Parkinson's disease detection method, system, equipment and medium

Abstract

The invention discloses a Parkinson's disease detection method, system and device and a medium, and relates to the technical field of intelligent medical treatment. Comprising the following steps: carrying out background denoising processing on an acquired audio signal to obtain a sample audio signal, and extracting audio characteristic parameters including audio total time, an instantaneous sound pressure value and glottis opening and closing time; based on these features, speed, sound power, and volume data of the patient, and a glottis cycle irregularity index are calculated. Furthermore, a fundamental frequency value is calculated by framing a sample audio signal, a total fundamental frequency is generated, and an audio disease index is obtained by combining a speech speed, a volume and a glottis period irregularity index. In addition, patient motion data including pace and stride is collected to correct the audio disease index, ultimately generating a comprehensive disease index. A multi-dimensional diagnosis mode enables detection of Parkinson's disease to form a more three-dimensional and comprehensive evaluation system, and more reliable reference is provided for clinicians.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and in particular to a Parkinson's disease detection method, system, equipment and medium. Background Art

[0002] As the global aging phenomenon intensifies, Parkinson's disease, as a common neurological disease, has attracted more and more attention. Parkinson's disease is mainly manifested by motor dysfunction. Patients may experience symptoms such as resting tremor, muscle stiffness, and bradykinesia during exercise. These symptoms not only affect the quality of life of patients, but may also cause patients to face serious safety hazards in their daily lives. Therefore, early and accurate diagnosis is crucial for the treatment, care and rehabilitation of Parkinson's patients. However, traditional Parkinson's disease diagnosis methods mostly rely on clinical observations by experts and self-reporting by patients, which have the disadvantages of strong subjectivity and poor accuracy, which brings difficulties to early detection and intervention.

[0003] In recent years, with the rapid development of digital signal processing and machine learning technologies, audio signal analysis has gradually been applied to the medical field. The patient's voice characteristics are believed to be closely related to their motor function, and certain characteristics of speech, such as speech rate, volume, and glottal period, can reflect the health of the patient's nervous system. Although some studies have attempted to use audio features to assist in the diagnosis of Parkinson's disease, these methods are often too complex, lack systematicity, and fail to fully consider the correlation between motion data and audio features. Therefore, there is an urgent need for a comprehensive detection method based on audio signal features and motion data to improve the diagnostic accuracy of Parkinson's disease.

[0004] In the prior art, the publication number CN118105032A discloses a coal mine supervision method based on video images, a combined feature parameter extraction system for Parkinson's disease detection and a Parkinson's disease prediction system. The combined feature parameter extraction system for Parkinson's disease detection includes: a speech signal feature parameter extraction module, an electroencephalogram signal feature parameter extraction module, a combined feature parameter acquisition module, and a dimension reduction module; the combined feature parameter acquisition module is used to combine speech signal feature parameters and electroencephalogram signal feature parameters; the dimension reduction module is used to reduce the dimension of the combined feature parameter vector using the linear minimum mean square error criterion to obtain the combined feature parameter vector after dimension reduction. However, in this scheme, although the feature extraction of the combined speech signal and the electroencephalogram signal can improve the comprehensiveness of the detection, it also introduces the complexity of data processing. The electroencephalogram signal usually contains a lot of noise and artifacts, requiring complex preprocessing and signal denoising steps. For clinical applications, this complexity may lead to an extension of processing time, thereby affecting the efficiency of real-time monitoring and diagnosis. In the combined feature parameter acquisition module, how to select and combine the most effective speech signal features and electroencephalogram signal features is a challenge. If irrelevant or redundant features are selected, it may lead to overfitting of the model, thereby reducing the accuracy of prediction. Therefore, the criteria and methods of feature selection need to be more clear and systematic to ensure that the extracted features can effectively reflect the characteristics of Parkinson's disease.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The object of the present invention is to provide a Parkinson's disease detection method, system, device and medium to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions: A method for detecting Parkinson's disease, comprising the following specific steps: Collecting an audio signal of a patient to be tested reading a text of a known number of words, preprocessing the collected audio signal to obtain a sample audio signal, extracting audio features based on the sample audio signal to obtain audio feature parameters, wherein the preprocessing is background denoising, and the audio feature parameters include total audio time, instantaneous sound pressure value, and glottal opening and closing time; The patient's speech speed data is calculated based on the audio feature parameters and the number of words in the text, the sound power of the sample audio signal is calculated based on the instantaneous sound pressure value in the audio feature parameters, the volume of the sample audio signal is characterized by the sound power to obtain the volume data, and the glottal period irregularity index is calculated based on the glottal opening and closing time; Divide the sample audio signal into frames, calculate the fundamental frequency value of each frame, generate the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame, and generate the audio disease index by combining the total fundamental frequency of the sample audio signal with the speech rate data, volume data and glottal period irregularity index; The motion data of the patient to be detected is collected, and the audio disease index is corrected based on the motion data to obtain a comprehensive disease index. The comprehensive disease index is compared with the disease index threshold, and different diagnosis results are generated according to different comparison results, wherein the motion data of the patient to be detected includes walking speed and stride.

