KR20220007537A - Method and device for predicting Alzheimer's disease based on phonetic features - Google Patents

Abstract

Provided are a method and device for predicting Alzheimer's disease based on a vocal characteristic. According to an embodiment of the present invention, the device for predicting Alzheimer's disease comprises: a voice inputting unit configured to generate a voice sample by recording a voice of a target; a data inputting unit configured to receive demographic information of the target; a vocal characteristic extracting unit extracting the vocal characteristic from the generated voice sample; and a prediction model pre-trained to predict Alzheimer's disease of the target based on the vocal characteristic and the demographic information.

Description

Method and device for predicting Alzheimer's disease based on phonetic features

The present invention is a method for diagnosing Alzheimer's disease (AD), which accounts for the highest rate among the causative diseases of dementia, and provides assistance in determining the presence or absence of Alzheimer's disease in medical personnel through a non-invasive method using the patient's voice To a method, it is possible to provide a method and an apparatus capable of predicting a dementia state due to Alzheimer's disease through a non-invasive method for people with a high risk of dementia.

Although the proportion of using AI speakers is gradually increasing along with the 4th industry, it has not yet been popularized. Regarding the delay in popularization of AI speakers, various reasons exist, but the majority of consumers are of the opinion that there is not enough content to use AI speakers. In other words, the development of various contents using AI speakers is required.

Here, AI speakers and the healthcare industry are recognized as 'two rabbits' that capture both profitability and public interest, and many domestic and foreign researchers are developing various contents related to healthcare using AI speakers.

The elderly population over 65 in Korea accounted for 7.1% of the total population as of 2000, and is expected to exceed 20% by 2026. In the case of dementia, a representative geriatric disease, in 2000, 8.3% or 280,000 of the elderly aged 65 and over were counted, and in 2010, it is estimated to reach 430,000, or 8.6% of the total, and is on the rise. With the increase of the elderly population, the prevalence of dementia shows a steadily increasing trend due to the increase in life expectancy.

Alzheimer's disease (AD) is the most common causative disease accounting for about 70% of all dementias, and it is estimated that about 600,000 Alzheimer's disease patients will occur in 2020. Alzheimer's disease is a representative degenerative dementia disease that causes deterioration of cognitive function and daily life through degeneration of cerebral nerve cells. Since Alzheimer's disease is an irreversible disease and a curable treatment has not been developed, early diagnosis and early intervention for Alzheimer's disease is currently the best solution. As the prevalence of Alzheimer's disease increases, public health and medical expenditures are also having a significant impact.

Speech and speech disorders are one of the symptoms accompanying dementia due to cognitive function and neurodegeneration, and are frequently used criteria to differentiate normal aging and diseases that cause dementia, and often appear as markers of early symptoms. Therefore, in automatic speech recognition research, various studies are being carried out to discover speech and language markers for dementia diagnosis. Typically, studies that predict dementia using data from Pitt corpus of DementiaBank (e.g., Cookie Theft illustration) are mainly conducted.

Considering that cerebrospinal fluid examination or positron emission tomograph (PET), which is used to accurately diagnose Alzheimer's disease, a typical dementia-causing disease, is limited to a specific space and time by experts, conversation using AI speakers and smart devices There is a need for a method and apparatus for predicting whether a subject has Alzheimer's disease.

Korean Patent Publication No. 10-2014-0119486

The present invention relates to a method for diagnosing Alzheimer's disease using the acoustic characteristics of speech, and the presence and possibility of Alzheimer's disease of the conversation partner by collecting the conversation partner's responses using an AI speaker or a smart device through a voice recognition microphone It relates to a method and apparatus for predicting

According to an embodiment of the present invention, an apparatus for predicting Alzheimer's disease based on speech characteristics includes a voice input unit configured to generate a voice sample by recording a subject's voice; a data input configured to receive demographic information of the subject; a voice characteristic extracting unit for extracting voice characteristics from the generated voice sample; and a predictive model trained in advance to predict whether the subject has Alzheimer's disease or not based on the voice characteristics and the demographic information.

A method for predicting Alzheimer's disease based on speech characteristics according to an embodiment of the present invention includes: generating a voice sample by recording a subject's voice; receiving demographic information of the subject; extracting a voice characteristic from the generated voice sample; and predicting whether the subject has Alzheimer's disease by inputting the voice characteristics and the demographic information into a pre-trained prediction model.

