CN112472065A - Disease detection method based on cough sound recognition and related equipment thereof - Google Patents
Disease detection method based on cough sound recognition and related equipment thereof Download PDFInfo
- Publication number
- CN112472065A CN112472065A CN202011298720.2A CN202011298720A CN112472065A CN 112472065 A CN112472065 A CN 112472065A CN 202011298720 A CN202011298720 A CN 202011298720A CN 112472065 A CN112472065 A CN 112472065A
- Authority
- CN
- China
- Prior art keywords
- training
- sound
- characteristic
- mel frequency
- detected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
- A61B5/0823—Detecting or evaluating cough events
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Abstract
The invention provides a disease detection method based on cough sound recognition, which comprises the following steps: acquiring sound information to be detected or training sound information, and extracting a characteristic Mel frequency cepstrum coefficient of the sound information to be detected or extracting a training Mel frequency cepstrum coefficient of the training sound information; according to the characteristic Mel frequency cepstrum coefficient, drawing a characteristic Mel frequency spectrogram by taking time and frequency as axes, or respectively drawing a plurality of training Mel frequency spectrograms by taking time and frequency as axes according to a plurality of training Mel frequency cepstrum coefficients; and training the training Mel frequency spectrograms through a preset convolutional neural network to obtain a characteristic convolutional neural network model. Compared with the related technology, the invention has simple detection process and high detection efficiency on the related diseases.
Description
[ technical field ] A method for producing a semiconductor device
The invention relates to the field of target detection, in particular to a disease detection method based on cough sound recognition and related equipment thereof.
[ background of the invention ]
Because infectious diseases have great harm to human bodies and are widely related to people, the social public increasingly attaches importance to prevention, control and treatment of the infectious diseases, especially to the infectious diseases with extremely strong infectivity, if the rapid detection of the infectious diseases can be realized in as short a time as possible, individuals, hospital personnel and countries can be helped to make the most effective and the most rapid protection measures, so that how to realize the rapid detection of the infectious diseases becomes an important research direction in the field of medical treatment and health, and especially in the Covid-19 new crown epidemic outbreak in 2019, the rapid detection technology for the Covid-19 disease is very important.
The detection method aiming at the Covid-19 disease in the related technology mainly comprises X-ray shooting and nucleic acid detection aiming at the Covid-19 virus; after the X-ray shooting, hospital personnel make a preliminary diagnosis result after analyzing the chest radiograph of the detected object; in the nucleic acid detection of the Covid-19 virus, a pharyngeal swab or a nasal swab sample is required to be collected and analyzed, and a diagnosis result is made according to a check detection result, wherein the process generally takes 4-6 hours.
However, in the related art, no matter the detection is performed by taking X-ray photographs or the detection of nucleic acid against Covid-19 virus, the professional requirement of the operation is high, the operation process is complicated, and methods such as nucleic acid detection and X-ray photographs require expensive equipment, so that the used group is limited, the popularization is not facilitated, in addition, both the detection methods need to consume a lot of time, hospital personnel are limited, and the detection efficiency is directly low due to insufficient manpower.
Therefore, there is a need to provide a new method for detecting a cough sound based condition, a fingerprint detection device and a computer readable storage medium to solve the above technical problems.
[ summary of the invention ]
The invention aims to provide a cough sound identification-based disease detection method, a fingerprint detection device and a computer readable storage medium, which solve the problems of low fingerprint trace detection definition, low detection speed and insufficient real-time detection requirement.
In order to achieve the above object, the present invention provides a method for detecting a disease based on cough sound recognition, the method comprising the steps of:
step S1, acquiring sound information to be detected in real time, and extracting a characteristic Mel frequency cepstrum coefficient of the sound information to be detected; the sound information to be detected is sent out by the detected object;
step S2, drawing a characteristic Mel frequency spectrogram by taking time and frequency as axes according to the characteristic Mel frequency cepstrum coefficient;
step S3, identifying the characteristic Mel frequency spectrum diagram by using a characteristic convolution neural network model, and outputting a disease diagnosis result according to an identification result; the method for acquiring the characteristic convolution neural network model comprises the following steps:
step S1a, acquiring a plurality of training voice information, and respectively extracting training Mel frequency cepstrum coefficients of the training voice information; the training voice information is sent by a patient, and the training voice information and the training Mel frequency cepstrum coefficients are arranged in a one-to-one correspondence mode;
step S2a, respectively drawing a plurality of training Mel frequency spectrograms by taking time and frequency as axes according to a plurality of training Mel frequency cepstrum coefficients; the training Mel frequency cepstrum coefficients and the training Mel frequency spectrograms are arranged in a one-to-one correspondence mode;
and S3a, training the training Mel frequency spectrograms through a preset convolutional neural network to obtain the characteristic convolutional neural network model.