[0008] Furthermore, audio feature extraction is performed based on the sample audio signal to obtain audio feature parameters, wherein the audio file is read by audio processing software, and the amplitude of the audio digital sample is taken as the sound pressure from the digital audio signal, and the instantaneous sound pressure value is the amplitude of the audio digital sample at the corresponding moment; the glottis opening and closing time refers to the glottis opening and closing time within a glottis cycle, which is determined by the waveform shape of the sample audio signal.

[0009] Furthermore, the speech speed data of the patient is calculated based on the audio feature parameters and the number of words in the text, wherein the speech speed data is characterized by the number of words read per minute. The specific formula for calculating the speech speed data of the patient is: ; In the formula, is the speech rate data, is the total number of words read aloud, is the total audio time in seconds; The sound power of the sample audio signal is calculated based on the instantaneous sound pressure value in the audio characteristic parameter, wherein the formula for calculating the sound power of the sample audio signal is: ; In the formula, is the sound power of the sample audio signal, is the root mean square sound pressure, is the density of the air medium, is the propagation speed of sound waves in air; The density of the air medium Corrected based on air humidity, density of air medium The calculation is based on the formula: ; In the formula, and are the gas constants for dry air and water vapor, respectively. is the temperature of the audio signal collection location, and are the partial pressures of dry air and water vapor, respectively, where and The specific calculation formula is: ; ; In the formula, is standard atmospheric pressure, is the saturated water vapor pressure at the current temperature, The humidity of the place where the audio signal is collected; RMS sound pressure The calculation is based on the formula: ; In the formula, is the total number of instantaneous sound pressure sampling points, For the The instantaneous sound pressure value of the sampling point, is the index of the sampling point, where ; The volume of the sample audio signal is characterized by the sound power to obtain volume data, wherein the formula for calculating the volume data is: ; In the formula, For volume, is the reference sound power.

[0010] Furthermore, the glottal period irregularity index is calculated based on the glottal opening and closing time, wherein the glottal period irregularity index is calculated based on the formula: ; In the formula, is the glottal period irregularity index, represents the standard deviation of the glottal period in the sample audio signal, represents the average time of the glottal cycle in the sample audio signal, where the kth glottal cycle is calculated based on the formula: ; In the formula, is the duration of the kth glottal cycle, and are the glottal opening and closing times in the kth glottal cycle in the sample audio signal, respectively, where k is the index of the glottal cycle, , is the total number of glottal cycles in the sample audio signal.

[0011] Furthermore, the sample audio signal is divided into frames, and the fundamental frequency value of each frame is calculated. The logic is as follows: Frames are divided in time, and the length of each frame is , the frame shift is , and get the frame signal, where the formula based on the frame signal is: ; In the formula, Indicates The audio signal of the frame, t is the time variable, , is the frame index, ,in, is the total number of frames divided from the entire audio signal; For each frame Calculate the autocorrelation function, where the specific formula is: ; In the formula, is the autocorrelation function, is the time delay, where ; Identify the peak delay of the autocorrelation function , the specific formula is: ; In the formula, Represents the autocorrelation function The independent variable at the maximum value The value of The fundamental frequency value of each frame is calculated based on the peak delay according to the formula: ; In the formula, is the sampling frequency, is the fundamental frequency value of the mth frame; The formula for generating the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame is: ; In the formula, is the overall fundamental frequency of the sample audio signal.

[0012] Furthermore, the formula for generating the audio disease index is based on the total fundamental frequency of the sample audio signal combined with the speech rate data, volume data and glottal period irregularity index: ; In the formula, is the audio morbidity index, is the glottal period irregularity index, For volume, is the speech rate data, 、 and are the glottal period irregularity index, the total fundamental frequency of the sample audio signal, and the weight coefficients of the sum of volume and speech rate, among which, and 、 and Both are greater than 0.

