KR102154763B1 - System and method for predicting the development of dementia using smart devices - Google Patents

Abstract

The present invention relates to a dementia occurrence prediction system. The dementia occurrence prediction system includes: a plurality of monitoring collection devices collecting condition data of a user including living data, bio-signal data and behavior data of the user; a communication part obtaining the condition data collected from the plurality of monitoring collection devices through a network; and a dementia prediction device predicting the dementia occurrence of the user by applying the condition data to an artificial intelligence-based learning model. The dementia prediction device derives a feature variable related with dementia based on the condition data, and predicts the dementia occurrence of the user based on a relation between the derived feature variable and the probability of dementia occurrence. Therefore, the dementia occurrence prediction system is capable of enabling self-monitoring for population-based health improvement.

Description

A system and method for predicting onset of dementia using a smart device {SYSTEM AND METHOD FOR PREDICTING THE DEVELOPMENT OF DEMENTIA USING SMART DEVICES}

The present application relates to a system and method for predicting the onset of dementia using a smart device.

The prevalence of dementia among the elderly over 65 in Korea was 9.2% (540,000) in 2012, and 1.27 million in 2030 and 2.7 million in 2050, which is predicted to double every 20 years.

The average annual economic cost per person with dementia in Korea was 18.51 million won as of 2010, and the total medical expenses due to dementia increased from about 41.5 billion won in 2004 to about 4528 billion won in 2009 and about 1.45 trillion won in 2014. This is very big.

In terms of public interest, domestic U-health services are partially under way by the government, but the medical law has not been revised, the insurance fee has not been developed, the technical standard has not been established, the lack of experts, lack of basic statistics, business models, etc. It is difficult to revitalize it in earnest as this cannot be derived.

In the case of a dementia patient management service using an existing smart device, it is a device for protecting and caring for patients diagnosed with dementia. The existing dementia patient management service using a smart device provides functions such as sending a notification to the guardian when they leave the safe area set by the guardian, and a roaming notification and a rescue request when they wander without a purpose. It is necessary to develop a smart device capable of self-monitoring targeting high-risk groups before diagnosis of dementia.

The technology behind the present application is disclosed in Korean Patent Publication No. 10-1951797.

The present application is to solve the problems of the prior art described above, and collects predictive factors for dementia that can be collected in daily life using various smart sensors, and designs a model of the signal acquired through the sensor based on the collected data. , It aims to provide a system and method for predicting the onset of dementia using a smart device that can predict and manage dementia by building a platform that can analyze the collected data, and using intelligent learning algorithms.

However, the technical problem to be achieved by the embodiments of the present application is not limited to the technical problems as described above, and other technical problems may exist.

As a technical means for achieving the above-described technical problem, the dementia outbreak prediction system according to an embodiment of the present application includes a plurality of monitoring collections that collect user's state data including user's life data, biometric signal data, and behavior data. A device, a communication unit that obtains the state data collected from the plurality of monitoring and collection devices through a network, and a dementia prediction device that predicts the onset of dementia of a user by applying the state data to an artificial intelligence-based learning model, wherein the The device for predicting dementia may derive a feature variable associated with dementia based on the state data, and predict the user's onset of dementia based on a relationship between the derived feature variable and the likelihood of dementia.

In addition, the dementia prediction apparatus includes a preprocessor for performing preprocessing by applying the obtained state data to a preprocessing algorithm, a feature variable extraction unit for extracting a feature variable from the preprocessed state data, and the invention of dementia from the extracted feature variables. A principal component analysis unit that obtains a principal component variable necessary for prediction, and a prediction unit that predicts the onset of dementia by applying the principal component variable obtained by the principal component analysis unit to a dementia prediction model.

In addition, the dementia onset prediction system may further include a performance evaluation unit that evaluates the performance of the dementia prediction model based on a medical diagnosis result and a prediction result of the dementia prediction model.

In addition, the communication unit includes a first communication unit that communicates with a first monitoring collection device of the plurality of monitoring collection devices through a network, and a second communication unit that communicates with a second monitoring collection device of the plurality of monitoring collection devices through a network. Including, wherein the first communication unit stores the data obtained from the first monitoring and collecting device in response to a preset time and transmits the data to the dementia prediction device, and the second communication unit includes the Data collected from the second monitoring collection device may be acquired.

In addition, the plurality of monitoring and collecting devices may include a user terminal that collects life data of the user, a wearable terminal that is worn on a part of the user's body to obtain the biosignal data, and is provided indoors and outdoors in an area where the user is located. It may include a sensor unit that collects the user's behavior data from the sensor.

In addition, the communication unit includes a storage unit for storing the state data obtained from the plurality of monitoring and collection devices, wherein the communication unit stores the state data in the storage unit for a preset time and then transmits the data to the dementia prediction device. can do.