Preferably, in the step S3, if the characteristic convolutional neural network model successfully identifies the characteristic mel frequency spectrum diagram, a confirmed disease result is output, and if the characteristic convolutional neural network model cannot identify the characteristic mel frequency spectrum diagram, an unconfirmed disease result is output.
Preferably, the step S1 includes the following steps:
collecting cough sounds emitted by a detected object in real time, and performing framing processing on the cough sounds to obtain a plurality of frames of sound samples to be detected; wherein the time length of each frame of the sound sample to be detected is 2-5 seconds;
carrying out noise reduction processing on the sound sample to be detected to obtain the sound information to be detected;
carrying out Fourier transform processing on the sound information to be detected to obtain a characteristic Mel frequency cepstrum coefficient; and/or the presence of a gas in the gas,
the step S1a includes the following steps:
collecting cough sounds emitted by a patient in real time, and performing framing processing on the cough sounds to obtain a plurality of frames of training sound samples; wherein the time length of each frame of the training sound sample is 2-5 seconds;
carrying out noise reduction processing on the training sound sample to obtain the training sound information;
and carrying out Fourier transform processing on the training sound information to obtain training Mel frequency cepstrum coefficients.
Preferably, in step S3a, the characteristic convolutional neural network model is formed by jointly merging a plurality of initiation layers, each initiation layer is formed by a plurality of convolutions with different scales, and each convolution includes one of short-time sound information and long-time sound information.
The invention provides a disease detection system based on cough sound recognition, which comprises:
the sound acquisition module is used for acquiring cough sound emitted by the detected object or the patient in real time;
the data processing module is used for preprocessing the cough sound emitted by the detected object to obtain sound information to be detected and extracting a characteristic Mel frequency cepstrum coefficient of the sound information to be detected, or preprocessing the cough sound emitted by the patient to obtain training sound information and extracting a training Mel frequency cepstrum coefficient of the training sound information; the system is used for drawing a characteristic Mel frequency spectrogram by taking time and frequency as axes according to the characteristic Mel frequency cepstrum coefficient, or respectively drawing a plurality of training Mel frequency spectrograms by taking time and frequency as axes according to a plurality of training Mel frequency cepstrum coefficients;
the model training module is used for training a plurality of training Mel frequency spectrograms through a preset convolutional neural network to obtain the characteristic convolutional neural network model; and the number of the first and second groups,
and the judgment and identification module is used for identifying the characteristic Mel frequency spectrogram by using a characteristic convolution neural network model and outputting a disease diagnosis result according to an identification result.
Preferably, the judgment and identification module is configured to output a confirmed disease result when the characteristic convolutional neural network model successfully identifies the characteristic mel frequency spectrum, and output an unconfirmed disease result when the characteristic convolutional neural network model cannot identify the characteristic mel frequency spectrum.
Preferably, the data processing module is configured to perform framing processing on the cough sound of the object to be tested to obtain multiple frames of sound samples to be tested, where the time length of each frame of the sound sample to be tested is 2 to 5 seconds, perform noise reduction processing on the sound sample to be tested to obtain the sound information to be tested, and perform fourier transform processing on the sound information to be tested to obtain a characteristic mel-frequency cepstrum coefficient; the method is further used for framing the patient cough sound to obtain a plurality of frames of the training sound samples, wherein the time length of each frame of the training sound samples is 2-5 seconds, denoising is conducted on the training sound samples to obtain the training sound information, and Fourier transform is conducted on the training sound information to obtain training Mel frequency cepstrum coefficients.
Preferably, the characteristic convolution neural network model is formed by jointly combining a plurality of initiation layers, each initiation layer is formed by convolution of a plurality of different scales, and each convolution comprises one of short-time sound information and long-time sound information.
The invention provides a cough sound recognition-based disease detection system, which comprises a processor and a memory, wherein a control program of a cough sound recognition-based disease detection method is stored in the memory, and the control program realizes the steps of the cough sound recognition-based disease detection method when being executed by the processor.
The present invention provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the steps of the cough sound recognition-based disorder detection method of the present invention.