[0013] Furthermore, the motion data of the patient to be tested is collected, and the audio disease index is corrected based on the motion data to obtain a comprehensive disease index, wherein the formula for calculating the comprehensive disease index is: ; In the formula, is the comprehensive disease index, is the symmetry of the stride, is the average walking speed of the patients to be tested, is the maximum walking speed of the patient to be tested, and are the weight coefficients of the ratio of average speed to maximum speed and stride symmetry, respectively. and and All are greater than 0; The formula for calculating the symmetry of the stride is: ; In the formula, and The stride lengths for the left and right foot respectively; The comprehensive disease index is compared with the disease index threshold, and different diagnosis results are generated according to different comparison results. The logic for generating different diagnosis results is as follows: when When the patient has a positive score of 0.1, the patient is judged to be in the low-risk group, indicating that there are no significant signs of Parkinson's disease; when When the patient is diagnosed with mild Parkinson's disease, the patient is classified as a marginal risk group. when When the patient is diagnosed with Parkinson's disease, the patient is judged to be in the high-risk group. In the formula, is the disease index threshold.

[0014] The present invention further provides a Parkinson's disease detection system, which is used to execute the above-mentioned Parkinson's disease detection method, comprising: An audio data acquisition module is used to collect audio signals of a patient to be tested reading a text of a known number of words, preprocess the collected audio signals to obtain sample audio signals, extract audio features based on the sample audio signals, and obtain audio feature parameters, wherein the preprocessing is background denoising, and the audio feature parameters include total audio time, instantaneous sound pressure value, and glottal opening and closing time; A feature data calculation module is used to calculate the patient's speech speed data based on the audio feature parameters and the number of text words, calculate the sound power of the sample audio signal based on the instantaneous sound pressure value in the audio feature parameters, characterize the volume of the sample audio signal by the sound power to obtain volume data, and calculate the glottal period irregularity index based on the glottal opening and closing time; An audio disease analysis module is used to divide the sample audio signal into frames, calculate the fundamental frequency value of each frame, generate the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame, and generate an audio disease index by combining the total fundamental frequency of the sample audio signal with the speech rate data, volume data and glottal period irregularity index; The diagnosis result generation module is used to collect the motion data of the patient to be detected, correct the audio disease index based on the motion data to obtain a comprehensive disease index, compare the comprehensive disease index with the disease index threshold, and generate different diagnosis results according to different comparison results, wherein the motion data of the patient to be detected includes walking speed and stride.

[0015] The present invention also provides a Parkinson's disease detection device, comprising one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned Parkinson's disease detection method.

[0016] The present invention also provides a Parkinson's disease detection medium, on which a computer program is stored, and when the program is executed by a processor, the Parkinson's disease detection method as described above is implemented.

[0017] Compared with the prior art, the present invention has the following beneficial effects: This solution solves the problems of subjectivity and lack of accuracy in traditional diagnostic methods by systematically analyzing the audio signals of the patients being tested. Specifically, the solution can objectively and clearly evaluate the patient's voice characteristics through preprocessing of audio signals, feature extraction, and quantitative analysis based on these features. It provides accurate voice feature parameters, especially by calculating audio feature parameters such as instantaneous sound pressure value, glottal opening and closing time. In addition, the solution fully considers the influence of motion data in the process of generating audio disease index, which helps to improve the sensitivity and specificity of diagnosis. By comprehensively considering the patient's walking speed and stride, the patient's overall health status can be more accurately reflected. This multi-dimensional diagnostic method makes the detection of Parkinson's disease no longer limited to a single audio feature analysis, but forms a more three-dimensional and comprehensive evaluation system, which effectively promotes early screening and management, provides more reliable reference for clinicians, and thus improves the quality of life and treatment effect of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the overall method flow of the present invention; Figure 2 It is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION

[0019] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0020] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] Example: See also Figure 1 , the present invention provides a technical solution: A method for detecting Parkinson's disease, comprising the following specific steps: Step 1: Collect an audio signal of a patient to be tested reading a text of a known number of words, preprocess the collected audio signal to obtain a sample audio signal, extract audio features based on the sample audio signal, and obtain audio feature parameters, wherein the preprocessing is background denoising, and the audio feature parameters include the total audio time, instantaneous sound pressure value, and glottis opening and closing time.

[0022] Use high-quality recording equipment, such as professional microphones (e.g., dynamic microphones or condenser microphones), to ensure the clarity of the audio signal. Choose appropriate audio interfaces and audio capture devices, such as USB audio interfaces, to ensure the stability of signal transmission. Choose a quiet room for audio capture to avoid interference from background noise, such as external traffic noise, fan noise, etc. Use acoustic processing materials (such as sound-absorbing panels) to reduce echoes and noise in the room. Prepare the text to be read aloud, and make sure that the number of words in the text is known and moderate (e.g., 100-200 words) to facilitate subsequent analysis. The text should cover a variety of speech features, such as different tones, speaking speeds, and rhythms, in order to obtain rich audio feature information.

