KR20210017616A - Diagnosis device for Parkinson's disease - Google Patents

Abstract

The present invention relates to a diagnosis device of a Parkinson′s disease, capable of diagnosing and predicting an initial idiopathic Parkinson′s disease, which comprises: a recording unit for recording a sentence; a storage unit for storing a pre-constructed diagnosis prediction model; an analysis unit for analyzing the uttered sentence; a diagnosis unit for diagnosing the presence or absence of Parkinson′s disease in a patient; and a control unit for controlling the recording unit, the storage unit, the analysis unit, and the diagnosis unit, separately.

Description

Diagnosis device for Parkinson's disease

The present invention relates to an apparatus for diagnosing Parkinson's disease, and more particularly, it is possible to diagnose Parkinson's disease by analyzing a patient's speech sentence in terms of phonetic prosody, and the phonetic rhyme of speech sentences for patients with idiopathic Parkinson's disease in Korea. The present invention relates to a diagnostic device for Parkinson's disease that can diagnose and predict early idiopathic Parkinson's disease through analysis.

As we enter an aging society, the seriousness of social coping with the increase of Parkinson's disease patients is becoming an issue. In particular, there is no protection management system for Parkinson's disease patients for the recognition, protection, and management of Parkinson's disease symptoms, exclusively for Parkinson's patients at home or patient management institutions.

Therefore, there is a growing need for techniques to recognize signs of Parkinson's disease and recognize progression of Parkinson's disease symptoms only for Parkinson disease patients, which can be used at home or in Parkinson disease patient management institutions.

Korean Patent Publication No. 2010-0026302 discloses a management system for the elderly using a portable terminal, i.e., monitoring the information of the elderly including the location information of the elderly, transmitting the information of the elderly to the terminal of the expert, and sending instructions from the expert. It describes a technology that receives and transmits to the terminal of the elderly, and Korean Patent Publication No. 2010-0026302 still has to rely on experts for patient management, and there is a fatal problem in that the patient's symptoms are not automatically recognized. In view of the problems caused by the increase in Parkinson's disease patients due to the recent rapid entry of the aging society and the increase in social costs for Parkinson disease patients, the need for a technology to manage Parkinson's disease patients more efficiently is emerging.

More than 70% of patients with Parkinson's disease are known to have speech problems, which is the presence or absence of Parkinson's disease by the way patients who may have Parkinson's disease speak (speech way) or the sentence the patient says (speech sentence). It means that you can also grasp indirectly.

Specifically, this prediction is possible through the phonetic rhyme analysis of the patient's speech sentence. More importantly, the existing analysis tools are mainly based on foreign entrepreneurs' research on foreigners, so it is difficult to directly apply it to Korean patients. Since it is often inappropriate, there is an urgent need to develop a diagnostic prediction tool based on the results of research conducted on Parkinson's disease patients in Korea.

On the other hand, recently, based on artificial intelligence, a system for predicting or discriminating Parkinson's disease, mental illness, or abnormal signs through voice analysis, which anyone can diagnose easily and easily, has been developed, and is also applied to mobile applications.

Artificial intelligence (AI), currently used in industry, is based on supervised learning, which inputs various data into a computer and infers a single result through repeated training processes. Over the past few years, such supervised learning technology has evolved rapidly. As the ability to infer results is greatly improved, the fields of use are gradually becoming diversified, and operation costs are becoming cheaper, so voice identification products based on artificial intelligence have recently been developed and released.

The present invention relates to an apparatus for diagnosing Parkinson's disease, and more particularly, it is possible to diagnose Parkinson's disease by analyzing a patient's speech sentence in terms of phonetic prosody, and the phonetic rhyme of speech sentences for patients with idiopathic Parkinson's disease in Korea. The present invention relates to a diagnostic device for Parkinson's disease that can diagnose and predict early idiopathic Parkinson's disease through analysis.

In order to solve the above technical problem, the present invention includes a recording unit for recording a sentence uttered by a patient who wants to be diagnosed with Parkinson's disease, a storage unit for storing a pre-built diagnostic prediction model, and the recording unit. An analysis unit for analyzing the recorded spoken sentence, and a diagnosis unit for diagnosing the presence or absence of Parkinson's disease of the patient from the results analyzed by the analysis unit using the diagnosis prediction model stored in the storage unit, and the recording unit and storage unit In the diagnostic apparatus for Parkinson's disease comprising a control unit that controls each of the analysis unit and the diagnosis unit, wherein the diagnosis prediction model is composed of patients who pre-designate the sentence to be uttered by the patient, and have been diagnosed with Parkinson's disease. Record diagnostic sentences uttered by the control group consisting of patients and normal people, analyze the prosody of the recorded uttered sentences, build through multivariate statistical analysis and automatic machine learning algorithms using prosody analysis results, and detect Parkinson's disease A receiving unit for receiving a Parkinson's disease patient's physiological signal collected by sensors corresponding to the device, an ambient environment information signal, a position signal, a motion signal, an audio signal, and an image signal corresponding to the patient, and analyzing the signals, A judgment unit for determining characteristics of the Parkinson's disease patient, including whether the Parkinson's disease patient is unconscious while the signal is being generated, and whether or not to speak a language appropriate to the situation, and the characteristics corresponding to each of the Parkinson's disease symptoms and the Parkinson's disease patient It provides a diagnostic apparatus for Parkinson's disease, comprising: a symptom recognition unit that compares features of, determines changes in cognition, memory, and expression based on the comparison result, and recognizes Parkinson's symptoms based on the change. .

According to an embodiment of the present invention, the symptom recognition unit calculates a grade of Parkinson's disease symptom based on the recognition result of the Parkinson's disease symptom, analyzes the relationship between the environmental information signal and the Parkinson disease symptom, and Based on the relationship, the frequency of occurrence of the Parkinson's disease symptoms and the variance rate of the Parkinson's disease symptoms according to the surrounding environment are determined, and the variability rate of Parkinson's disease symptoms by time period, the variability rate of Parkinson disease symptoms by temperature, Parkinson by humidity. It may be to judge the rate of variability in disease symptoms.

According to another embodiment of the present invention, the determination is to determine whether a physiological phenomenon has occurred based on the physiological signal, and when the physiological phenomenon occurs, a management service request signal is not received or a predetermined time has elapsed. When the temperature physiological signal value and the humidity physiological signal value exist within a specific range, it is determined that a physiological phenomenon has occurred in the patient's unconscious state, and based on the position signal, it is determined whether the patient is out of position, and the position When a departure occurs, it may be to check whether a safe return or not based on the position signal.