According to an embodiment of the present application, a method for predicting the onset of dementia includes the steps of collecting, by a plurality of monitoring and collecting devices, user's state data including the user's life data, biosignal data, and behavioral data, from the plurality of monitoring and collecting devices. Acquiring the collected state data through a network and predicting the onset of dementia of the user by applying the state data to an artificial intelligence-based learning model, wherein the step of predicting the onset of dementia includes the state data Based on, a characteristic variable related to dementia may be derived, and the onset of dementia of the user may be predicted based on a relationship between the derived characteristic variable and the likelihood of developing dementia.

In addition, the step of predicting the onset of dementia may include performing pre-processing by applying the obtained state data to a pre-processing algorithm, extracting a feature variable from the pre-processed state data, and predicting the invention of dementia from the extracted feature variable. It may include obtaining a principal component variable required for and predicting the onset of dementia by applying the principal component variable to a dementia prediction model.

The above-described problem solving means are merely exemplary and should not be construed as limiting the present application. In addition to the above-described exemplary embodiments, additional embodiments may exist in the drawings and detailed description of the invention.

According to the above-described problem solving means of the present application, there is an effect that self-monitoring for population-based health promotion is possible through the development of a smart device reflecting the risk factors of dementia.

According to the above-described problem solving means of the present application, it can be used as a mechanism for treatment and management as well as prevention of dementia, and the consumer himself/herself manages health-related data for future PHR (Personal Health Record) System. It is predicted that the foundation will be worth researching.

However, the effect obtainable in the present application is not limited to the effects as described above, and other effects may exist.

1 is a schematic configuration diagram of a system for predicting onset of dementia according to an embodiment of the present application.
2 is a schematic block diagram of an apparatus for predicting dementia according to an embodiment of the present application.
3 is a diagram illustrating a model for analyzing predictive variables of dementia of a device for predicting dementia according to an exemplary embodiment of the present disclosure.
4 is a diagram illustrating a process of obtaining a feature value using a first sensor of a system for predicting onset of dementia according to an embodiment of the present application.
5 is an operation flowchart of a method for predicting onset of dementia according to an embodiment of the present application.
6 is an operation flowchart of an embodiment of a method for predicting the onset of dementia according to an embodiment of the present application.

Hereinafter, exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present application. However, the present application may be implemented in various different forms and is not limited to the embodiments described herein. In addition, in the drawings, parts not related to the description are omitted in order to clearly describe the present application, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the present specification, when a part is said to be "connected" with another part, it is not only "directly connected", but also "electrically connected" or "indirectly connected" with another element interposed therebetween. "Including the case.

Throughout this specification, when a member is positioned "on", "upper", "upper", "under", "lower", and "lower" of another member, this means that a member is located on another member. It includes not only the case of contacting but also the case of another member being present between the two members.

Throughout the specification of the present application, when a certain part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

We use smart devices to accurately collect dementia-related data related to movements and bio-signals in daily life that subjects are not aware of, and build a platform that can analyze the collected data to collect, analyze, and provide dementia prediction data. It relates to a system for predicting the onset of dementia to enable it.

This is an integrated monitoring and analysis system that uses a wearable sensor and a fixed motion sensor to obtain characteristic variables related to dementia and probability values of onset probability from information obtained through daily life information and bio-signal information monitoring of potential dementia patients. .

1 is a schematic configuration diagram of a dementia onset prediction system 1 according to an embodiment of the present application.

Referring to FIG. 1, the dementia incidence prediction system 1 may include a dementia prediction device 10, a communication unit 20, and a plurality of monitoring and collection devices 30. However, the configuration of the dementia onset prediction system 1 is not limited thereto. For example, the dementia invention prediction system 1 may include an external server (not shown). The external server (not shown) may include a hospital server, a public institution server (eg, a police station, a fire station, a ward office, a town office, etc.). In addition, the dementia onset prediction system 1 may include a guardian terminal (not shown).

According to an exemplary embodiment of the present disclosure, the plurality of monitoring collection devices 30 may collect user's state data including user's life data, biometric signal data, and behavior data. The plurality of monitoring collection devices 30 may be devices that collect a user's life pattern, a living environment, and a bio signal. The plurality of monitoring and collection devices 30 may collect user life data and biometric signal data indoors and outdoors in order to collect indirect data related to a risk factor of dementia. The plurality of monitoring collection devices 30 may be provided so that a blind spot is not included in an area where the user lives (eg, a house, a company, etc.).

As an example, referring to FIG. 1, the plurality of monitoring collection devices 30 may include a user terminal 31, a wearable terminal 32, and a sensor unit 33.

According to the exemplary embodiment of the present application, the user terminal 31 may collect life data of the user. The user's life data may be data related to the user's use of the mobile phone. As an example, the device for predicting dementia 10 may provide a menu for collecting life data to the user terminal 31. For example, the user terminal 31 downloads and installs an application program provided by the dementia prediction apparatus 10, and a life data collection menu may be provided through the installed application. The dementia prediction apparatus 10 may include all types of servers, terminals, or devices that transmit and receive data, contents, and various communication signals to and from the user terminal 31 through a network, and store and process data.