[0023] The audio signal may be affected by background noise during the acquisition process. Effective background denoising is crucial to improving the accuracy of subsequent audio feature extraction. The following are the specific steps for background denoising: Use software analysis tools (such as Audacity, Adobe Audition, or MATLAB) to perform preliminary analysis on the recorded audio signal and identify the spectral characteristics of the background noise. Find the silent paragraphs (parts without voice activity) to use as noise samples. Extract noise samples from the identified silent paragraphs and establish a noise feature model. These samples will be used for subsequent noise removal. Save the denoised audio signal as a new audio file, which is the sample audio signal.

[0024] Audio feature extraction is performed based on the sample audio signal to obtain audio feature parameters, wherein the audio file is read by audio processing software, and the amplitude of the audio digital sample is used as the sound pressure from the digital audio signal, and the instantaneous sound pressure value is the amplitude of the audio digital sample at the corresponding moment; the glottis opening and closing time refers to the glottis opening and closing time within a glottis cycle, which is determined by the waveform shape of the sample audio signal.

[0025] The total audio time refers to the duration of the recorded audio signal, which can be directly obtained through the audio processing library or software. The instantaneous sound pressure value describes the sound pressure state of the audio signal at a specific moment, usually expressed as the amplitude value of the signal. The instantaneous sound pressure value can be obtained by performing a short-time Fourier transform (STFT) on the audio signal or directly calculating the amplitude of the signal. Specifically, it includes calculating the instantaneous sound pressure value: directly extracting the instantaneous amplitude from the original audio signal; taking the absolute value: the instantaneous sound pressure value usually takes the absolute value of the audio signal amplitude. The glottal opening and closing time is a parameter closely related to speech characteristics, which is usually calculated by analyzing the audio envelope of the audio signal, using Hilbert transform or other methods to calculate the envelope of the audio signal, defining a threshold, detecting the opening and closing of the glottis from the envelope, determining the moment of glottal opening and closing, and calculating the time difference, and finally obtaining the glottal opening and closing time.

[0026] Step 2: Calculate the patient's speaking speed data based on the audio feature parameters and the number of words in the text, calculate the sound power of the sample audio signal based on the instantaneous sound pressure value in the audio feature parameters, characterize the volume of the sample audio signal by the sound power to obtain volume data, and calculate the glottal period irregularity index based on the glottal opening and closing time.

[0027] The patient's speech speed data is calculated based on the audio feature parameters and the number of words in the text, wherein the speech speed data is characterized by the number of words read per minute. The specific formula for calculating the patient's speech speed data is: ; In the formula, is the speech rate data, is the total number of words read aloud, is the total audio time in seconds; The sound power of the sample audio signal is calculated based on the instantaneous sound pressure value in the audio characteristic parameter, wherein the formula for calculating the sound power of the sample audio signal is: ; In the formula, is the sound power of the sample audio signal, is the root mean square sound pressure, is the density of the air medium, The speed of sound waves in the air medium is about 343m / s in the air; The density of the air medium Corrected based on air humidity, density of air medium The calculation is based on the formula: ; In the formula, and are the gas constant of dry air and the gas constant of water vapor, respectively. Generally, it is 287.05 J / (kg·K), the gas constant of water vapor Generally, it is 461.52 J / (kg·K), is the temperature of the audio signal collection location, and are the partial pressures of dry air and water vapor, respectively, where and The specific calculation formula is: ; ; In the formula, is standard atmospheric pressure, is the saturated water vapor pressure at the current temperature, is the humidity of the audio signal collection site; standard atmospheric pressure Generally, it is 1013.25hPa, which is the saturated water vapor pressure at the current temperature. It can be obtained by consulting the corresponding saturated water vapor pressure table at different temperatures. The general range is 6.11hPa to 1013.25hPa, which is determined according to different temperatures.

[0028] RMS sound pressure The calculation is based on the formula: ; In the formula, is the total number of instantaneous sound pressure sampling points, For the The instantaneous sound pressure value of the sampling point, is the index of the sampling point, where ; The volume of the sample audio signal is characterized by the sound power to obtain volume data, wherein the formula for calculating the volume data is: ; In the formula, For volume, is the reference sound power.

[0029] Among them, the reference sound power The correction is made according to the gender and age of the patients to be tested. The specific formula is: ; In the formula, is the initial reference sound power, generally 10−12 W / m 2 , is the gender index, where males take 1.2 and females take 1.0. is the age of the patient to be tested, is the age effect coefficient.