A recording medium according to an embodiment of the present invention is a computer-readable recording medium storing computer-readable instructions, the instructions, when executed by at least one processor, cause the at least one processor to perform the steps, recording the subject's voice to generate a voice sample; receiving demographic information of the subject; extracting a voice characteristic from the generated voice sample; and predicting whether the subject has Alzheimer's disease by inputting the voice characteristics and the demographic information into a pre-trained prediction model.

The method and apparatus for predicting Alzheimer's disease based on voice characteristics according to an embodiment of the present invention is a method and apparatus for predicting whether a conversation target has Alzheimer's disease using an AI speaker and a smart device It is expected to help overcome the disease, and can support the diagnosis of Alzheimer's disease by medical personnel.

It can be supported to prepare an appropriate intervention plan for the continuously increasing dementia population by predicting the risk of Alzheimer's disease using the acoustic characteristics of speech provided by the present invention.

1 is a block diagram illustrating the configuration of an apparatus for predicting Alzheimer's disease based on speech characteristics according to an embodiment of the present invention.
2 exemplarily shows the operation of the apparatus for predicting Alzheimer's disease based on speech characteristics according to an embodiment of the present invention.
3 is a block diagram illustrating a procedure of a method for predicting Alzheimer's disease based on speech characteristics according to an embodiment of the present invention.
4 is a graph showing Alzheimer's disease prediction results in consideration of demographics and negative characteristics.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. Although the present invention has been described with reference to the embodiment shown in the drawings, which will be described as one embodiment, the technical idea of the present invention and its core configuration and operation are not limited thereby.

1 is a block diagram showing the configuration of an apparatus for predicting Alzheimer's disease according to an embodiment of the present invention. 2 exemplarily shows an operation process of an apparatus for predicting Alzheimer's disease according to an embodiment of the present invention.

1 and 2, the Alzheimer's disease prediction apparatus 10 according to an embodiment of the present invention includes a voice input unit 100, a data input unit 110, a voice characteristic extractor 120, a prediction model 130, and and a data storage unit 140 .

The Alzheimer's disease prediction apparatus 10 may be entirely hardware, or may have aspects that are partly hardware and partly software. For example, the Alzheimer's disease prediction apparatus 10 and each unit included therein of the present specification may collectively refer to an apparatus for exchanging data of a specific format and content in an electronic communication method and software related thereto. As used herein, terms such as “unit”, “module”, “server”, “device”, “device” or “terminal” refer to a combination of hardware and software driven by the hardware. it is intended to be For example, the hardware herein may be a data processing device including a CPU or other processor. In addition, software driven by hardware may refer to a running process, an object, an executable file, a thread of execution, a program, and the like.

In addition, each module constituting the apparatus 10 for predicting Alzheimer's disease is not necessarily intended to refer to physically distinct separate components. In FIG. 1 , the voice input unit 100 , the data input unit 110 , the voice characteristic extractor 120 , the predictive model 130 , and the data storage unit 140 are shown as separate blocks that are separated from each other, but this is Alzheimer's disease prediction The devices constituting the device 10 are only functionally divided by the operations performed by the device. Accordingly, in some embodiments, the voice input unit 100, the data input unit 110, the voice characteristic extraction unit 120, the predictive model 130, and the data storage unit 140 are all or partly integrated in the same device. can be For example, the Alzheimer's disease prediction apparatus 10 may be implemented through a device having voice recognition and data processing capabilities, such as an AI speaker. However, the present invention is not limited thereto, and one or more of the components may be implemented as separate devices physically separated from other units, or may be components communicatively connected to each other in a distributed computing environment.

The voice input unit 100 generates a voice sample of the subject. The voice input unit 100 may be configured to generate a voice sample by recording the subject's voice at a predetermined frequency. The voice input unit 100 may include a condenser microphone and a device for controlling the same, and may generate a voice sample by recording the subject's voice at a frequency of 16 Hz or higher. The subject performs at least one of picture description, standard paragraph utterance, and story recall tasks in a quiet room where recording is possible to perform spontaneous utterance or aloud utterance, and the voice input unit 100 records the subject's utterance to obtain a voice sample. will create

The data input unit 110 receives demographic information of a subject. Demographic information includes at least the age, gender, and education years (degree of education received) of the subject. Data input through the data input unit 110 means data from which the age, gender, and education years of the subject can be confirmed, or such information can be extracted. Illustratively, demographic information may be acquired through questionnaire data of the subject. The questionnaire data may include information related to the patient's age, gender, and education and training, and correspond to data generated using the medical judgment of a trained specialist, and may be data collected under the management of other medical staff and the patient's guardian. . The data input unit 110 may receive demographic information through the above-described questionnaire, but is not limited thereto. In some embodiments, the subject's demographic information may be information directly input by the subject.