In addition, the user terminal 31 may provide user information. In other words, the user terminal 31 may provide user information through an application provided by the apparatus 10 for predicting dementia. For example, the user information may include the user's gender, age, height, weight, race, nationality, smoking amount, alcohol consumption, family history, occupation, lifestyle habits, exercise status, personal disease-related history, drug use status information, etc. have.

The user terminal 31 and the guardian terminal (not shown) are devices that are interlocked with the dementia prediction apparatus 10 through a network, and include, for example, a smartphone, a smart pad, a tablet PC, and PCS. (Personal Communication System), GSM (Global System for Mobile communication), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple) Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet) may be all kinds of wireless communication devices such as terminals.

In the description of the embodiment of the present application, the guardian terminal (not shown) may be a terminal owned by a subject corresponding to the guardian (parent, child, family, social worker, etc.) of the user of the user terminal 31. According to an embodiment of the present application, a guardian terminal (not shown) is a user terminal (not shown) when the user terminal 31 first subscribes to an application-based service provided by the dementia prediction apparatus 10 of the present application or while using an application-based service. It may be set based on the parental registration input applied by 31) and registered in the dementia prediction apparatus 10. In particular, in the description of the embodiment of the present application, a guardian terminal (not shown) may acquire various types of information and alarm information provided to the user terminal 31 together with the user terminal 31.

According to another embodiment of the present application, the device for predicting dementia 10 not only predicts the incidence of dementia of the user by applying the state data provided through the communication unit 20 to an artificial intelligence-based learning model, but also determines the user's behavior pattern. By analyzing the emergency situation notification information can be provided to the guardian terminal (not shown). In addition, the device for predicting dementia 10 may analyze a user's behavior pattern and, in case of an emergency, may provide emergency situation information to the communication unit 20. The communication unit 20 may provide emergency situation information to a guardian terminal (not shown) and an external server. That is, the dementia prediction apparatus 10 may analyze a behavior pattern, a life pattern, etc. of a dementia patient, and provide emergency notification information when a risk factor occurs.

According to the exemplary embodiment of the present disclosure, the wearable terminal 32 may be worn on a user's body part to obtain biometric signal data. The wearable terminal 32 is a device that is attached to a human body and can collect data related to heart rate, motion, and stress, and can acquire bio signals such as PPG, electrocardiogram, skin electrical conductivity, and body temperature. The biosignal data may include heart rate information, respiration volume information, blood oxygen concentration information, pulse information, blood pressure information, electrocardiogram information, and the like. The wearable terminal 32 may acquire current state information of the user. The biosignal data may be data obtained from the wearable terminal 32 worn by the user. For example, the wearable terminal 32 may be a device worn on a user's body part (eg, wrist).

In addition, the user terminal 31 and the wearable terminal 32 may include GPS. The user terminal 31 and the wearable terminal 32 may provide location information of the user. For example, the location information may be GPS information, but is not limited thereto. As another example, the location information may be latitude and longitude information. Alternatively, the location information may include address information, and location coordinates in a form different from the latitude and longitude information.

According to an exemplary embodiment of the present disclosure, the sensor unit 33 may collect user behavior data from sensors provided indoors and outdoors in an area where the user is located. The sensor unit 33 may be installed in a smart home to collect information on a user's indoor space location over time. As an example, the sensor unit 33 may be provided in a user's living space. The sensor unit 33 may include a heat sensor, a motion sensor, a position sensor, a motion sensor, a human body sensor, a temperature sensor, a weight sensor, a pressure sensor, and the like. The sensor unit 33 may be provided on a front door, a window, a door, etc. included in the smart home to detect the opening of the door. In addition, the sensor unit 33 may acquire the user's location movement line based on the heat sensor. In addition, the weight detection sensor is provided in a toilet, a toilet, a sofa, a bed, a chair, and the like, so as to obtain information about the time the user stayed in one place. In addition, the sensor unit 33 may include an image acquisition device (eg, CCTV). The sensor unit 33 may collect user behavior data including images, images, and the like from the image acquisition device.

According to the exemplary embodiment of the present application, the communication unit 20 may acquire state data collected from the plurality of monitoring and collection devices 30 through a network. Exemplarily, the communication unit 20 receives the status data collected from the plurality of monitoring and collection devices 30 through Bluetooth and ZigBee communication and temporarily stores the dementia prediction device 10 through Wi-Fi or LTE communication. Can be sent to. Here, the communication unit 20 may acquire status data collected from at least one of the plurality of monitoring collection devices 30 provided at locations capable of transmitting via Bluetooth and ZigBee communication through Bluetooth and ZigBee. In addition, the communication unit 20 may transmit the state data acquired from the plurality of monitoring and collection devices 30 to the dementia prediction device 10 through Wi-Fi or LTE communication. Here, the communication unit 20 may obtain the status data collected from the plurality of monitoring collection devices 30 through Wi-Fi or LTE communication even if the plurality of monitoring collection devices 30 are not located in a specific area.