[0030] As you age, the quality, frequency range, and volume of your voice may change. Older people's vocal cords may become more rigid, which can affect the voice's ability to produce sound. The vocal cords of men and women are structurally different, with men typically having a lower fundamental frequency and greater vocal cord mass, so under the same vocalization conditions, men will generally produce more vocal power than women.

[0031] It is assumed that age decreases linearly or exponentially from a certain baseline value (e.g., 30 years old). To characterize the effect of age on the reference sound power The influence of age Generally between 0.1 and 0.2, used to indicate the effect of age on the reference sound power. degree of impact.

[0032] The glottal period irregularity index was calculated based on the glottal opening and closing time, where the formula for calculating the glottal period irregularity index is: ; In the formula, is the glottal period irregularity index, represents the standard deviation of the glottal period in the sample audio signal, represents the average time of the glottal cycle in the sample audio signal, where the kth glottal cycle is calculated based on the formula: ; In the formula, is the duration of the kth glottal cycle, and are the glottal opening and closing times in the kth glottal cycle in the sample audio signal, respectively, where k is the index of the glottal cycle, , is the total number of glottal cycles in the sample audio signal.

[0033] in, represents the standard deviation of the glottal period, Representing the average time of the glottal cycle, calculating the standard deviation and the average are mature conventional technical means and will not be elaborated here.

[0034] Step 3: Divide the sample audio signal into frames, calculate the fundamental frequency value of each frame, generate the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame, and generate the audio disease index by combining the total fundamental frequency of the sample audio signal with the speech rate data, volume data and glottal period irregularity index.

[0035] The sample audio signal is divided into frames and the fundamental frequency value of each frame is calculated. The logic is as follows: Frames are divided in time, and the length of each frame is , the frame shift is , and get the frame signal, where the formula based on the frame signal is: ; In the formula, Indicates The audio signal of the frame, t is the time variable, , is the frame index, ,in, is the total number of frames divided from the entire audio signal; The frame length of the audio signal Usually between 20ms and 40ms. For speech signals, the commonly used frame length is 25ms or 30ms; Usually set to the frame length 50% to 75% of the total.

[0036] For each frame Calculate the autocorrelation function, where the specific formula is: ; In the formula, is the autocorrelation function, is the time delay, where ; Identify the peak delay of the autocorrelation function , the specific formula is: ; In the formula, Represents the autocorrelation function The independent variable at the maximum value The value of The fundamental frequency value of each frame is calculated based on the peak delay according to the formula: ; In the formula, is the sampling frequency, is the fundamental frequency value of the mth frame; The sampling frequency The determination can be done by the formula: ; The formula for generating the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame is: ; In the formula, is the overall fundamental frequency of the sample audio signal.

[0037] The formula for generating the audio disease index is based on the total fundamental frequency of the sample audio signal combined with the speech rate data, volume data and glottal period irregularity index: ; In the formula, is the audio morbidity index, is the glottal period irregularity index, For volume, is the speech rate data, 、 and are the glottal period irregularity index, the total fundamental frequency of the sample audio signal, and the weight coefficients of the sum of volume and speech rate, among which, and 、 and Both are greater than 0.

[0038] The total fundamental frequency refers to the basic frequency of the sound signal, which is usually related to the vibration frequency of the vocal cords. Parkinson's disease affects the muscle control of sound generation, which may cause changes in the tension and vibration amplitude of the vocal cords, thereby affecting the fundamental frequency. Many Parkinson's patients have a lower fundamental frequency, which may be due to the weakening of muscle tension, resulting in the inability of the vocal cords to vibrate fully; therefore, changes in the fundamental frequency can directly affect the disease index. The logarithmic transformation of the fundamental frequency can stabilize its numerical range and reduce the impact of extreme values. The audio disease index The larger the value, the higher the probability of suffering from Parkinson's disease. Audio Disease Index Inversely proportional.

[0039] Speech rate refers to the number of words spoken per minute. People with Parkinson's disease often exhibit slow speech (i.e. bradykinesia), which may result in a significantly lower speech rate. A lower speech rate will result in an increased disease index, reflecting the patient's possible motor impairment. Therefore, speech rate data Audio Disease Index Inversely proportional.

[0040] Volume refers to the intensity of sound. Parkinson's patients may experience a phenomenon of weakened voice, called "low-volume speech" or "whispering". Lower volume values ​​will lead to an increase in the disease index, indicating that the patient may have a decline in vocal function, so volume Audio Disease Index Inversely proportional. At the same time, the volume is expressed by taking the square root With speech rate data Audio Disease Index The reason for using square root to process volume and speech rate is that it can smooth the impact of these two variables and avoid large changes in the amplitude of the impact. This approach can reduce the impact of unbalanced factors.