Compared with the related technology, the disease detection method based on cough sound recognition can be applied to detection of the Covid-19 disease, in the actual use process, the cough sound of a Covid-19 patient is collected and is used as the original training sample of the preset convolutional neural network to obtain the characteristic convolutional neural network model capable of recognizing the Covid-19 disease, the cough sound of the tested object is directly collected during detection, the method can process and analyze the cough sound of the tested object, the characteristic convolutional neural network model is used for recognition, the diagnosis result of the Covid-19 disease of the tested person is output according to the recognition result, here, the diagnosis of the Covid-19 disease is carried out in a sound characteristic recognition mode, the rapid and convenient detection of the Covid-19 virus pneumonia is realized, the disease detection efficiency is improved, and the required equipment has low cost, and the detection operation is simple and convenient, thereby being beneficial to popular use and social popularization.
[ description of the drawings ]
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:
FIG. 1 is a schematic flow chart of a method for detecting a disease condition based on cough recognition according to the present invention;
FIG. 2 is a schematic flow chart of a method for detecting a cough sound-based condition according to the present invention;
FIG. 3 is a schematic diagram of a cough recognition-based disease detection system according to the present invention;
FIG. 4 is a schematic diagram of obtaining Mel frequency cepstrum coefficients after the voice information is processed by MFCC feature extraction;
FIG. 5 is a schematic diagram of the construction of each initiation layer of the characteristic convolutional neural network model of the present invention;
FIG. 6 is a schematic structural diagram of a characteristic convolutional neural network model of the present invention.
[ detailed description ] embodiments
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1-2, the present invention provides a method for detecting a disease condition based on cough sound recognition, the method comprising the following steps:
step S1, acquiring sound information to be detected in real time, and extracting a characteristic Mel frequency cepstrum coefficient of the sound information to be detected; the sound information to be detected is sent out by the detected object;
step S2, drawing a characteristic Mel frequency spectrogram by taking time and frequency as axes according to the characteristic Mel frequency cepstrum coefficient;
and step S3, identifying the characteristic Mel frequency spectrogram by using a characteristic convolution neural network model, and outputting a disease diagnosis result according to an identification result.
The method for acquiring the characteristic convolution neural network model comprises the following steps:
step S1a, acquiring a plurality of training voice information, and respectively extracting training Mel frequency cepstrum coefficients of the training voice information; the training voice information is sent by a patient, and the training voice information and the training Mel frequency cepstrum coefficients are arranged in a one-to-one correspondence mode;
step S2a, respectively drawing a plurality of training Mel frequency spectrograms by taking time and frequency as axes according to a plurality of training Mel frequency cepstrum coefficients; the training Mel frequency cepstrum coefficients and the training Mel frequency spectrograms are arranged in a one-to-one correspondence mode;
and S3a, training the training Mel frequency spectrograms through a preset convolutional neural network to obtain the characteristic convolutional neural network model.
Referring to fig. 3, the present invention further provides a disease detection system 100 based on cough recognition, which includes a sound collection module 1, a data processing module 2, a model training module 3, and a judgment and recognition module 4, wherein the above modules may be integrated on the same terminal device, specifically, the data processing module 2 is respectively connected with the sound collection module 1, the model training module 3, and the judgment and recognition module 4 to realize data transmission, and the model training module 3 is connected with the judgment and recognition module 4.
The sound acquisition module 1 is used for acquiring cough sound in real time and sending the cough sound to the data processing module 2; for example, in the present embodiment, the sound collection module 1 is a microphone.
A data processing module 2, serving as a central center of data processing, configured to receive a cough sound, pre-process the cough sound to obtain sound information, and extract a Mel Frequency Cepstrum Coefficient (MFCC) of the sound information; the method is used for drawing a Mel frequency spectrogram by taking time and frequency as axes according to the Mel frequency cepstrum coefficient.
The model training module 3 is configured to train a plurality of training mel frequency spectrograms through a preset Convolutional Neural Network (CNN) to obtain a trained Convolutional Neural network model.
And the judgment and identification module 4 is used for identifying the Mel frequency spectrogram to be detected by using the trained convolutional neural network model and outputting a disease diagnosis result according to the identification result.