According to an embodiment of the present application, an example of a network of the communication unit 20 is a 3rd Generation Partnership Project (3GPP) network, a Long Term Evolution (LTE) network, a 5G network, a World Interoperability for Microwave Access (WIMAX) network, and a wired/wireless Internet. (Internet), LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), Bluetooth (Bluetooth) network, Wifi network, NFC (Near Field Communication) network , A satellite broadcasting network, an analog broadcasting network, a Digital Multimedia Broadcasting (DMB) network, and the like, but are not limited thereto.

As an example, the communication unit 20 may include a first communication unit (not shown), a second communication unit (not shown), and a storage unit (not shown).

According to an exemplary embodiment of the present disclosure, the first communication unit (not shown) may communicate with the first monitoring collection device among the plurality of monitoring collection devices 30 through a network. For example, the first monitoring collection device may be the sensor unit 33. The first communication unit (not shown) is a communication device that exchanges data collected indoors with an analysis result, and may be an indoor dedicated gateway. The first communication unit (not shown) receives data collected from the plurality of sensor units 33 installed in the space where the user is located (eg, home, company, etc.) through Bluetooth and ZigBee communication, and temporarily stores the data. It may be transmitted to the dementia prediction device 10 through -Fi or LTE communication. The state data collected through the first communication unit (not shown) may be temporarily stored in a storage unit (not shown). In addition, the first communication unit (not shown) collects through Bluetooth and ZigBee communication when the user terminal 31 and the wearable terminal 32 carried by the user are located in an area equipped with the first communication unit (not shown). State data can be obtained. In addition, the first communication unit (not shown) may store data obtained from the first monitoring and collection device corresponding to a preset time and transmit the data to the dementia prediction device 10. As an example, the preset time may be a time set for efficient power use of the first monitoring collection device, but is not limited thereto.

According to the exemplary embodiment of the present application, the second communication unit (not shown) may communicate with the second monitoring collection device among the plurality of monitoring collection devices through a network. For example, the second monitoring collection device may be the user terminal 31. The second communication unit (not shown) may be a communication device that transmits and receives data and analysis results collected outdoors. The second communication unit (not shown) may transmit data obtained from the wearable terminal 32 and the user terminal 31 to the dementia prediction device 10 through Wi-Fi or LTE communication when the user is out. In addition, the second communication unit (not shown) may acquire data collected from the second monitoring collection device in response to a preset time. As an example, the preset time may be a time set for efficient power use of the second monitoring collection device, but is not limited thereto. In other words, the first communication unit (not shown) and the second communication unit (not shown) store the collected data for a certain period of time (pre-set time) without sending the collected data in real time for efficient power use of the plurality of monitoring and collection devices 30 After that, it may be intermittently transmitted to the dementia prediction apparatus 10.

According to an exemplary embodiment of the present disclosure, the storage unit (not shown) may store state data acquired from the plurality of monitoring and collection devices 30. For example, the communication unit 20 may store the state data in a storage unit (not shown) for a preset time and then transmit it to the dementia prediction apparatus 30. In addition, the storage unit (not shown) stores the state data obtained by the first communication unit (not shown) and the second communication unit (not shown) from the plurality of monitoring collection devices 30 when network communication is disabled, When the network communication is changed to a possible state, the obtained state data may be transmitted to the dementia prediction apparatus 30.

According to the exemplary embodiment of the present application, the communication unit 20 may transmit health-related information and an analysis result of the dementia prediction apparatus 10 to a display unit (not shown). The display unit (not shown) may be an interface based on a screen and a touch screen. In addition, the communication unit 20 may transmit health-related information and an analysis result of the dementia prediction apparatus 10 to the user terminal 31.

According to another embodiment of the present application, the display unit (not shown) is, for example, a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, or a micro electric field system (MEMS). It may include a display. The display unit (not shown) may output health-related information and an analysis result of the device 10 for predicting dementia. The display unit (not shown) may include a touch screen, and may receive, for example, a touch, gesture, near point, or hovering input using an electronic pen or a part of a user's body.

According to the exemplary embodiment of the present application, the dementia prediction apparatus 10 may predict a user's onset of dementia by applying state data provided through the communication unit 20 to an artificial intelligence-based learning model. The dementia prediction apparatus 10 may manage the collected state data and analyze the collected state data using machine learning or deep learning. In addition, the device for predicting dementia 10 may derive a feature variable associated with dementia based on state data. In addition, the dementia prediction apparatus 10 may predict a user's onset of dementia based on a relationship between the derived specific variable and the likelihood of onset of dementia.

A more detailed description of the dementia prediction apparatus 10 will be described below with reference to FIG. 2.

2 is a schematic block diagram of an apparatus for predicting dementia according to an embodiment of the present application.

2, the dementia prediction apparatus 10 may include a preprocessor 11, a feature variable extraction unit 12, a principal component analysis unit 13, a prediction unit 14, and a performance evaluation unit 15. have.