[0041] The glottal period irregularity index is a measure of the periodicity of sound waves, indicating the regularity of vocal cord vibration. The voices of Parkinson's disease patients tend to show greater irregularity and instability, which can lead to an increase in this index. Using an exponential function can increase the sensitivity to the effects of this irregularity, especially in When the value exceeds a certain threshold, it indicates that the patient may have severe voice irregularity, which is clinically considered to be an important indicator of Parkinson's disease. Audio Disease Index Directly proportional.

[0042] Among them, the glottal periodic irregularity index is an indicator used to measure the periodicity and stability of glottal vibration. It is usually related to the quality, stability and variability of the audio signal. Therefore, its influence on the audio disease index is The fundamental frequency has the greatest impact on the pitch of the sound, but it is smaller than the glottal period irregularity, because the stability and regularity of the glottal vibration directly affect the quality and intelligibility of the sound. Although volume and speaking rate are very important for the communication of speech, the characteristics they represent are relatively more subjective and do not directly reflect the structure and stability of the sound. Therefore, setting and 、 and Both are greater than 0.

[0043] Step 4: Collect motion data of the patient to be tested, correct the audio disease index based on the motion data to obtain a comprehensive disease index, compare the comprehensive disease index with the disease index threshold, and generate different diagnostic results according to different comparison results, wherein the motion data of the patient to be tested includes walking speed and stride.

[0044] Methods for obtaining motion data of the patient to be tested include: Many smart watches and fitness bracelets are equipped with accelerometers and gyroscopes, which can monitor motion data in real time. Users only need to wear the device to automatically record the pace and stride. Or professional motion monitors can measure pace and stride more accurately, especially for use in clinical or sports science research. In the process of collecting motion data of the patient to be tested, the heart rate and blood pressure data of the patient to be tested during exercise are also collected.

[0045] The motion data of the patient to be tested is collected, and the audio disease index is corrected based on the motion data to obtain a comprehensive disease index, where the formula for calculating the comprehensive disease index is: ; In the formula, is the comprehensive disease index, is the symmetry of the stride, is the average walking speed of the patients to be tested, is the maximum walking speed of the patient to be tested, and are the weight coefficients of the ratio of average speed to maximum speed and stride symmetry, respectively. and and All are greater than 0; The formula for calculating the symmetry of the stride is: ; In the formula, and The stride lengths for the left and right foot respectively; Symmetry of stride The larger the value, the more obvious the difference between the left and right foot strides, that is, the asymmetry of the stride. This usually reflects some potential movement problems. The smaller the value, the more consistent the stride length of the left and right feet, which means that the gait is more symmetrical, which is usually associated with normal motor function, that is, the symmetry of the stride length. The larger the value, the higher the probability of illness, and the comprehensive disease index The larger the value, the higher the probability of illness, so the symmetry of the stride and comprehensive morbidity index Directly proportional.

[0046] Average walking speed of the patient to be tested The highest walking speed of the patient to be tested pass Indicates the difference, that is, the closer its value is to 1, the smaller the difference is, and the larger the difference is, The smaller the value of , the higher the probability of disease. and comprehensive morbidity index Inversely proportional.

[0047] The comprehensive disease index is compared with the disease index threshold, and different diagnosis results are generated according to different comparison results. The logic for generating different diagnosis results is as follows: when When the patient has a positive score of 0.1, the patient is judged to be in the low-risk group, indicating that there are no significant signs of Parkinson's disease; when When the patient is diagnosed with mild Parkinson's disease, the patient is classified as a marginal risk group. when When the patient is diagnosed with Parkinson's disease, the patient is judged to be in the high-risk group. In the formula, is the disease index threshold.

[0048] Among them, the disease index threshold The calculation is based on the formula: ; In the formula, is the preset disease index threshold, is the blood pressure change during exercise, which is calculated by collecting the blood pressure data of the patient to be tested during exercise. is the maximum heart rate of the patient to be tested during exercise, is the resting heart rate of the patient to be tested.