According to another embodiment of the present invention, the symptom recognition unit recognizes a Parkinson's disease symptom based on an increase rate of unconscious position deviation, and when the increase rate is more than a specific value, the family history information of the Parkinson disease patient, and the daily information of the Parkinson disease patient , Recognizing the symptoms of Parkinson's disease based on the input information, the change in the cognitive ability, the change in the memory, and the change in the expressive power by receiving the examination information of the Parkinson's disease patient.

The present invention relates to an apparatus for diagnosing Parkinson's disease, and more particularly, it is possible to diagnose Parkinson's disease by analyzing a patient's speech sentence in terms of phonetic prosody, and the phonetic rhyme of speech sentences for patients with idiopathic Parkinson's disease in Korea. The present invention relates to a diagnostic device for Parkinson's disease that can diagnose and predict early idiopathic Parkinson's disease through analysis.

1 is a block diagram of a diagnostic apparatus for diagnosing Parkinson's disease according to an embodiment of the present invention,
2 is a block diagram showing a Parkinson's disease patient management platform and apparatus according to an embodiment of the present invention,
3 is a block diagram showing the vascular platform for Parkinson's disease patients of FIG. 2,
Figure 4 is a block diagram showing the Parkinson's disease recognition device shown in Figure 3,
Figure 5 is a block diagram showing a Parkinson's disease patient management and provision apparatus shown in Figure 3,
6 is a diagram schematically illustrating a process of diagnosing a diagnosis of Parkinson's disease using the diagnostic apparatus of FIG. 1,
7 is a flowchart illustrating a method for diagnosing Parkinson's disease according to an embodiment of the present invention,
FIG. 8 is a diagram conceptually showing the technical contents of diagnosis, evaluation, judgment, and measures for Parkinson's disease and the application of the contents embedded in a mobile phone.

Hereinafter, the present invention will be described in detail.

However, it should be noted that the technical terms used in the present invention are used only to describe specific embodiments, and are not intended to limit the present invention, and the technical terms used in the present invention have special different meanings in the present invention. Unless defined, the present invention should be interpreted as a meaning generally understood by those of ordinary skill in the art, and should not be interpreted as an excessively comprehensive or excessively reduced meaning, When a technical term used in the present invention is an incorrect technical term that does not accurately express the spirit of the present invention, it should be understood by being replaced with a technical term that can be correctly understood by a person skilled in the art, and the general term used in the present invention is It should be interpreted as defined or according to the context before and after, and should not be interpreted as an excessively reduced meaning, and the singular expression used in the present invention includes a plurality of expressions unless the context clearly indicates otherwise, In the present invention, terms such as “consisting of” or “comprising” should not be construed as necessarily including all of the various components or steps described in the invention, and some components or some steps are included. It may not be performed, or it should be construed that additional components or steps may be further included, and when it is determined that a detailed description of a related known technology may obscure the gist of the present invention in describing the present invention, the detailed description thereof Can be omitted.

In addition, reference numerals shown in the drawings are denoted by the same reference numerals unless there is a special reason, and the corresponding description may be omitted from the overlapping part.

1 is a block diagram of a diagnostic apparatus for diagnosing Parkinson's disease according to an embodiment of the present invention, and FIG. 2 is a block diagram showing a Parkinson's disease patient management platform and apparatus according to an embodiment of the present invention, and FIG. 2 is a block diagram showing the Parkinson's disease patient vascular platform, Figure 4 is a block diagram showing the Parkinson's disease recognition device shown in Figure 3, Figure 5 shows the Parkinson's disease patient management providing device shown in Figure 3 It is a block diagram, and FIG. 6 is a diagram schematically illustrating a process of diagnosing Parkinson's disease using the diagnostic apparatus of FIG. 1, and FIG. 7 is a diagram for explaining a method for diagnosing Parkinson's disease according to an embodiment of the present invention. It is a flow chart, refer to it.

Parkinson's disease diagnosis apparatus 100 according to the present invention includes a recording unit for recording a sentence uttered by a patient who is to be diagnosed with Parkinson's disease, a storage unit for storing a pre-built diagnostic prediction model, and the recording unit An analysis unit that analyzes the uttered sentences recorded in the storage unit, and a diagnosis unit and the recording unit to diagnose the presence or absence of Parkinson's disease of the patient from the results analyzed by the analysis unit using the diagnosis prediction model stored in the storage unit, and storage A diagnostic apparatus for Parkinson's disease comprising a control unit that controls each of a unit, an analysis unit, and a diagnosis unit, wherein the diagnosis prediction model designates the sentence to be uttered by the patient in advance, and includes patients diagnosed with Parkinson’s disease. The diagnostic sentences uttered by the composed patient group and the control group composed of normal people are recorded, the prosody of the recorded uttered sentences is analyzed, and the prosody analysis result is used to build through multivariate statistical analysis and automatic machine learning algorithms, and Parkinson's disease A receiver configured to receive a Parkinson's disease patient's physiological signal collected by sensors corresponding to a sensing device, an ambient environment information signal corresponding to the patient, a location signal, a motion signal, an audio signal, and an image signal, and analyze the signals, and the A judgment unit for determining characteristics of the Parkinson's disease patient, including whether the Parkinson's disease patient is unconscious while a physiological signal is being generated, and whether or not to speak a language appropriate to the situation, and the characteristics corresponding to each of Parkinson's disease symptoms and the Parkinson's disease The characteristic includes a symptom recognition unit that compares characteristics of patients, determines changes in cognition, memory, and expression based on the comparison result, and recognizes Parkinson's symptoms based on the changes.

In the present invention, throughout the specification, artificial learning and machine learning can be performed by artificial intelligence, and the "artificial intelligence" is also referred to as AI (Artificial Intelligence), which artificially implements some or all of human intellectual capabilities. It is a field of computer science that focuses on solving cognitive problems primarily linked to human intelligence, such as learning, problem solving, and pattern recognition.

In addition, the mental disease or abnormal symptom may be a symptom of depression, anxiety, early Parkinson's disease, or suicide, as well as Parkinson's disease.