According to the exemplary embodiment of the present application, the preprocessor 11 may perform preprocessing by applying the obtained state data to a preprocessing algorithm. The state data may include life data collected from the user terminal 31, biosignal data obtained from the wearable terminal 32, and behavior data of the user obtained from the sensor unit 33. The preprocessor 11 may perform preprocessing by applying state data necessary for predicting dementia among the acquired state data to a preprocessing algorithm. The state data acquired from the plurality of monitoring and collecting devices 30 may be time series data in various types. The preprocessor 11 may pre-process the state data in the form of a time series in the form of N repetition cycles. In addition, the preprocessor 11 may perform preprocessing by applying a data preprocessing algorithm to be applied to an artificial intelligence-based learning model. Also, the preprocessor 11 may perform preprocessing of normalizing the acquired state data. The preprocessor 11 may perform data normalization to unify variable values of the acquired state data on a predetermined basis. In order to effectively reduce the error of the acquired state data, the preprocessor 11 uses a preprocessing algorithm (for example, MIN/MAX) to convert data including non-constant parts, that is, non-constant variable values. Can be normalized. The preprocessor 11 may convert the variable value of the collected state data into any one value within a range between the minimum value 0 and the maximum value 1 by normalization through the MIN/MAX method.

In addition, the preprocessor 11 may perform preprocessing for redundantly or multiplexed preprocessing of missing value processing and filtering. In addition, the preprocessor 11 may perform a preprocessing process based on the usage purpose of the state data. In addition, when the state data is numeric data, the preprocessor 11 may perform preprocessing of replacing the missing value with at least one of a maximum value, a mode value, a minimum value, a median value, an average value, 0, and deletion. Missing values can mean missing values, empty values, or missing values due to measurement errors

As another example, when the acquired state data is image data, the preprocessor 11 may perform data conversion in a specific format (eg, JPEG format). In addition, the preprocessor 11 may exclude the image when an image of low quality or low resolution, such as an out-of-focus, an artifact, or a sound range, is acquired during the image acquisition process. Exemplarily, the preprocessor 11 may perform a 5-step preprocessing process. First, the preprocessor 11 may perform a cropping step. In the cropping step, an unnecessary part (eg, background) of the edge may be cut out with the user as the center of the acquired image. In addition, the preprocessor 11 may perform preprocessing of cropping an image to a size applicable to an artificial intelligence-based learning model.

Next, the preprocessor 11 may perform a parallel shift step. The preprocessor 11 may move the obtained image in parallel in the vertical, left, and right directions. In addition, the preprocessor 11 may perform a flipping step. For example, the preprocessor 11 may perform preprocessing to reverse an image acquired vertically. In addition, the preprocessor 11 may perform preprocessing of flipping the acquired image in at least one of the vertical directions and then flipping it in at least one of the left and right directions.

In addition, the preprocessor 11 may perform color adjustment. For example, in the color adjustment step, the preprocessor 11 may perform color adjustment of the image based on the color extracted by using an average subtraction method to the average RGB value of the entire data set. In addition, the preprocessor 11 may further perform a resizing step. The resizing step may be a step of expanding and reducing the obtained image to a preset size. Also, the preprocessor 11 may perform image data amplification preprocessing to increase the number of data of the acquired image data.

As an example, in the case of using a deep learning algorithm including a convolutional neural network, it is advantageous to achieve good performance as the amount of data increases, but when the amount of collected image data is small, it is very difficult to perform learning using a convolutional neural network. It can be insufficient. Meanwhile, the preprocessor 11 may perform a data augmentation process by applying at least one of rotation, flipping, cropping, and noise mixing of the acquired image data.

According to an exemplary embodiment of the present disclosure, the feature variable extraction unit 12 may extract a feature variable from the preprocessed state data. For example, characteristic variables can be obtained from variables such as sleep time and cycle, toilet visit time and cycle, stay time and cycle by indoor location, outing time and cycle, meal time and cycle, biosignal data, cell phone use, etc. Statistical values that are available, that is, mean, standard deviation, median, etc. can be included. The feature variable extractor 12 may obtain a first feature value for converting preprocessed state data, that is, time series data into a feature variable. In addition, the feature variable extraction unit 12 may extract the feature variable by applying the pre-processed state data to the feature extraction algorithm. In this case, the feature value extracted from the feature variable extracting unit 12 may be the first feature value.

Since time series data has a dynamic property, a feature extraction algorithm different from that of a static image is required. In time series signals, a Fourier transform is mainly used to represent a frequency indicating how many times a wave occurs per unit time and an amplitude indicating the height of the wave. In time series, signals are usually divided into frames to extract features. Let k be the number of points in one frame. Since randomness can occur at the boundary point, it is common to allow some overlap. Feature vectors can be extracted from each divided frame. In the end, not one feature vector is obtained, but a row of feature vectors is obtained. That is, after Fourier transform is performed in each frame, d power spectra are selected and used as features. Extracting features from a time series signal also has a cepstrum feature that is superior to Fourier features.