Data input through the voice input unit 100 and the data input unit 110 may be stored in the data storage unit 140 . The data storage unit 140 may be configured to store input data or to provide a temporary or temporary storage space required for data processing of a predictive model to be described later.

The voice characteristic extractor 120 may extract the patient's voice characteristic from the inputted subject's voice sample. The voice characteristic extractor 120 may extract a voice characteristic related to a phonological characteristic, a root characteristic, and a spectral characteristic of a voice sample. Specifically, the speech characteristic extracting unit 120 provides information related to the fundamental frequency of the subject's voice, utterance (speech rate, utterance time, utterance length) information related to rest (degree of rest, number of rest periods, length of rest period) , Shimmer, Jitter, formant, harmonic-to-noise ratio, loudness, spectral centroid Mel frequency cepstrum coefficient (MFCC , Mel Frequency Cepstral Coefficients), identity vector (i-vector), articulation rate, zero-crossing rate (zcr, zero-crossing rate), voicing probability (vp, voicing probability), line spectral paris (LSP, line spectral paris), periodic variation At least one of (period perturbation), amplitude perturbation quotient (APQ), stiffness, energy, intensity (voice loudness, Intensity), and entropy may be extracted as a voice characteristic .

Here, the voice characteristic extractor 120 may first perform a pre-processing process for quantifying the voice sample. Through the pre-processing process, the time, frequency, etc. of the voice sample may be adjusted to be constant. In addition, it is possible to distinguish a human voice from a non-human voice through a preprocessing process. The speech characteristic extraction unit 120 may include an artificial neural network model (eg, a convolutional neural network) trained to select only human voices from the input voice samples, and may include the human voice model described above through the learned artificial neural network model. Preprocessing for selecting only negatives may be performed. The speech characteristic extracting unit 120 includes a fundamental frequency, a speech rate, a speech time, a speech length, a degree of pause, the number of pauses, a length of a pause period, a shimmer, and a jitter in the preprocessed voice sample. , formant, harmonic-to-noise ratio, loudness, spectral centroid, Mel Frequency Cepstral Coefficients (MFCC), identity vector (i) -vector), articulation rate, zero-crossing rate (zcr, zero-crossing rate), voicing probability (vp, voicing probability), line spectral paris (LSP, line spectral paris), period perturbation, amplitude fluctuation index (APQ) , amplitude perturbation quotient), stiffness, energy, intensity (volume of a voice, Intensity), and a voice characteristic including at least one of entropy may be extracted. The voice characteristic extraction unit 120 may extract voice characteristics corresponding to phonological, root, and spectral regions using a publicly available voice characteristic extraction program (eg, Praat).

The predictive model 130 may be in a state previously trained to predict whether a subject has Alzheimer's disease through the voice characteristics extracted by the voice characteristic extracting unit 120 and the demographic information provided through the data input unit 110 . The predictive model 130 may include at least one analysis model among a linear regression model, a logistic regression model, a machine learning model, and a neural network model.

The predictive model 130 includes a multivariate logistic regression model, and the multivariate logistic regression model may be configured as shown in Equation 1 below.

[Equation 1]

내지

는 독립 변수의 회귀 계수인 상수 값에 해당하며,

는 초기 상수 값에 해당하고,

는 치매 위험 확률 값에 해당한다)(Here, X ₁ to X _p are input values input to the predictive model as independent variables, respectively, corresponding to p voice characteristics and demographic information,

inside

corresponds to the constant value that is the regression coefficient of the independent variable,

is the initial constant value,

corresponds to the dementia risk probability value)

)이 0.5 이상인 경우 알츠하이머병으로 환자의 상태를 결정하며, 계산된 치매 위험 확률 값이 0.5 미만인 경우 정상인지기능으로 환자의 상태를 결정할 수 있다. The predictive model 130 constructed based on Equation 1 above may output a dementia risk probability value and output state information obtained by evaluating a patient's condition based on the dementia risk probability value. The dementia risk probability value means the probability of being diagnosed with dementia, and the status information indicates that the patient is diagnosed according to the diagnostic criteria of a specialist, and the patient's status may be determined as Alzheimer's disease or normal cognitive function. For example, the predictive model 130 calculates the dementia risk probability value (

) is 0.5 or more, the patient's status is determined as Alzheimer's disease, and when the calculated dementia risk probability value is less than 0.5, the patient's status can be determined by normal cognitive function.