In the present invention, the "depression" is a disease in which a'negative emotion' appears due to a change in the brain's ability to control emotions, and is a disease that more than 100 million people around the world suffer from.

The exact cause of depression has not been identified, and various hormones related to nerves such as dopamine and norepinephrine, including serotonin and melatonin, affect depression.

In addition, the onset of depression is influenced internally and externally. In the case of depression due to a social structural factor, complete healing can be achieved only through community cooperation, such as correcting the social atmosphere or expanding the social safety net.

In the present invention, the "anxiety" is a mental disorder that causes disorders in daily life due to various types of abnormal and pathological anxiety and fear.

The causes of anxiety are very diverse and vary from person to person, so it is difficult to define simply, and medically, the lack or excess of neurotransmitters in the cranial nerves responsible for emotional parts such as anxiety or depression, genetic predisposition, functional or structural of the brain Changes can be the cause, and social psychological aspects, past experiences, post-traumatic stress disorder or acute stress disorder can be the cause.

In the present invention, the "Parkinson's disease" is a disease in which cognitive function is impaired and personality changes occur, and a wide range of brains in which memory and thinking abilities gradually decline in the long term and have an effect on daily life. Means damage.

About 50 to 70% of Parkinson's disease corresponds to Alzheimer's disease, and about 20% have vascular Parkinson's disease, which occurs when blood is not properly supplied to the brain due to impaired blood supply or disease.

The system for determining early signs of mental illness or abnormality using artificial intelligence-based voice analysis of the present invention includes a recording unit, a storage unit, an analysis unit, a diagnosis unit, a control unit, and a discrimination unit.

The recording unit 200 may use a cardioid dynamic headset microphone (Shure WH20 XLR, USA) and Praat (ver 5132, Amsterdam, Netherlands), and may be digitized under a 22,050Hz standardization rate and 16-bit quantization conditions.

The storage unit 300 stores a diagnostic prediction model constructed to diagnose Parkinson's disease.

The recording unit 200 is characterized by recording the user's usual voice for a week to 2 months, and preferably recording for a month, but is not limited thereto.

The storage unit 300 is characterized in that the recording unit collects the user's usual voice recorded for one week to two months, preferably one month.

The storage unit 300 may be configured as a ROM to store the diagnostic prediction model.

As another example, the storage unit 300 is composed of a ROM and a RAM (RAM), the ROM stores the diagnostic prediction model, and the RAM contains the diagnosis result of the diagnosis unit 400 and the analysis unit 500 Analysis results can also be saved.

The analysis unit 400 is characterized by analyzing the sound level, speech speed, or frequency of a specific word of the collected voice.

In this case, the "pitch level" refers to the level of the pitch of the sound, and the "speech rate" refers to the speed at which you speak.

The analysis unit analyzes whether the sound level of the recorded voice is high or low, whether the speech speed is fast or slow, or whether a specific word is frequently used.

In addition, the analysis unit is characterized by analyzing the frequency of use of words related to Parkinson's disease, depression, anxiety, early Parkinson's disease, or suicidal signs.

The analysis unit 400 analyzes the sentence uttered from the patient recorded in the recording unit 200 using the diagnostic prediction model stored in the storage unit 300 and provides an analysis result for diagnosis of Parkinson's disease.

The analysis unit 400 may be provided with an analysis tool to automatically perform the diagnosis of Parkinson's disease.

The determination unit 700 is characterized by predicting or discriminating a mental illness or abnormal symptom from the voice data analyzed by the analysis unit through statistical analysis and artificial intelligence algorithm.

Specifically, the determination unit is characterized in that after accumulating the voice data analyzed by the analysis unit and data matching the actual user's mood, and then statistically analyzing the accumulated data to predict or discriminate a mental illness or abnormal symptom. .

On the other hand, there is a study report showing that the voice of depressed patients has a low pitch (Moses, 1954; Eldred&Price, 1958). Also, looking at previous studies on speech rate, depressed patients showed individual depressive mood. There is also a study report that the speech speed increased when the mood recovered from the patient (Klos, Ellgring&Scherer, 1988). The more severe the severity of depression, the slower the speech speed and the lower pitch range.

Therefore, on the basis of the preceding studies, the discriminating unit is when the voice data analyzed by the analysis unit has a low pitch, a slow utterance rate, or the frequency of use of words related to depression, anxiety, early Parkinson's disease, or suicidal signs. When it is high, after accumulating data matching the actual user's mood, the accumulated data is statistically analyzed to predict or discriminate mental illness or abnormal symptoms.

As another example, a storage unit may be constructed in the control unit 600 to store a diagnosis prediction model, or a diagnosis prediction model and a diagnosis result of the diagnosis unit 400 and an analysis result of the analysis unit 500 may be stored.

The diagnosis unit 500 diagnoses the presence or absence of Parkinson's disease using the analysis result of the analysis unit 400, and the diagnosis device 100 further includes an output unit (not shown), and the analysis result or diagnosis Results can be displayed, and such an output unit can include a display or a printer.

In addition, the Parkinson's disease diagnosis apparatus 100 of the present invention may sense a signal corresponding to the state of a Parkinson's disease patient using sensors and transmit it to the Parkinson's disease patient management platform 120, and the Parkinson's disease detection signal sensing unit, It may be composed of a sensing signal processing unit and a sensing signal transmission unit.

The Parkinson's disease detection signal sensing unit is at least one of a physiological signal, an audio signal, a video signal, an image signal, an environmental signal (including temperature, humidity, and time), a signal corresponding to the amount of activity and motion intensity, and a position signal of the human body of a Parkinson disease observer A signal including a can be sensed.

At this time, sensors that sense physiological signals, voice signals, movement signals, and position signals can be applied to a wearable device, and sensors that sense voice signals, environmental signals, movement signals, and position signals are personal portable smart devices. It can be applied to phones.

In addition, sensors that sense voice signals and environmental signals can be applied to PCs or servers to which the Parkinson's disease patient management platform 120 is applied, and sensors that sense image signals can be applied to CCTV, and Parkinson's symptom detection signals In order to configure the sensing unit, sensors for sensing physiological signals, motion signals, voice signals, and location signals may be included, and sensors for sensing image signals or environmental signals may be used.

At this time, the signal corresponding to the condition of the Parkinson's disease patient may be a signal having various information, and may not be limited to the information possessed by the signal, for example, a bowel signal, a temperature signal, a humidity signal, etc. It may correspond to a signal corresponding to the state of.