For example, characteristic (characteristic) variables are average (AVERAGE), standard deviation (STANDARD DEVIATION), COEFFICIENT OF VARIATION, ROOT MEAN SQURE, CREST, SRA ( SQURE ROOT OF AMPLITUDE), MARGINAL FACTOR, PEAK TO PEAK VALUE, IMPACT FACTOR, KURTOSIS VALUE, SKEWNESS VALUE, but may include at least one of , But is not limited thereto.

According to the exemplary embodiment of the present application, the principal component analysis unit 13 may obtain a principal component variable necessary for predicting the invention of dementia from the extracted feature variable. As an example, the principal component analysis unit 13 may analyze the principal components of the distribution when several data generated by the aforementioned plurality of feature variables are gathered to form a single distribution. Here, the main component means a direction vector in which the variance of data is the largest in that direction. This principal component variable is newly created for a given number of data. In other words, the main component analysis unit 13 includes the extracted feature variables, for example, sleep time and cycle, toilet visit time and cycle, stay time and cycle by indoor place, outing time and cycle, meal time and cycle, and biosignal data. Principal component variables can be obtained by analyzing the main components of this distribution when data corresponding to the mean, standard deviation, and median are gathered to form a single distribution. have.

In addition, the principal component analysis unit 13 applies principal component analysis to more efficiently use the variable acquired through the preprocessing algorithm, converts it into a principal component variable with high correlation, reduces the dimension, and obtains an optimized secondary feature value. It can be converted into an input variable of a predictive model. The principal component analysis unit 13 may obtain a principal component variable required for prediction of the invention of dementia by analyzing one feature value extracted by the feature variable extraction unit 12 as a principal component.

For example, principal component analysis (PCA) is a technique that rotates a data set so that variables are not statistically correlated. After rotation, depending on how important it is to describe the data, it often selects only some of the new features. This new characteristic is called a principal component (each principal component is affected by all variables of the existing data), and if only a part is selected, the dimension is reduced, so it is sometimes used for dimension reduction.

According to the exemplary embodiment of the present application, the prediction unit 14 may predict the onset of dementia by applying the principal component analysis obtained by the principal component analysis unit 13 to the dementia prediction model. The prediction unit 14 may predict the incidence of dementia by converting it into a principal component variable in the principal component analysis unit 13 to reduce the dimension and input the optimized principal component variable into the dementia prediction model. As an example, the prediction unit 14 may predict the incidence of dementia by applying the principal component variable obtained from the principal component analysis unit 13 to a dementia prediction model built based on a classification algorithm.

As another example, the dementia prediction model may be built by a machine learning unit (not shown). A machine learning unit (not shown) can construct a model for predicting dementia through learning that receives a plurality of state data sets as inputs through a preprocessing process and outputs items related to predicting the incidence of dementia. The dementia prediction model may be an example, and the dementia prediction model may be a learning model built based on machine learning (machine learning) and deep learning. In addition, the dementia prediction module may be a learning model (module) built based on at least one of a regression model, a classification model, a cluster model, and a deep learning model.

As another example, the dementia prediction model model may be generated based on a regression algorithm, a classification algorithm, a cluster algorithm, and a deep learning algorithm. Unsupervised learning refers to an algorithm that learns while analyzing or clustering the data itself, rather than building data for learning. The machine learning unit (not shown) may cluster and calculate the analysis patterns based on the clustering algorithm, and detect a new analysis pattern based on the degree of separation between the clusters of each analysis pattern. For example, as a clustering algorithm for unsupervised learning, a logistic regression algorithm, a random forest algorithm, a support vector machine (SVM) algorithm, a decision making algorithm, and a clustering algorithm may be used.

In addition, the machine learning unit (not shown) uses clustering algorithms such as Extra Tree algorithm, XG Boost algorithm and Deep Learning algorithm, K-means clustering algorithm, SOM (Self-Organizing-Maps) algorithm, EM & Canopy algorithm, in addition to the above-described algorithm. Through this, unsupervised learning can be performed. The Random Forest algorithm is an algorithm that averages each prediction result into one outcome variable by composing a forest of numerous Decision Trees, and the SVM algorithm determines the classification to which the data belongs by separating the boundary of the largest width in the distribution space of the data. It is a non-probabilistic algorithm. The Extra Tree algorithm is similar to the Random Forest, but the speed is faster than the Random Forest, and the XGBoost algorithm is a boost method that applies the result of the Tree of XGBoost to the next tree if the Tree of the Random Forest is independent. Deep Learning Algorithm is an algorithm that learns by controlling the effect of the pattern of variables on the result based on a multi-layered Neural Network with weights. In addition, the K-means clustering algorithm is a traditional classification technique that repetitively subdivides the target group into K clusters based on the average value (similarity) of the distance, and the SOM algorithm uses the input pattern of the training set as a weight. It is a technique of learning and clustering. In addition, the EM & Canopy algorithm refers to a method of clustering by updating parameter values through an iterative process starting from the one with the maximum possibility with a given initial value.