However, the predictive model 130 of the present invention is not limited thereto. The predictive model 130 may be configured as a multinomial multivariate logistic regression model, and the predictive model 130 may determine the patient's condition as one of Alzheimer's disease, mild cognitive impairment, and normal cognitive function based on the calculated dementia risk probability values. have.

Hereinafter, a method for predicting Alzheimer's disease based on voice characteristics according to another embodiment of the present invention will be described.

3 is a flowchart of a method for predicting Alzheimer's disease based on voice characteristics according to an embodiment. The method may be performed in the prediction apparatus according to FIGS. 1 and 2 , and FIGS. 1 and 2 and related descriptions may be referred to for the description of the present embodiment.

Referring to FIG. 3 , the method for predicting Alzheimer's disease based on voice characteristics according to an embodiment includes generating a voice sample by recording a subject's voice (S100); Receiving demographic information of the subject (S110); extracting a voice characteristic from the generated voice sample (S120); and predicting whether the subject has Alzheimer's disease by inputting the voice characteristics and the demographic information into a pre-learned prediction model (S130).

Steps (S100) and (S110) in each step of the method for predicting Alzheimer's disease based on negative characteristics according to the embodiment are sequentially described and explained for convenience of explanation, and are limited to the order of description and are not sequentially performed. In some embodiments, step S110 may be performed before step S100. In addition, the step of learning the prediction model before the prediction method is performed may be performed first.

A voice sample is generated by recording the subject's voice (S100).

Subjects perform at least one of picture description, standard paragraph utterance, and story recall tasks in a quiet room with recording capability to utter spontaneous or aloud utterances, and voice samples are generated by recording the subject's utterances. The voice input unit 100 may include a condenser microphone and a device for controlling the same, and may generate a voice sample by recording the subject's voice at a frequency of 16 Hz or higher.

Receives demographic information of the subject (S110).

The demographic information includes the age, sex, and education years of the subject. Illustratively, demographic information may be acquired through questionnaire data of the subject. The questionnaire data may include information related to the patient's age, gender, and education and training, and correspond to data generated using the medical judgment of a trained specialist, and may be data collected under the management of other medical staff and the patient's guardian. . The data input unit 110 may receive demographic information through the above-described questionnaire, but is not limited thereto. In some embodiments, the subject's demographic information may be information directly input by the subject.

Next, a voice characteristic is extracted from the generated voice sample (S120).

A voice characteristic of the patient may be extracted from the inputted subject's voice sample. An artificial neural network model may be utilized in a preprocessing process for quantifying a patient's voice sample. Through the pre-processing process, the time, frequency, etc. of the voice sample may be adjusted to be constant. In addition, when there are a plurality of input voice samples, a pre-processing process for selecting the voice samples may be performed. The speech characteristic extracting unit 120 includes a fundamental frequency, a speech rate, a speech time, a speech length, a degree of pause, the number of pauses, a length of a pause period, a shimmer, and a jitter in the preprocessed voice sample. , formant, harmonic-to-noise ratio, loudness, spectral centroid, Mel Frequency Cepstral Coefficients (MFCC), identity vector (i) -vector), articulation rate, zero-crossing rate (zcr, zero-crossing rate), voicing probability (vp, voicing probability), line spectral paris (LSP, line spectral paris), period perturbation, amplitude fluctuation index (APQ) , amplitude perturbation quotient), stiffness, energy, intensity (volume of a voice, Intensity), and a voice characteristic including at least one of entropy may be extracted.

Next, the voice characteristics and the demographic information are input to the pre-trained prediction model to predict whether the subject has Alzheimer's disease (S130).

The predictive model 130 may be in a state previously trained to predict whether a subject has Alzheimer's disease through the voice characteristics extracted by the voice characteristic extracting unit 120 and the demographic information provided through the data input unit 110 . The predictive model 130 may include at least one analysis model among a linear regression model, a logistic regression model, a machine learning model, and a neural network model.

The predictive model 130 includes a multivariate logistic regression model, and the multivariate logistic regression model may be configured as shown in Equation 1 below.