At this time, the Parkinson's disease patient management platform 120 recognizes signs and symptoms of Parkinson's disease based on the signal received from the device 100, and can perform preventive and situational measures using this. The management platform consists of a Parkinson's disease symptom recognition device and a Parkinson's disease patient management service providing device, and FIG. 3 is a block diagram showing the Parkinson's disease patient tube platform of FIG. 2, and the Parkinson's patient management platform 120 is It consists of a delay dong unit 210, Parkinson's disease recognition device 220, and Parkinson's disease patient management providing device 230.

The Parkinson's disease detection interworking unit 210 may establish a connection between the Parkinson's disease diagnosis apparatus and a Parkinson's disease patient service platform, process a communication protocol, and receive a multi-modal multi-channel message.

Here, since the signals transmitted from the Parkinson's disease diagnosis apparatus 110 are a plurality of signals, the Parkinson's disease detection interworking unit 210 may parse, store, and manage multiple detection signals.

On the other hand, looking at the Parkinson's disease recognition device 220 and the Parkinson's disease patient management providing device 230, the Parkinson's disease recognition device according to an embodiment of the present invention includes a receiving unit 310, a determination unit 320, and a symptom recognition unit ( 330), wherein the receiving unit 310 receives physiological signals, surrounding environment information signals, position signals, movement signals, audio signals, and image signals of Parkinson's patients collected by sensors corresponding to the Parkinson's disease diagnosis apparatus, The physiological signal may mean a signal generated when a Parkinson's disease patient presses urine or feces.

Here, the surrounding environment information signal may be a signal including any one or more of temperature, humidity, and time around the patient with Parkinson's disease, and the audio signal may mean a signal recorded by a patient with Parkinson's disease, and the image signal is a Parkinson's disease. In order to monitor the patient's symptoms and analyze the condition, it may refer to a signal taken of a Parkinson's disease patient.

Here, the motion signal may mean a signal including a radius of a motion of a Parkinson's disease patient, a pace of a step, and a direction of a step, and the receiver 310 synchronizes a signal for symptom analysis in the same time zone for a plurality of received signals. You can also do.

In addition, the determination unit 320 analyzes the signals to determine the characteristics of the Parkinson's disease patient, including whether the Parkinson's disease patient is in an unconscious state while the physiological signal is being generated, and whether or not speaking appropriate language. (320) analyzes the characteristics of signals for recognizing symptoms and signs of Parkinson's disease patients, and extracts and analyzes information that has an influence on the onset of Parkinson's disease.

In addition, whether the Parkinson's disease patient is unconscious may mean whether it is caused by the conscious behavior of the Parkinson's disease patient when a physiological phenomenon occurs.For example, Parkinson's disease patients contain urine or feces without their intention. Physiological phenomena may occur, and the determination unit 320 may determine this, and the determination unit 320 does not request a management service due to the occurrence of the physiological phenomena or the sensed temperature/humidity physiological signal values are If maintained, it may be determined that the patient is in an unconscious (unconscious) state, and when occurrence of a physiological phenomenon is detected by a physiological signal, the determination unit 320 may determine that the physiological phenomenon has occurred in an unconscious state.

In addition, the symptom recognition unit 330 compares the characteristics corresponding to each of the symptoms of Parkinson's disease with the characteristics of the Parkinson's disease patient, determines changes in cognition, memory, and expression based on the comparison result, and based on the change, Parkinson's disease Symptoms can be recognized, and the symptom recognition unit 330 can also analyze the influence of environmental information on the invention of Parkinson's disease symptoms, analyze the relationship between environmental information signals and Parkinson's disease symptoms, and analyze surroundings based on the relationship. It is possible to determine the incidence of Parkinson's disease symptoms according to the environment and the variability of Parkinson's disease symptoms. For example, when the temperature is high, the incidence of Parkinson's disease symptoms is high, or when the humidity is high, it may be determined that the incidence of Parkinson's disease symptoms is high.

In addition, the symptom recognition unit 330 may determine the variance rate of Parkinson's disease symptoms by time period, the variance rate of Parkinson’s disease symptoms by temperature, and the variance rate of Parkinson’s disease symptoms by humidity, and the symptom recognition unit 330 determines Using the results, a Parkinson's disease patient can create an environment in which symptoms occur to a minimum, and the symptom recognition unit 330 may calculate a grade of Parkinson's disease symptoms based on the recognition results of Parkinson's disease symptoms. The characteristics corresponding to each disease symptom are compared with the characteristics of Parkinson's disease patients, and if more than a certain number of the same symptoms occur, the Parkinson disease grade can be set as an upper grade, and if the number is less than a certain number, the middle or lower grade In addition, the symptom recognition unit 330 extracts features for each Parkinson's disease symptom, extracts fusion features necessary for Parkinson's disease symptom recognition, and calculates Parkinson's disease symptoms through a process of integrated analysis of the association. May be.

In addition, the symptom recognition unit 330 receives family history information of Parkinson's disease patients, daily life information of Parkinson's disease patients, and examination information of Parkinson's patients, and is based on the received information, changes in cognition, changes in memory, and changes in expression. You can also recognize the symptoms of Parkinson's disease.

Here, the family history information of Parkinson's disease patients can be stored and managed according to items necessary for Parkinson's influence analysis, such as direct line and disease name, and examination information of a specialized institution is a periodic MRI image reading result and ultrasound reading of a specialized institution. It is possible to store and manage inputs according to items necessary for analysis of signs and symptoms of Parkinson's disease, such as information and opinions such as results, blood test values, and expert judgment records.

In addition, the information for determining the degree of cognitive decline may mean information on the degree to which family, relatives, objects, and locations are recognized, and the information for determining the degree of memory decline is shown in a picture or video, and after a certain period of time. It can mean information about the degree of guessing the picture that was shown, or information about the degree of finding the word that was shown after a certain period of time, and it can be output through the speaker to help you guess the word that was heard to collect memory information. The information for determining the degree of decline in expressive power can be stored by inputting information such as situations such as wrong words and wrong actions and the contents of expressions that do not make sense, and the number of times to collect information on language proficiency that does not make sense. have.