Referring to fig. 1 to 3, the present invention provides a method for detecting a disease based on cough sound recognition, which is applied to the disease detection system 100, and the method is described in detail below with reference to the specific structure of the disease detection system 100:
the method comprises the following steps of firstly, obtaining a characteristic convolution neural network model, and comprising the following substeps:
step S1a, acquiring, by the sound acquisition module 1, the cough sound emitted by the patient in real time and sending the acquired cough sound to the data processing module 2, preprocessing the cough sound emitted by the patient by the data processing module 2 to obtain training sound information, and extracting a training mel-frequency cepstrum coefficient of the training sound information, specifically, a process of acquiring the training mel-frequency cepstrum coefficient after the training sound information is processed by MFCC feature extraction is shown in fig. 4; the patient is a patient with infectious diseases, including, but not limited to, influenza, SARS (atypical pneumonia virus) Disease, MERS (Middle East Respiratory Syndrome Coronavirus) Disease, and Covid-19(corona diseases 2019, 2019 Coronavirus) Disease, which are extremely contagious diseases accompanied by cough symptoms, and the patient is preferably a patient with infectious diseases, such as influenza, SARS (Middle East Respiratory Syndrome Coronavirus), and is specifically selected according to actual use conditions, for example, a global Covid-19 Disease epidemic situation is developed in 2019, and the patient is preferably a patient with Covid-19 in the present embodiment, in order to better prevent and control the spread of the epidemic situation.
The training sound information and the training Mel frequency cepstrum coefficients are arranged in a one-to-one correspondence mode.
Specifically, the step S1a specifically includes the following sub-steps:
the sound collection module 1 collects the cough sound sent by the patient in real time, the data processing module 2 receives the cough sound, and the cough sound is subjected to framing processing to obtain a plurality of frames of training sound samples; wherein, the time length of each frame of the training sound sample is 2-5 seconds, and the time length of each frame can be specifically selected according to the requirements of practical application; performing noise reduction processing on the training voice sample through the data processing module 2 to obtain the training voice information; in the present embodiment, the time length of each training voice information is determined by the training voice sample of each frame, and for example, in the present embodiment, the time length of the training voice sample of each frame may be 2 seconds, 3 seconds, 4 seconds, or 5 seconds, wherein the training voice information corresponding to the training voice sample having the time length of 2 seconds or 3 seconds is used as the short-time voice information, and the training voice information corresponding to the training voice sample having the time length of 4 seconds or 5 seconds is used as the long-time voice information.
And performing Fast Fourier Transform (FFT) on the training sound information through the data processing module 2 to obtain a training mel-frequency cepstrum coefficient.
Step S2a, respectively drawing a plurality of training Mel frequency spectrograms by the data processing module 2 according to a plurality of training Mel frequency cepstrum coefficients and taking time and frequency as axes, and sending the training Mel frequency spectrograms to the model training module 3; and the training Mel frequency cepstrum coefficients and the training Mel frequency spectrograms are arranged in a one-to-one correspondence mode.
Step S3a, training the training mel-frequency spectrograms through a preset convolutional neural network by the model training module 3, to obtain the characteristic convolutional neural network model, and referring to fig. 6 for the structure of the characteristic convolutional neural network model. More specifically, the characteristic convolution neural network model is formed by jointly combining a plurality of initiation layers, each initiation layer is formed by convolution of a plurality of different scales, and each convolution comprises one of short-time sound information and long-time sound information, so that the characteristic convolution neural network model can cover training sound information of various different time lengths, and the identification range of the characteristic convolution neural network model to sound information to be detected of different time lengths is effectively widened. Secondly, identifying the cough sound of the detected object in real time according to the characteristic convolution neural network model, and outputting a diagnosis result, wherein the method comprises the following substeps:
step S1, the sound collection module 1 collects the cough sound emitted by the detected object in real time, the data processing module 2 preprocesses the cough sound to obtain the sound information to be detected, and extracts the characteristic mel-frequency cepstrum coefficient of the sound information to be detected.
Specifically, the step S1 specifically includes the following sub-steps:
the sound collection module 1 collects the cough sound emitted by the detected object in real time and sends the cough sound to the data processing module 2, and the data processing module 2 performs framing processing on the cough sound to obtain a plurality of frames of sound samples to be detected; in the embodiment, the time length of each frame of the sound sample to be detected may be 2 to 5 seconds, and the time length may be determined according to an actual situation, and in the embodiment, the time length of each frame of the sound sample to be detected may be 2 seconds, 3 seconds, 4 seconds, or 5 seconds;
performing noise reduction processing on the sound sample to be detected through the data processing module 2 to obtain the sound information to be detected;
and carrying out Fourier transform processing on the sound information to be detected through the data processing module 2 to obtain a characteristic Mel frequency cepstrum coefficient.