According to an exemplary embodiment of the present disclosure, the performance evaluation unit 15 may evaluate the performance of the dementia prediction model based on the medical diagnosis result and the prediction result of the dementia prediction model. As an example, the medical diagnosis result may be a Ground Truth value. The ground truth value is actually measured data, is data obtained through various types of measurements and observations performed in the field, and may mean verification data that can confirm work, but is not limited thereto. Also, the medical diagnosis result may mean diagnosis data of patients actually treated by a doctor. The medical diagnosis result may include information on a user's genetic factors, environmental factors, biological factors, and the like.

In addition, the performance evaluation unit 15 may adjust the machine learning parameter to obtain an optimized predictive model for dementia based on the evaluation result. As an example, the performance evaluation unit 15 may increase the performance of the dementia prediction model by applying hyper parameters for each algorithm (by learning model). Hyper-parameters are values that influence the performance of the dementia prediction model, such as the number of neurons in each layer, batch size, learning rate when updating parameters, and weight reduction. The performance evaluation unit 15 may set a range of hyper parameter values in order to optimize the hyper parameter. In addition, the performance evaluation unit 15 may randomly extract a value of the hyper parameter within a set range. (Step 1) In addition, the performance evaluation unit 15 may perform model training using the previously sampled hyperparameter values, and then evaluate the accuracy with verification data. (Step 2) The verification data may be data for evaluating the appropriateness of the hyper parameter. The performance evaluation unit 15 may repeat steps 1 and 2 described above for a predetermined number of times to determine accuracy and reset the range of the hyper parameter. The performance evaluation unit 15 may reset the hyper parameter in a direction of narrowing the range. The performance evaluation unit 15 may select a value of the hyper parameter within a narrowed range. An embodiment in which the performance of the dementia prediction model is improved by applying the above-described hyper parameter is only an example, is not limited thereto, and more various embodiments may be applied.

3 is a diagram illustrating a model for analyzing predictive variables of dementia of a device for predicting dementia according to an exemplary embodiment of the present disclosure.

For example, referring to FIG. 3, the device for predicting dementia 10 may acquire state data in the form of time series data from the plurality of monitoring and collecting devices 30. The dementia prediction apparatus 10 may obtain a first feature value for converting time series data into a feature variable by applying a data preprocessing algorithm to analyze the collected state data. In addition, the dementia prediction apparatus 10 applies principal component analysis to convert the variable into a principal component variable with high correlation by applying principal component analysis in order to use the variable acquired through the preprocessing algorithm more efficiently, and obtains optimized secondary feature values by reducing the dimension. have. In addition, the dementia prediction apparatus 10 may convert the secondary feature value into an input variable of a classification algorithm. The dementia prediction apparatus 10 may predict and classify a user's incidence of dementia using a dementia prediction model. In addition, the dementia prediction apparatus 10 may evaluate the performance of a dementia prediction model (learning model) by comparing it with a ground truth value obtained through a medical diagnosis, and adjust machine learning parameters to obtain an optimized model through the evaluation result.

4 is a diagram illustrating a process of obtaining a feature value using a first sensor of a system for predicting onset of dementia according to an embodiment of the present application.

For example, referring to FIG. 4, the dementia incidence prediction system 1 may obtain a feature value using a fixed motion sensor. For example, a plurality of fixed motion sensors may be provided around a toilet, in a toilet, outside a toilet, and so on. The dementia invention prediction system 1 can analyze data in the initial stage, in the initial state before the user goes to the bathroom, around the bathroom (0) and the bathroom (0). In addition, the dementia invention prediction system 1 may change the state to a'around the toilet' state when the motion sensor around the toilet recognizes the user. The dementia invention prediction system 1 can return to the initial stage when the toilet sensor and the toilet sensor are turned off at the same time in the current state. In addition, the dementia invention prediction system 1 may change the state of the toilet inside the toilet when the motion sensor inside the toilet recognizes the user. In addition, the dementia invention prediction system 1 may change the state outside the toilet if a sensor around the toilet recognizes the user after a certain period of time in the toilet and the sensor inside the toilet recognizes that there is no user. For reference, the dementia invention prediction system 1 may predict the user's location by analyzing state data acquired from a plurality of sensors provided around the toilet.

Hereinafter, based on the details described above, the operation flow of the present application will be briefly described.

5 is an operation flowchart of a method for predicting the onset of dementia according to an exemplary embodiment of the present application, and FIG. 6 is a flowchart illustrating an exemplary method for predicting an onset of dementia according to an exemplary embodiment of the present application.

The method for predicting the onset of dementia illustrated in FIGS. 5 and 6 may be performed by the system 1 for predicting the onset of dementia described above. Accordingly, even if omitted below, the description of the dementia onset prediction system 1 may be equally applied to the description of the dementia onset prediction method.

For example, referring to FIG. 5, in step S510, the plurality of monitoring and collection devices 30 may collect the user's state data including the user's life data, biosignal data, and behavior data.

In step S520, the dementia incidence prediction system 1 may acquire state data collected from the plurality of monitoring and collection devices 30 through a network.