Meanwhile, the Parkinson's disease diagnosis apparatus may further include a Parkinson's disease patient information management unit (not shown) that manages information on Parkinson's disease patients, and the Parkinson's disease patient information management unit may manage information generated through self-diagnosis, You can manage the degree of psychological state of Parkinson's disease patients, manage cumulative information about Parkinson's symptoms, manage activity radius information of Parkinson's disease patients, and provide information corresponding to Parkinson's observers and situational information. It can also be accumulated and managed.

In addition, the Parkinson's disease patient management service providing device 230 is composed of an analysis unit 410 and a service providing unit 420, the analysis unit 410 is the Parkinson's disease patients collected by sensors corresponding to the It receives physiological signals, surrounding environment information signals, location signals, motion signals, audio signals, and video signals, and analyzes the signals to analyze the current state of Parkinson's patients, and the service provider 420 When it is determined that the current state of the dementia patient is in a specific situation based on the result of comparing the state and the current state of the Parkinson's disease patient, a warning service corresponding to the specific situation and a notification service to acquaintances of the Parkinson disease patient are provided. .

Here, the warning service refers to a service that displays the situation of a Parkinson's disease patient on a device possessed by a Parkinson's disease patient.For example, if the patient has a physiological phenomenon unconsciously, a warning service notifying that a physiological phenomenon has occurred is provided. In addition, when a Parkinson's disease patient deviates from a designated location, a warning service notifying that a patient with Parkinson's disease has deviated from the designated location may be provided.

In addition, the service providing unit 420 determines whether to provide a notification service based on the level of the symptoms of Parkinson's disease, and when the level of the symptoms of Parkinson's disease corresponds to any one of the upper grade or the middle grade, the warning service and A notification service can be provided, and if the Parkinson's disease symptom falls under a relatively low grade, only a warning service can be provided.For example, if the Parkinson's disease grade corresponds to an intermediate grade or an upper grade, the patient may be affected by behavior. When an abnormality occurs, a physiological phenomenon occurs, or a location deviation occurs, a notification service may be provided to an acquaintance or guardian of the patient.

In addition, the notification service may include the time and place of occurrence, and such notification service may be delivered to a guardian or acquaintance using SMS, or may be delivered using an SNS service corresponding to a guardian or acquaintance. The providing unit 420 may provide a self-diagnosis that can be directly diagnosed by the Parkinson's disease patient himself or his family. For example, it may provide a memory diagnosis, a gait diagnosis, a cognitive diagnosis, a reaction rate diagnosis, and the like.

In addition, the service providing unit 420 may provide a service for alleviating or preventing Parkinson's disease symptoms, for example, memory rehabilitation, active force rehabilitation, cognitive rehabilitation, reaction force rehabilitation, etc., and the service providing unit 420 A service can be provided to the guardian, and a patient status notification service for notification of Parkinson's disease patients in an emergency situation, a service that monitors the status of Parkinson's patients, and the location of Parkinson's patients can be provided. An interlocking function may also be provided, and the service provider 420 may also provide a test schedule notification service, an exercise time notification service, a bedtime notification service, and a meal time notification service, and the service provider 420 warns of danger. An alarm service, a location tracking service, an automatic aid request service, and a movement/away detection service may also be provided.

Next, a method of providing a service of the Parkinson's disease diagnosis apparatus of the present invention will be described.

First, the Parkinson's disease diagnostic device collects signals using sensors, and the Parkinson's disease detection device can collect physiological signals, surrounding environment information signals, audio signals, video signals, and motion signals using sensors, and the physiological signals are Parkinson's. It may mean a signal that is generated when the sick patient urinates or stool, and the surrounding environment information signal may be a signal including any one or more of temperature, humidity, and time around the patient with Parkinson's disease, and the negative signal is Parkinson's disease. It may refer to a signal recorded by the patient's voice, and the image signal may refer to a signal photographed of a Parkinson's disease patient to monitor symptoms and analyze the condition, and the motion signal may refer to the movement of the Parkinson disease patient. It may mean a signal including a radius of, walking speed, and walking direction.

Subsequently, based on the collected signals, the characteristics of the Parkinson's disease patient are determined, and the determination unit 320 analyzes the signals to determine whether the Parkinson's disease patient is unconscious while the physiological signal is being generated, and whether or not speaking appropriate language. To determine the characteristics of the Parkinson's disease patient including, the determination unit 320 analyzes the characteristics of the signal for recognizing the symptoms and signs of the Parkinson's disease patient, and extracts and analyzes information that has an influence on the onset of Parkinson's disease. In addition, the unconsciousness of a Parkinson's disease patient may mean whether it is caused by conscious behavior of a Parkinson's disease patient when a physiological phenomenon occurs.For example, Parkinson's disease patients contain urine or feces without their intention. A physiological phenomenon may occur, and the determination unit 320 may determine this, and the determination unit 320 may determine that the patient is in an unconscious state when the patient's movement is less than a specific value based on the position signal. , When the movement of the patient is less than a certain value, when the occurrence of a physiological phenomenon is detected by a physiological signal, the determination unit 320 may determine that a physiological phenomenon has occurred in an unconscious state, and the determination unit 320 If there is no request for a management service due to the occurrence or the sensed temperature/humidity physiological signal value is maintained even after a certain period of time, the patient may determine that a physiological phenomenon has occurred in an unconscious (unconscious) state. When the occurrence is detected, the determination unit 320 may determine that a physiological phenomenon has occurred in an unconscious state.

Next, the Parkinson's disease symptoms are recognized by determining changes in cognition, memory, and expression, and whether the Parkinson's disease patient is unconscious while the physiological signal determined by the determination unit 320 is being generated, and whether or not the appropriate language is spoken. Changes in cognition, memory, and expression are determined by using the characteristics of the included Parkinson's disease patient, and the determination unit 320 may determine the degree of decline in expressive power by checking the frequency of words that do not fit the situation, A warning service and notification service corresponding to a specific situation can be provided, and the warning service refers to a service that displays the situation of a Parkinson disease patient on a device possessed by a Parkinson disease patient. For example, a physiological phenomenon in which the patient is unconscious If this occurs, a warning service notifying that a physiological phenomenon has occurred can be provided, and if a Parkinson's disease patient has deviated from the designated location, a warning service notifying that the designated location has been deviated can be provided. It determines whether to provide a notification service based on it. If the level of the Parkinson's disease symptom corresponds to either the upper or the middle level, a warning service and a notification service can be provided, and the level of the Parkinson's disease symptom is relatively low. If it falls under the lower level, only warning service can be provided.