[Equation 1]

내지

는 독립 변수의 회귀 계수인 상수 값에 해당하며,

는 초기 상수 값에 해당하고,

는 치매 위험 확률 값에 해당한다)(Here, X ₁ to X _p are input values input to the predictive model as independent variables, respectively, corresponding to p voice characteristics and demographic information,

inside

corresponds to the constant value that is the regression coefficient of the independent variable,

is the initial constant value,

corresponds to the dementia risk probability value)

The predictive model 130 constructed based on Equation 1 above may output a dementia risk probability value and output state information obtained by evaluating a patient's condition based on the dementia risk probability value. The dementia risk probability value means the probability of being diagnosed with dementia, and the status information indicates that the patient is diagnosed according to the diagnostic criteria of a specialist, and the patient's status may be determined as Alzheimer's disease or normal cognitive function. For example, the predictive model 130 determines the patient's status as Alzheimer's disease when the calculated dementia risk probability value is 0.5 or more, and determines the patient's status with normal cognitive function when the calculated dementia risk probability value is less than 0.5. have.

The voice characteristic-based Alzheimer's disease prediction method according to these embodiments may be implemented as an application or implemented in the form of program instructions that may be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination.

Examples of the computer-readable recording medium include hard disks, magnetic media such as floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floppy disks. media), and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like.

Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules for carrying out the processing according to the present invention, and vice versa.

So far, the present invention has been looked at with respect to preferred embodiments thereof. Those of ordinary skill in the art to which the present invention pertains will understand that the present invention can be implemented in a modified form without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments are to be considered in an illustrative rather than a restrictive sense. The scope of the present invention is indicated in the claims rather than the foregoing description, and all differences within the scope equivalent thereto should be construed as being included in the present invention.

Experimental example

An experiment was performed to construct a predictive model of the Alzheimer's disease prediction method and apparatus based on the voice characteristics according to the above-described embodiment, and to verify the constructed predictive model.

The subject data for the construction of the predictive model was obtained from a total of 210 patients, and voice and diagnostic information were obtained from patients visiting Boramae Hospital and those registered at Dongjak-gu Dementia Safety Center. Among the total subjects, 106 people in the Alzheimer's dementia group and 104 in the normal group were. To collect the subject's voice, the voice was collected while interacting with the examiner and the subject.

In order to quantify the subject's voice, preprocessing was performed on the voice sample using an artificial neural network model (convolutional neural network) first. That is, when there are a large number of input voice samples, a preprocessing process for selecting human voice samples is performed first to prevent noise data from being input for learning. For the pre-processed voice, the voice characteristics of the voice were extracted using an automated voice characteristic extraction method. Fundamental frequency (f0) mean, f0 std), speech-related information (speech rate, speech time, speech length), information related to pauses (pause rate, pause count, pause duration mean, pause duration standard deviation (std)), Shimmer, Jitter, formant, harmonic-to-noise ratio, loudness, spectral centroid mean, spectral centroid std) were extracted, respectively.

Demographic information (age, gender, education year) and each extracted voice characteristic are taken as input values, the Alzheimer's dementia prediction model outputs the subject's dementia risk probability value, and the patient based on the dementia risk probability value A predictive model that outputs state information that evaluates the state of The predictive model was implemented as a multivariate logistic regression model and constructed to output dementia risk probability values. The predictive model may determine the patient's status as Alzheimer's disease when the calculated dementia risk probability value is 0.5 or more, and determine the patient's status with normal cognitive function when the calculated dementia risk probability value is less than 0.5.

The prediction performance was tested through the built prediction model. Specifically, with respect to the prediction model, the AUC (area under the curve) of the ROC (receiver operating characteristic) curve of the index loader of the prediction performance was calculated. AUC is a value obtained by obtaining the area under the ROC curve. It is a representative index indicating the overall performance of the predictive model. The closer to 1, the better the performance. 4 is a graph showing the results of the constructed prediction model, and according to the optimal cutoff score according to the Youden Index, the performance is shown in Table 1 below.

cutoff AUC AUC.se Sensitivity Specificity PPV NPV 0.437 0.816 0.029 0.802 0.699 0.733 0.774

The predictive power was AUC = 0.816, the sensitivity was 0.802, and the specificity was 0.802. PPV (positive predictive value) was 0.733 and NPV (negative predictive value) was 0.774.

Conventionally, the test performed for the diagnosis of Alzheimer's disease was a method using a cerebral deposited amyloid PET and cerebrospinal fluid test. The use of amyloid PET is expensive, and it is difficult to use except in hospitals that operate specialized medical centers such as tertiary hospitals, and there are risk factors such as exposure to radiation. For cerebrospinal fluid analysis, there are limitations in that invasive lumbar puncture is used, labor is required for work and analysis, and there is a difference in reliability depending on the institution.