[0049] See also Figure 2 The present invention also provides a Parkinson's disease detection system, which is used to execute the above-mentioned Parkinson's disease detection method, including: An audio data acquisition module is used to collect audio signals of a patient to be tested reading a text of a known number of words, preprocess the collected audio signals to obtain sample audio signals, extract audio features based on the sample audio signals, and obtain audio feature parameters, wherein the preprocessing is background denoising, and the audio feature parameters include total audio time, instantaneous sound pressure value, and glottal opening and closing time; A feature data calculation module is used to calculate the patient's speech speed data based on the audio feature parameters and the number of text words, calculate the sound power of the sample audio signal based on the instantaneous sound pressure value in the audio feature parameters, characterize the volume of the sample audio signal by the sound power to obtain volume data, and calculate the glottal period irregularity index based on the glottal opening and closing time; An audio disease analysis module is used to divide the sample audio signal into frames, calculate the fundamental frequency value of each frame, generate the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame, and generate an audio disease index by combining the total fundamental frequency of the sample audio signal with the speech rate data, volume data and glottal period irregularity index; The diagnosis result generation module is used to collect the motion data of the patient to be detected, correct the audio disease index based on the motion data to obtain a comprehensive disease index, compare the comprehensive disease index with the disease index threshold, and generate different diagnosis results according to different comparison results, wherein the motion data of the patient to be detected includes walking speed and stride.

[0050] The present invention also provides a Parkinson's disease detection device, comprising one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned Parkinson's disease detection method.

[0051] The present invention also provides a Parkinson's disease detection medium, on which a computer program is stored, and when the program is executed by a processor, the Parkinson's disease detection method as described above is implemented.

[0052] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0053] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0054] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0055] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for detecting Parkinson's disease, characterized in that: The specific steps include: Collecting an audio signal of a patient to be tested reading a text of a known number of words, preprocessing the collected audio signal to obtain a sample audio signal, extracting audio features based on the sample audio signal to obtain audio feature parameters, wherein the preprocessing is background denoising, and the audio feature parameters include total audio time, instantaneous sound pressure value, and glottal opening and closing time; The patient's speech speed data is calculated based on the audio feature parameters and the number of words in the text, the sound power of the sample audio signal is calculated based on the instantaneous sound pressure value in the audio feature parameters, the volume of the sample audio signal is characterized by the sound power to obtain the volume data, and the glottal period irregularity index is calculated based on the glottal opening and closing time; Divide the sample audio signal into frames, calculate the fundamental frequency value of each frame, generate the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame, and generate the audio disease index by combining the total fundamental frequency of the sample audio signal with the speech rate data, volume data and glottal period irregularity index; The motion data of the patient to be detected is collected, and the audio disease index is corrected based on the motion data to obtain a comprehensive disease index. The comprehensive disease index is compared with the disease index threshold, and different diagnosis results are generated according to different comparison results, wherein the motion data of the patient to be detected includes walking speed and stride.

2. A Parkinson's disease detection method according to claim 1, characterized in that: Audio feature extraction is performed based on the sample audio signal to obtain audio feature parameters, wherein the audio file is read by audio processing software, and the amplitude of the audio digital sample is used as the sound pressure from the digital audio signal, and the instantaneous sound pressure value is the amplitude of the audio digital sample at the corresponding moment; the glottis opening and closing time refers to the glottis opening and closing time within a glottis cycle, which is determined by the waveform shape of the sample audio signal.

3. A Parkinson's disease detection method according to claim 1, characterized in that: The patient's speech speed data is calculated based on the audio feature parameters and the number of words in the text, wherein the speech speed data is characterized by the number of words read per minute. The specific formula for calculating the patient's speech speed data is: ； In the formula, is the speech rate data, is the total number of words read aloud, is the total audio time in seconds; The sound power of the sample audio signal is calculated based on the instantaneous sound pressure value in the audio characteristic parameter, wherein the formula for calculating the sound power of the sample audio signal is: ； In the formula, is the sound power of the sample audio signal, is the root mean square sound pressure, is the density of the air medium, is the propagation speed of sound waves in air; The density of the air medium Corrected based on air humidity, density of air medium The calculation is based on the formula: ； In the formula, and are the gas constants for dry air and water vapor, respectively. is the temperature of the audio signal collection location, and are the partial pressures of dry air and water vapor, respectively, where and The specific calculation formula is: ； ； In the formula, is standard atmospheric pressure, is the saturated water vapor pressure at the current temperature, The humidity of the place where the audio signal is collected; RMS sound pressure The calculation is based on the formula: ； In the formula, is the total number of instantaneous sound pressure sampling points, For the The instantaneous sound pressure value of the sampling point, is the index of the sampling point, where ; The volume of the sample audio signal is characterized by the sound power to obtain volume data, wherein the formula for calculating the volume data is: ； In the formula, For volume, is the reference sound power.

4. A Parkinson's disease detection method according to claim 2, characterized in that: The glottal period irregularity index was calculated based on the glottal opening and closing time, where the formula for calculating the glottal period irregularity index is: ； In the formula, is the glottal period irregularity index, represents the standard deviation of the glottal period in the sample audio signal, represents the average time of the glottal cycle in the sample audio signal, where the kth glottal cycle is calculated based on the formula: ； In the formula, is the duration of the kth glottal cycle, and are the glottal opening and closing times in the kth glottal cycle in the sample audio signal, respectively, where k is the index of the glottal cycle, , is the total number of glottal cycles in the sample audio signal.