For example, if the Parkinson's disease level corresponds to the middle or high grade, if the patient has an abnormal behavior, a physiological phenomenon, or a location deviation, a notification service may be provided to the patient's acquaintances or guardians. In addition, the notification service may include the time and place of occurrence, and the notification service may be delivered to a guardian or acquaintance using SMS, or may be delivered using an SNS service corresponding to a guardian or acquaintance, and Parkinson's disease. The patient himself or his family can provide self-diagnosis that can be directly diagnosed.For example, it can provide memory diagnosis, gait diagnosis, cognitive ability diagnosis, reaction rate diagnosis, etc., and alleviate or prevent Parkinson's disease symptoms. Services can be provided, memory rehabilitation, active rehabilitation, cognitive rehabilitation, reaction rehabilitation, etc. can be provided, and services can be provided to guardians, and a patient status notification service for notification when Parkinson's patients are in an emergency , Parkinson's disease patient's condition, Parkinson's disease patient's location monitoring service can be provided, and the function of linking to the guardian's device can also be provided, test schedule notification service, exercise time notification service, bedtime notification service, A meal time notification service may also be provided, and a danger warning alarm service, a location tracking service, an automatic aid request service, and a movement/outbound detection service may also be provided.

6 is a diagram schematically illustrating a process of diagnosing a diagnosis of Parkinson's disease using the diagnostic apparatus of FIG. 1, as model construction data 310, normal people required to build a diagnostic prediction model 340 ( Idiopathic Parkinson's disease using the confirmation information 320 of the hereinafter referred to as "control group") and patients (hereinafter referred to as "patient group") and the measured values 330 for each variable of the control group and patient group for building the model. A diagnostic prediction model 340 for diagnosing is constructed.

Using the constructed diagnostic prediction model 340, the presence or absence of Parkinson's disease of a patient who visited the outpatient hospital 350 is diagnosed to receive a diagnosis of Parkinson's disease.

Specifically, a measured value 360 for each variable for diagnosing Parkinson's disease of the patient is obtained, and the presence or absence of Parkinson's disease in the patient is diagnosed from the measured value 360 obtained from the patient using the diagnostic prediction model 340 ( 370)

7 is a diagram showing a flow chart for explaining a method for diagnosing Parkinson's disease according to an embodiment of the present invention. The method for diagnosing Parkinson's disease includes a diagnosis sentence designation operation 410, a speech diagnosis sentence recording operation 420, and diagnosis. Using the predictive model construction operation 430, the diagnosis sentence recording operation 440 of the patient who visited the hospital to receive the diagnosis of Parkinson's disease, the prosody analysis operation 450 of the diagnosis sentence uttered by the patient, and the analyzed prosody. The patient's Parkinson's disease diagnosis operation 460 may be performed in the order.

First, looking at the diagnosis sentence designation operation (S410), the diagnosis of idiopathic Parkinson's disease is not performed through medical procedures or tests, but is performed by analyzing the patient's utterance method. It is designated, and the diagnosis sentence is determined in consideration of the convenience of the patient and the rapidity of analysis, and it is desirable to minimize the number of sentences as much as possible. For example, one or two sentences or three or four sentences are preferable.

In addition, since Parkinson's disease patients have difficulty with pitch, intensity, and length variables, it is desirable to designate a sentence that can show the difference according to these variables at once, for example, a Korean sentence, and the change in the tone curve of the sentence. In order to induce, the diagnosis sentence may be composed of a plain preface, an interrogative sentence, and an exclamation sentence. For example, a plain preface such as "The morning is cold and the day is hot" may be designated as the diagnosis sentence. It is because it has a certain length back and forth based on the middle rest.

Meanwhile, in the present invention, the reason for diagnosing the presence or absence of Parkinson's disease by analyzing the prosody of the spoken sentence is as follows.

Patients diagnosed with idiopathic Parkinson's disease are known to have the following characteristics when speaking.

First, the range of the pitch and intensity of the spoken sentence is reduced.

In general, when people speak a sentence, the pitch changes continuously due to a number of complex factors, such as highlighting, contrasting, or expressing prominently in a specific word in the sentence.

Usually, intonation or intonation means the change in pitch over time, and in the case of a normal person, there is a distinct rise and fall in pitch (sound curve) over time.

On the other hand, patients with idiopathic Parkinson's disease tend to show a monotonous pattern like a mechanical sound because the range of changes in this sound level is considerably reduced compared to that of a normal person.

Also, the change in intensity tends to be reduced compared to that of a normal person.

Second, in the case of Parkinson's disease patients, the rate of utterance increases, which can be seen to be gradually increased due to the effect of akinesia.

On the other hand, the reduction of the pitch and intensity range may appear even in the case of normal people when suffering from the disease, but the reduction in the sound and intensity range seen by Parkinson's disease patients is beyond the degree that is simply seen in general patients, and is so much that it can be expressed as a mechanical sound. It is very severe and is very similar to uttering in a state of no emotion at all, so this is used for diagnosis in the present invention.

Therefore, after specifying the sentence for diagnosis as above, the uttered sentence is recorded, and a diagnostic predictive model is constructed and stored based on various predictive phonetic variables, and the prominent prosody characteristics of the uttered sentence are stored. The patient can be diagnosed.

Next, the recording operation (S420) will be described. First, a patient group and a control group for constructing a diagnostic prediction model are selected.

In the present invention, as a patient group, as a patient admitted to the hospital to receive a diagnosis of Parkinson's disease, idiopathic Parkinson's disease patients (IPD, Idiopathic Parkinson's disease), who are in a state before taking an anti-Parkinson disease drug, can be selected as a patient group, and a control group With, it is possible to select normal elderly people who do not have problems with cognitive ability and hearing that cause difficulties in daily life, and who do not have discomfort in physical activity.

When the selected patient group and control group read the specified sentence aloud, it is recorded through a recording unit. In this case, the recording time of the sentence for diagnosis may be approximately 5 minutes. This recording operation can be performed in a quiet place, for example in a separate recording room. Specifically, in the recording process, when the control group and the patient group for constructing the diagnostic prediction model read the diagnosis sentence aloud through a microphone, for example, a headset-type microphone in a recording room, the spoken sentence is recorded.