In step S530, the dementia onset prediction system 1 may predict the onset of dementia by applying the collected state data to an artificial intelligence-based learning model.

For example, referring to FIG. 6, in step S531, the dementia incidence prediction system 1 may perform preprocessing by applying the obtained state data to a preprocessing algorithm.

In step S532, the dementia onset prediction system 1 may extract feature variables from the preprocessed state data.

In step S533, the dementia onset prediction system 1 may obtain a principal component variable required for prediction of the invention of dementia from the extracted feature variables.

In step S534, the dementia onset prediction system 1 may predict the onset of dementia by applying the principal component variable to the dementia prediction model.

In the above description, steps S510 to S530 and S531 to S534 may be further divided into additional steps or may be combined into fewer steps, according to an embodiment of the present disclosure. In addition, some steps may be omitted as necessary, and the order between steps may be changed.

The method for predicting the onset of dementia according to an exemplary embodiment of the present disclosure may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those produced by a compiler but also high-level language codes that can be executed by a computer using an interpreter or the like. The above-described hardware device may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

In addition, the above-described method for predicting the onset of dementia may also be implemented in the form of a computer program or application executed by a computer stored in a recording medium.

The foregoing description of the present application is for illustrative purposes only, and those of ordinary skill in the art to which the present application pertains will be able to understand that it is possible to easily transform it into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present application is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present application.

1: Dementia onset prediction system
10: dementia prediction device
20: communication department
30: multiple monitoring collection devices
31: user terminal
32: wearable terminal
33: sensor unit

Claims (9)

In the dementia incidence prediction system,
A plurality of monitoring collection devices for collecting user's state data including user's life data, biometric signal data, and behavior data;
A communication unit that obtains the status data collected from the plurality of monitoring collection devices through a network; And
A dementia prediction device for predicting the onset of dementia of a user by applying the state data to an artificial intelligence-based learning model,
Including,
The device for predicting dementia,
Deriving a characteristic variable related to dementia based on the state data, predicting the onset of dementia of the user based on the relation between the derived characteristic variable and the likelihood of dementia,
A preprocessor for performing preprocessing by applying the obtained state data to a preprocessing algorithm;
A feature variable extraction unit for extracting a feature variable from the preprocessed state data;
A principal component analysis unit that obtains a principal component variable necessary for predicting the onset of dementia from the extracted characteristic variables;
A prediction unit that predicts the onset of dementia by applying the principal component variable obtained by the principal component analysis unit to a dementia prediction model; And
A performance evaluation unit that evaluates the performance of the dementia prediction model based on the medical diagnosis result and the prediction result of the dementia prediction model,
Including,
The performance evaluation unit,
The dementia onset prediction system to reset the range of the hyper parameter of the dementia prediction model based on the performance evaluation result of the dementia prediction model. delete delete The method of claim 1,
The communication unit,
A first communication unit communicating with a first monitoring collection device among the plurality of monitoring collection devices through a network; And
A second communication unit that communicates with a second monitoring collection device among the plurality of monitoring collection devices through a network,
Including,
The first communication unit stores data obtained from the first monitoring and collection device according to a preset time and then transmits the data to the dementia prediction device,
The second communication unit is to obtain the data collected from the second monitoring collection device in response to a preset time, dementia outbreak prediction system. The method of claim 1,
The plurality of monitoring collection devices,
A user terminal collecting life data of the user;
A wearable terminal worn on a part of the user's body to obtain the biosignal data; And
A sensor unit that collects user behavior data from sensors provided indoors and outdoors in the area where the user is located,
That includes, dementia incidence prediction system. The method of claim 5,
The communication unit,
Including a storage unit for storing the state data obtained from the plurality of monitoring collection device,
The communication unit stores the state data for a predetermined time in the storage unit and then transmits the data to the dementia prediction device. In the dementia onset prediction method in which each step is performed by a computer-implemented dementia onset prediction system,
Collecting, by a plurality of monitoring collection devices, user's state data including user's life data, biometric signal data, and behavior data;
Obtaining the status data collected from the plurality of monitoring collection devices through a network; And
Predicting a user's onset of dementia by applying the state data to an artificial intelligence-based learning model,
Including,
Predicting the onset of dementia,
Deriving a characteristic variable related to dementia based on the state data, predicting the onset of dementia of the user based on the relation between the derived characteristic variable and the likelihood of dementia,
Performing preprocessing by applying the obtained state data to a preprocessing algorithm;
Extracting a feature variable from the preprocessed state data;
Obtaining a principal component variable necessary for predicting the onset of dementia from the extracted feature variables;
Predicting the onset of dementia by applying the principal component variable to a dementia prediction model; And
Evaluating the performance of the dementia prediction model based on the medical diagnosis result and the prediction result of the dementia prediction model;
Including,
Evaluating the performance of the dementia prediction model,
Dementia incidence prediction method to reset the range of the hyper parameter of the dementia prediction model based on the performance evaluation result of the dementia prediction model. delete A computer-readable recording medium storing a program for executing the method of claim 7 on a computer.

Images