Step S2, drawing a characteristic mel frequency spectrum diagram by the data processing module 2 according to the characteristic mel frequency cepstrum coefficient and using time and frequency as axes, and sending the characteristic mel frequency spectrum diagram to the model training module 3.
And step S3, recognizing the characteristic Mel frequency spectrum diagram by the model training module 3 through a characteristic convolution neural network model, and outputting a disease diagnosis result according to the recognition result through the judgment and recognition module 4.
Specifically, in the step S3, if the characteristic convolutional neural network model successfully identifies the characteristic mel frequency spectrum, the judgment and identification module 4 outputs a confirmed disease result, that is, the detected object has Covid-19 disease, and if the characteristic convolutional neural network model cannot identify the characteristic mel frequency spectrum, the judgment and identification module 4 outputs an unconfirmed disease result, that is, the detected object does not have Covid-19 disease.
The invention provides a cough sound recognition-based disease detection system, which comprises a processor and a memory, wherein a control program of a cough sound recognition-based disease detection method is stored in the memory, and the control program realizes the steps of the cough sound recognition-based disease detection method when being executed by the processor.
The present invention provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the steps of the cough sound recognition-based disorder detection method of the present invention.
Compared with the related technology, the disease detection method based on cough sound recognition can be applied to detection of the Covid-19 disease, in the actual use process, the cough sound of a Covid-19 patient is collected and is used as the original training sample of the preset convolutional neural network to obtain the characteristic convolutional neural network model capable of recognizing the Covid-19 disease, the cough sound of the tested object is directly collected during detection, the method can process and analyze the cough sound of the tested object, the characteristic convolutional neural network model is used for recognition, the diagnosis result of the Covid-19 disease of the tested person is output according to the recognition result, here, the diagnosis of the Covid-19 disease is carried out in a sound characteristic recognition mode, the rapid and convenient detection of the Covid-19 virus pneumonia is realized, the disease detection efficiency is improved, and the required equipment has low cost, and the detection operation is simple and convenient, thereby being beneficial to popular use and social popularization.
While the foregoing is directed to embodiments of the present invention, it will be understood by those skilled in the art that various changes may be made without departing from the spirit and scope of the invention.
Claims (10)
1. A method for detecting a condition based on cough sound recognition, the method comprising the steps of:
step S1, acquiring sound information to be detected in real time, and extracting a characteristic Mel frequency cepstrum coefficient of the sound information to be detected; the sound information to be detected is sent out by the detected object;
step S2, drawing a characteristic Mel frequency spectrogram by taking time and frequency as axes according to the characteristic Mel frequency cepstrum coefficient;
step S3, identifying the characteristic Mel frequency spectrum diagram by using a characteristic convolution neural network model, and outputting a disease diagnosis result according to an identification result; the method for acquiring the characteristic convolution neural network model comprises the following steps:
step S1a, acquiring a plurality of training voice information, and respectively extracting training Mel frequency cepstrum coefficients of the training voice information; the training voice information is sent by a patient, and the training voice information and the training Mel frequency cepstrum coefficients are arranged in a one-to-one correspondence mode;
step S2a, respectively drawing a plurality of training Mel frequency spectrograms by taking time and frequency as axes according to a plurality of training Mel frequency cepstrum coefficients; the training Mel frequency cepstrum coefficients and the training Mel frequency spectrograms are arranged in a one-to-one correspondence mode;
and S3a, training the training Mel frequency spectrograms through a preset convolutional neural network to obtain the characteristic convolutional neural network model.
2. The cough sound identification-based disease detection method of claim 1, wherein in the step S3, if the characteristic convolutional neural network model successfully identifies the characteristic mel frequency spectrum diagram, a confirmed disease result is output, and if the characteristic convolutional neural network model cannot identify the characteristic mel frequency spectrum diagram, an unconfirmed disease result is output.
3. The cough sound recognition based condition detection method according to claim 1, wherein the step S1 includes the steps of:
collecting cough sounds emitted by a detected object in real time, and performing framing processing on the cough sounds to obtain a plurality of frames of sound samples to be detected; wherein the time length of each frame of the sound sample to be detected is 2-5 seconds;
carrying out noise reduction processing on the sound sample to be detected to obtain the sound information to be detected;
carrying out Fourier transform processing on the sound information to be detected to obtain a characteristic Mel frequency cepstrum coefficient; and/or the presence of a gas in the gas,
the step S1a includes the following steps:
collecting cough sounds emitted by a patient in real time, and performing framing processing on the cough sounds to obtain a plurality of frames of training sound samples; wherein the time length of each frame of the training sound sample is 2-5 seconds;
carrying out noise reduction processing on the training sound sample to obtain the training sound information;
and carrying out Fourier transform processing on the training sound information to obtain training Mel frequency cepstrum coefficients.