In addition, in some cases, it may be performed in a quiet place such as a hospital room or home.

Next, an operation (S430) of constructing a diagnostic prediction model will be described.

The predictor variables used in the prosody analysis of the present invention can be largely divided into variables that take time flow into account (dynamic variables) and variables that exclude time flow (static variables).These dynamic variables include prosody-related variables such as pitch, intensity, and length. There are variables and slope-related variables, and static variables include numerical fluctuations of each variable without time flow.

First, among the dynamic variables used in model construction, there are fundamental frequency (F0), intensity, and duration as prosody-related variables, and the fundamental frequency (F0) refers to the sound level (Hz) for each segment of the sentence. , Intensity means the intensity (dB) for each segment section of the sentence, and the length means the length (sec) for each segment section of the sentence.

Meanwhile, the slope-related variables of the dynamic variable include a maximum fundamental frequency slope (MaxFO_slope), an average intensity slope (MeanEnergy_slope), a semitone slope (SemiT_slope), a maximum intensity slope (MaxEnergy_slope), and an average sound slope (MeanF0_slope).

The highest fundamental frequency slope (Hz/sec) means a value obtained by dividing the maximum sound level of a segment by the time at that point.

The average intensity slope (Hz/sec) refers to a value obtained by obtaining the average sound level of a segment and dividing it by the intermediate time of the segment.

Semitone slope (Hz/sec) refers to the value divided by the total amount of time change of the sentence by changing the pitch value into semitones, obtaining the time value and the pitch of each frame, processing the difference value between the front and back frames, and ( semitone/sec), the maximum intensity slope (dB/sec) means a value obtained by dividing the maximum sound level of a segment by time.

The average sound slope (dB/sec) refers to a value obtained by obtaining the average intensity of a segment and dividing it by the median value of the segment.

In addition, static variables include the highest fundamental frequency (Highest_F0), lowest fundamental frequency (Lowest_F0), frequency range (Range_F0), fundamental frequency average (Mean_F0), highest strength (HIghest_dB), lowest strength (Lowest_dB), intensity average (Mean_dB), sound slope (SlopeHz-time), and intensity slope (SlopedB-time).

The highest fundamental inclination (Hz) means the maximum sound level in the entire sentence, the lowest fundamental frequency (Hz) means the lowest sound level in the entire sentence, and the fundamental frequency range (Hz) is the difference between the highest fundamental frequency and the lowest fundamental frequency. Means.

In addition, the highest intensity (dB) means the highest intensity in the entire sentence, the lowest intensity (dB) means the lowest intensity in the entire sentence, and the intensity average (dB) means the average intensity in the overall average.

The sound slope (Hz/sec) refers to a value obtained by dividing the amount of change in sound intensity for the entire sentence by time, and the intensity slope (dB/sec) refers to the value obtained by dividing the amount of change in intensity for the entire sentence by the amount of time change.

In the present invention, when analyzing a recorded sentence, it is based on prosody analysis.

The sound quality analysis can selectively analyze only a certain section by excluding the passage of time for a single vocal vocalization.

Specifically, prosody analysis is natural phenomena such as syllable boundary fluctuations according to phonological phenomena, assimilation phenomena due to the influence of preceding consonants and trailing consonants, and prosody analysis can be performed based on this knowledge.

Analysis conditions and variables must be set in consideration of prosody analysis within the automated analysis frame in the analysis unit 400 of FIG. 1.

First, a method of constructing a diagnostic prediction model stored in the storage unit 300 using three dynamic variables, sound level, intensity, and length will be described.

For the analysis sentences of the control group and the patient group, measurement values for three variables of soundness, intensity, and length are obtained.

This uses a widely known sound analysis program such as Praat or Wavesurfwe.

Then, for the same sentence, based on the obtained Euclidean distances for each of the three variables, one of the patient groups and both the control group are compared to obtain a patient group score set of the corresponding patient.

The formula for calculating the Euclidean distance between two points on the n-dimensional is as follows.

In Equation 1, in the n-dimensional, P and Q are P=(p1, p2, p3,pn), respectively, and Q=(q1,q2,qn).

Specifically, in order to check the dynamic variable, the uttered sentence must be segmented by word or syllable. For example, if the sentence is divided by word when "morning is cold, it is hot during the day" as a diagnostic sentence, the segment is divided into sound waves and It can be determined by referring to the spectrogram at the same time. In other words, in order to obtain the variable values for sound and intensity, the uttered sentences of the patient group and the sentences of the control group are marked for each segmental unit, and the length is made the same using the PSOLA algorithm. give.

Then, the difference between the negative variable of the sentence of the patient group and the sentence of the control group is calculated, and the difference between the intensity variable of the sentence of the patient group and the sentence of the control group is calculated.

At this time, the soundness and intensity are compared while the length is controlled. That the length is controlled means that the utterance lengths of the patient group and the control group are made the same for each segment.

This means that even in the same sentence, the length of the uttered sentence is different for each person, and different utterance lengths result in a difference in pitch and intensity because the distribution of breath is different.

Therefore, in order to compare the control group and the patient group, the difference in pitch and intensity between the two groups must be obtained by unifying the length of the sentence.

The length of the speech sentence of the control group is the standard, and the length of the patient group is increased or decreased by segment unit according to the control group.

The sentence length unification method can be manually or a dynamic time warping (DTW) technique can be used, and the n pitch points constituting the tone curve of the patient group and the control group are regarded as two points in the n-dimensional space. Then, the distance between these two points is calculated using the Euclidean distance formula and designated as a variable representing the difference in pitch between the two sentences.

On the other hand, the length variable is calculated by comparing the length for each segment unit before making the diagnosis sentence identical after the diagnosis sentence is displayed for each segment unit.

As an example, length comparison compares the length of each syllable targeting the original sentence before length duplication synthesis. Like the pitch and intensity, the length of n syllables in a sentence is considered as n coordinates and the other sentences are m Calculate the Euclidean distance by assuming an index

If the patient score set consisting of the three coordinates obtained above is expressed as coordinate points in a three-dimensional space consisting of three axes of pitch, intensity, and length, the degree to which the patient deviates from the prosody aspect of the control group with respect to the control group. Able to know. That is, the patient's score set can be expressed as far away from the origin (0,0,0) of the 3D space.