In contrast, the voice characteristic-based Alzheimer's disease prediction method and apparatus of the present invention diagnose Alzheimer's disease through a non-invasive test, and time and space that can be progressed through AI speakers/smartphones/tablets/PCs, etc. , a psychiatric screening test that transcends experts, allowing you to conveniently conduct the test at home or in a non-hospital setting. In addition, the risk of Alzheimer's disease can be determined at the primary and secondary clinics, and the false-positive can be minimized, which has the effect of saving costs and can be extended to treatment programs in the future.

That is, considering that the cerebrospinal fluid test or positron emission tomograph (PET) used to accurately diagnose Alzheimer's disease, a typical dementia-causing disease, is limited to a specific space and time by an expert, an AI speaker like the present invention and The test method for Alzheimer's disease in conversational subjects using smart devices is expected to help overcome the temporal, spatial, and resource limitations associated with existing tests.

As a result, Alzheimer's disease risk diagnosis using the acoustic characteristics of speech is expected to provide an appropriate intervention plan for the continuously increasing dementia population.

10: Alzheimer's disease prediction device
100: voice input unit
110: data input unit
120: voice feature extraction unit
130: predictive model
140: data storage unit

Claims (9)

a voice input unit configured to record the subject's voice to generate a voice sample;
a data input configured to receive demographic information of the subject;
a voice characteristic extracting unit for extracting voice characteristics from the generated voice sample; and
and a predictive model trained in advance to predict whether or not the subject has Alzheimer's disease based on the voice characteristics and the demographic information.
According to claim 1,
The demographic information is characterized in that it includes the age, sex, and education years of the subject, Alzheimer's disease prediction device.
According to claim 1,
The voice characteristic extracting unit includes a fundamental frequency of a voice, a utterance rate, a utterance time, a utterance length, a degree of pause, the number of pauses, a length of a pause period, a shimmer, a jitter, a formant, and a nasal sound. harmonic-to-noise ratio, loudness, spectral centroid Mel Frequency Cepstral Coefficients (MFCC), identity vector (i-vector), articulation velocity, zero Crossing rate (zcr, zero-crossing rate), negative probability (vp, voicing probability), line spectral paris (LSP), period perturbation, amplitude perturbation quotient (APQ), stiffness ( Stiffness), energy (Energy), intensity (voice loudness, Intensity), characterized in that for extracting at least one of entropy (Entropy) as the voice characteristic, Alzheimer's disease prediction apparatus.
4. The method of claim 3,
The speech characteristic extraction unit includes an artificial neural network model that performs preprocessing for selecting human speech from the speech sample,
The apparatus for predicting Alzheimer's disease, characterized in that the speech characteristic extractor extracts the speech characteristic from a preprocessed speech sample.
According to claim 1,
The predictive model is characterized in that it comprises at least one analysis model of a linear regression model, a logistic regression model, a machine learning model, and a neural network model, Alzheimer's disease prediction apparatus.
6. The method of claim 5,
The logistic regression model is a multivariate logistic regression model, and the multivariate logistic regression model is configured as in Equation 1 below, Alzheimer's disease prediction apparatus.

[Equation 1]

(Here, X ₁ to X _p are input values input to the predictive model as independent variables, respectively, corresponding to p voice characteristics and demographic information,

inside

corresponds to the constant value that is the regression coefficient of the independent variable,

is the initial constant value,

corresponds to the dementia risk probability value)
7. The method of claim 6,
The prediction model outputs a dementia risk probability value, and outputs state information that evaluates the patient's condition based on the dementia risk probability value,
The state information is Alzheimer's disease prediction apparatus, characterized in that determined as Alzheimer's disease or normal cognitive function.
recording the subject's voice to generate a voice sample;
receiving demographic information of the subject;
extracting a voice characteristic from the generated voice sample; and
and predicting whether the subject has Alzheimer's disease by inputting the voice characteristics and the demographic information into a pre-trained prediction model.
A computer-readable medium storing computer-readable instructions, the instructions, when executed by at least one processor, cause the at least one processor to perform steps, the steps comprising:
recording the subject's voice to generate a voice sample;
receiving demographic information of the subject;
extracting a voice characteristic from the generated voice sample; and
and predicting whether the subject has Alzheimer's disease by inputting the voice characteristics and the demographic information into a pre-trained prediction model.

Images