5. A Parkinson's disease detection method according to claim 3, characterized in that: The sample audio signal is divided into frames and the fundamental frequency value of each frame is calculated. The logic is as follows: Frames are divided in time, and the length of each frame is , the frame shift is , and get the frame signal, where the formula based on the frame signal is: ； In the formula, Indicates The audio signal of the frame, t is the time variable, , is the frame index, ,in, is the total number of frames divided from the entire audio signal; For each frame Calculate the autocorrelation function, where the specific formula is: ； In the formula, is the autocorrelation function, is the time delay, where ; Identify the peak delay of the autocorrelation function , the specific formula is: ； In the formula, Represents the autocorrelation function When the independent variable reaches its maximum value The value of The fundamental frequency value of each frame is calculated based on the peak delay according to the formula: ； In the formula, is the sampling frequency, is the fundamental frequency value of the mth frame; The formula for generating the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame is: ； In the formula, is the overall fundamental frequency of the sample audio signal.

6. A Parkinson's disease detection method according to claim 5, characterized in that: The formula for generating the audio disease index is based on the total fundamental frequency of the sample audio signal combined with the speech rate data, volume data and glottal period irregularity index: ； In the formula, is the audio morbidity index, is the glottal period irregularity index, For volume, is the speech rate data, 、 and are the glottal period irregularity index, the total fundamental frequency of the sample audio signal, and the weight coefficients of the sum of volume and speech rate, among which, and 、 and Both are greater than 0.

7. A Parkinson's disease detection method according to claim 6, characterized in that: The motion data of the patient to be tested is collected, and the audio disease index is corrected based on the motion data to obtain a comprehensive disease index, where the formula for calculating the comprehensive disease index is: ； In the formula, is the comprehensive disease index, is the symmetry of the stride, is the average walking speed of the patients to be tested, is the maximum walking speed of the patient to be tested, and are the weight coefficients of the ratio of average speed to maximum speed and stride symmetry, respectively. and and All are greater than 0; The formula for calculating the symmetry of the stride is: ； In the formula, and The stride lengths for the left and right foot respectively; The comprehensive disease index is compared with the disease index threshold, and different diagnosis results are generated according to different comparison results. The logic for generating different diagnosis results is as follows: when When the patient has a positive score of 0.1, the patient is judged to be in the low-risk group, indicating that there are no significant signs of Parkinson's disease; when When the patient is diagnosed with mild Parkinson's disease, the patient is classified as a marginal risk group. when When the patient is diagnosed with Parkinson's disease, the patient is judged to be in the high-risk group. In the formula, is the disease index threshold.

8. A Parkinson's disease detection system, characterized in that: The Parkinson's disease detection system is used to execute the Parkinson's disease detection method according to any one of claims 1 to 7, comprising: An audio data acquisition module is used to collect audio signals of a patient to be tested reading a text of a known number of words, preprocess the collected audio signals to obtain sample audio signals, extract audio features based on the sample audio signals, and obtain audio feature parameters, wherein the preprocessing is background denoising, and the audio feature parameters include total audio time, instantaneous sound pressure value, and glottal opening and closing time; A feature data calculation module is used to calculate the patient's speech speed data based on the audio feature parameters and the number of text words, calculate the sound power of the sample audio signal based on the instantaneous sound pressure value in the audio feature parameters, characterize the volume of the sample audio signal by the sound power to obtain volume data, and calculate the glottal period irregularity index based on the glottal opening and closing time; An audio disease analysis module is used to divide the sample audio signal into frames, calculate the fundamental frequency value of each frame, generate the total fundamental frequency of the sample audio signal based on the fundamental frequency value of each frame, and generate an audio disease index by combining the total fundamental frequency of the sample audio signal with the speech rate data, volume data and glottal period irregularity index; The diagnosis result generation module is used to collect the motion data of the patient to be detected, correct the audio disease index based on the motion data to obtain a comprehensive disease index, compare the comprehensive disease index with the disease index threshold, and generate different diagnosis results according to different comparison results, wherein the motion data of the patient to be detected includes walking speed and stride.

9. A Parkinson's disease detection device, characterized in that: It comprises one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the Parkinson's disease detection method as described in any one of claims 1 to 7.

10. A Parkinson's disease detection medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the Parkinson's disease detection method as claimed in any one of claims 1 to 7 is implemented.

Images