For each of the control groups, a score set is obtained for each member of the control group by one-to-one comparison with all other control groups in the same manner as above, and the score set is expressed as coordinate points in the same three-dimensional space.

Since the control group consists of normal people, the group is grouped near the origin of the 3D space, the control group is grouped at the root of the origin of the 3D space, and the patient group is distributed within the 3D space.

Therefore, the diagnostic prediction model is created based on how the values of predicted variables are distributed in the model space, and diagnosed using discriminant analysis, one of the multivariate statistical methods capable of classifying unknown values based on observed values. Although the construction of a predictive model is illustrated, it is also possible to construct a diagnostic prediction model using AI such as a machine learning method such as CART or a method such as neural network.

In addition, in the above, the use of three attributes of prosody, soundness, intensity, and length, to construct the diagnostic prediction model, but the number of variables used in the model construction is practically not limited.

Therefore, in addition to the three attributes of prosody, pitch, intensity, and length, slope-related variables that represent numerical fluctuations of variables for each segment (syllable or word) can be added, which can greatly increase diagnostic predictive power.

As described above, if the group group in the three-dimensional space of the control group and the patient group is set with a multivariate statistical method, for example, a discriminant function, and prediction is performed in the form of discriminant analysis for an unknown patient, this patient will be The probability value can be used to determine whether the probability of belonging to the normal group is high or whether the probability of belonging to the patient group is high.

Subsequently, an operation of recording a diagnosis sentence of a patient for diagnosis of Parkinson's disease is performed. Specifically, in order to receive a diagnosis of Parkinson's disease, a patient admitted to the hospital utters a designated diagnosis sentence in order to construct a diagnosis prediction model, and records it using a microphone or the like.

The recording operation can be performed in recording, or in a hospital room for hospitalized patients, or at home for discharged patients.

Next, the operation of analyzing the patient's diagnosis sentence can be performed.The clinician performs the prosody analysis of the diagnosis sentence uttered by the patient manually or semi-automatically, and analyzes it using the diagnosis prediction model stored in the storage unit. The results were exemplified by providing auxiliary information necessary for the diagnosis of idiopathic Parkinson's disease and analyzing the recorded diagnostic sentence by the clinician, but using the diagnostic prediction model stored in the storage unit using the analysis frame in the diagnosis unit 400 Of course, it is possible to automatically analyze the prosody of the diagnostic sentence uttered by.

Next, based on the analyzed result, an operation (S460) to diagnose the presence or absence of Parkinson's disease of the patient is performed, and the presence or absence of Parkinson's disease of the patient can be diagnosed manually or semi-automatically by, for example, a clinician or a doctor in charge. The device may have a separate diagnosis unit 500 and automatically diagnose the presence or absence of Parkinson's disease in the patient from the analysis result provided from the analysis unit and the diagnosis prediction model using a diagnosis tool. It is based on symptoms and can be provided, for example, in the following 5 items. If "0", the speech is "normal", and if "1", "the degree of expression is slightly reduced, and the number of words or voices Means the degree to which the size is reduced", "2" means "the degree to which speech is slow, single-tone or recognizable, and assessable as moderate damage", and "3" means "the degree to which it is difficult to understand as severe damage Means ", and "4" can mean "severe damage of an almost imperceptible degree".

Claims (4)

A recording unit for recording sentences uttered by a patient who wants to be diagnosed with Parkinson's disease, a storage unit for storing a pre-built diagnostic prediction model, an analysis unit for analyzing the uttered sentences recorded in the recording unit, And a diagnostic unit for diagnosing the presence or absence of Parkinson's disease of the patient from the result analyzed by the analysis unit using the diagnostic prediction model stored in the storage unit, and a control unit for controlling the recording unit, storage unit, analysis unit, and diagnosis unit, respectively. In the diagnostic device for Parkinson's disease,
The diagnostic prediction model pre-designates the sentence to be uttered by the patient, records a diagnostic sentence uttered by a patient group consisting of patients who have been confirmed for Parkinson's disease and a control group consisting of normal people, and recorded Analyzes the prosody of sentences, and builds through multivariate statistical analysis and automatic machine learning algorithms using the results of prosody analysis,
A receiver configured to receive a Parkinson's disease patient's physiological signal collected by sensors corresponding to the Parkinson's disease detection device, a surrounding environment information signal, a location signal, a motion signal, an audio signal, and an image signal, and analyze the signals. , A judgment unit for determining the characteristics of the Parkinson's disease patient, including whether the Parkinson's disease patient is unconscious while the physiological signal is being generated, and whether or not to speak a language appropriate for the situation, and the characteristics corresponding to each of Parkinson's disease symptoms and the Parkinson's disease diagnosis apparatus comprising a symptom recognition unit that compares characteristics of Parkinson's disease patients, determines changes in cognition, memory, and expression based on the comparison result, and recognizes Parkinson's symptoms based on the change. .
The method of claim 1,
The symptom recognition unit calculates a grade of Parkinson's disease symptom based on the recognition result of the Parkinson's disease symptom, analyzes the relationship between the environmental information signal and the Parkinson disease symptom, and based on the association relationship, the surrounding environment To determine the incidence of the Parkinson's disease symptoms and the variance rate of the Parkinson's disease symptoms according to the time period, the variability rate of Parkinson's disease symptoms by temperature, and the variance rate of Parkinson disease symptoms by humidity. Parkinson's disease diagnosis device, characterized in that.
The method of claim 2,
The determination is to determine whether a physiological phenomenon occurs based on the physiological signal, and when the physiological phenomenon occurs, a management service request signal is not received, or a temperature physiological signal value and a humidity physiological signal measured after a preset time has elapsed. If the value is within a specific range, the patient determines that a physiological phenomenon has occurred in an unconscious state,
A diagnostic apparatus for Parkinson's disease, characterized in that it determines whether the patient is out of position based on the position signal, and checks whether the patient is safely returned based on the position signal when the position out of position occurs.
The method of claim 2,
The symptom recognition unit recognizes a Parkinson's disease symptom based on an increase rate of unconscious position deviation, and when the increase rate is more than a specific value, the Parkinson's disease patient's family history information, the Parkinson's disease patient's daily information, and the Parkinson's disease patient's examination information An apparatus for diagnosing Parkinson's disease, characterized by recognizing the symptoms of Parkinson's disease based on input and received information, a change in cognitive ability, a change in memory, and a change in expression.

Images