Diagnosis Server
20: virtual reality providing device
200: memory
210: database
30: data collection device
300: processor
310: VR task setting module
320: dementia diagnosis module
330: explanatory power providing module

Claims (5)

The step of displaying the VR task by the virtual reality providing device (S1 step);
Collecting biometric data, behavioral data, and cognitive data while the patient is performing the VR task by a data collection device (step S2);
Determining whether or not there is dementia by inputting the biometric data, behavioral data, and cognitive data to the multimodal deep learning model learned by the dementia early diagnosis server (step S3); and
Including the step (S4 step) of providing explanatory power for the dementia or not by the dementia early diagnosis server
Methods for early diagnosis of dementia. According to claim 1,
The VR task in the step S1 is controlled by the VR task setting module,
The VR task controlled by the VR task setting module includes a task of ordering a menu through a kiosk,
The task of ordering a menu through the kiosk is,
The first task is to select a place to eat, the second task is to select the type of main meal, the third task is to select a side menu, the fourth task is to select a beverage, the fifth task is to select a payment method, and the password is entered. Including the 6th task
Methods for early diagnosis of dementia. According to claim 1,
In step S3,
Including building a multimodal deep learning model learned by the dementia diagnosis module of the dementia early diagnosis server,
The learned multimodal deep learning model is constructed by classifying biological data, behavioral data, and cognitive data of subjects belonging to different taxa as time series data and learning the convolutional neural network model
Methods for early diagnosis of dementia. According to claim 3,
In step S3,
The heart rate, skin conductance constituting the biometric data, the 3-axis movement of the hand, the 3-axis head, and the 3-axis movement of the eye constituting the behavioral data, and the error count and execution time constituting the cognitive data are the multimodal deep learning model. After being input into , learning through a convolution layer, batch normalization, and an action layer, and then determining whether the diagnosis subject has dementia through global average pooling (GAP).
Methods for early diagnosis of dementia. According to claim 1,
In step S4,
Mapping the behavior of the diagnosed subject to a general cognitive function (GCF) through a class activation map (CAM) by an explanatory power providing module,
The GCF (general cognitive function) refers to concentration, which refers to the ability to pay attention to and focus on stimuli and information in the surrounding environment, memory, which refers to self-regulation and memory abilities used in daily life, and mental ability to process given information or tasks. It can be classified into psychomotor speed, which means the speed of performance, problem solving, which is systematic and rational problem-solving ability, and language, which is necessary for communication with others.
Methods for early diagnosis of dementia.

Images