4. The method for detecting a cough sound recognition-based disease state according to claim 1, wherein in step S3a, the characteristic convolutional neural network model is formed by jointly combining a plurality of initiation layers, each initiation layer is formed by convolution of a plurality of different scales, and each convolution includes one of short-time sound information and long-time sound information.
5. A cough sound recognition based condition detection system, comprising:
the sound acquisition module is used for acquiring cough sound emitted by the detected object or the patient in real time;
the data processing module is used for preprocessing the cough sound emitted by the detected object to obtain sound information to be detected and extracting a characteristic Mel frequency cepstrum coefficient of the sound information to be detected, or preprocessing the cough sound emitted by the patient to obtain training sound information and extracting a training Mel frequency cepstrum coefficient of the training sound information; the system is used for drawing a characteristic Mel frequency spectrogram by taking time and frequency as axes according to the characteristic Mel frequency cepstrum coefficient, or respectively drawing a plurality of training Mel frequency spectrograms by taking time and frequency as axes according to a plurality of training Mel frequency cepstrum coefficients;
the model training module is used for training the training Mel frequency spectrograms through a preset convolutional neural network to obtain the convolutional neural network model; and the number of the first and second groups,
and the judgment and identification module is used for identifying the characteristic Mel frequency spectrogram by using a characteristic convolution neural network model and outputting a disease diagnosis result according to an identification result.
6. The fingerprint detection apparatus of claim 4, wherein the decision identification module is configured to output a confirmed disease result when the characteristic convolutional neural network model successfully identifies the characteristic Mel frequency spectrogram, and output an unconfirmed disease result when the characteristic convolutional neural network model fails to identify the characteristic Mel frequency spectrogram.
7. The fingerprint detection device according to claim 5, wherein the data processing module is configured to perform framing processing on the cough sound of the object to be detected to obtain multiple frames of sound samples to be detected, where a time length of each frame of the sound samples to be detected is 2-5 seconds, perform noise reduction processing on the sound samples to be detected to obtain the sound information to be detected, and perform fourier transform processing on the sound information to be detected to obtain the characteristic mel-frequency cepstrum coefficients; the method is further used for framing the patient cough sound to obtain a plurality of frames of the training sound samples, wherein the time length of each frame of the training sound samples is 2-5 seconds, denoising is conducted on the training sound samples to obtain the training sound information, and Fourier transform is conducted on the training sound information to obtain training Mel frequency cepstrum coefficients.
8. The fingerprint detection apparatus according to claim 5, wherein the characteristic convolutional neural network model is formed by jointly combining a plurality of initiation layers, each initiation layer is formed by convolution of a plurality of different scales, and each convolution includes one of short-time sound information and long-time sound information.
9. A cough sound recognition-based disease detection system, wherein the fingerprint detection device comprises a processor and a memory, the memory stores a control program of the cough sound recognition-based disease detection method, wherein the control program is executed by the processor to implement the steps of the cough sound recognition-based disease detection method according to any one of claims 1 to 4.
10. A computer-readable storage medium, storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the cough sound recognition based condition detection method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011298720.2A CN112472065A (en) | 2020-11-18 | 2020-11-18 | Disease detection method based on cough sound recognition and related equipment thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011298720.2A CN112472065A (en) | 2020-11-18 | 2020-11-18 | Disease detection method based on cough sound recognition and related equipment thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112472065A true CN112472065A (en) | 2021-03-12 |
Family
ID=74931797
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011298720.2A Pending CN112472065A (en) | 2020-11-18 | 2020-11-18 | Disease detection method based on cough sound recognition and related equipment thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112472065A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113096691A (en) * | 2021-03-22 | 2021-07-09 | 深圳市安保科技有限公司 | Detection method, device, equipment and computer storage medium |
| CN113724731A (en) * | 2021-08-30 | 2021-11-30 | 中国科学院声学研究所 | Method and device for audio discrimination by using audio discrimination model |
| CN113804767A (en) * | 2021-08-16 | 2021-12-17 | 东南大学 | Bolt failure detection method |
| CN114202892A (en) * | 2021-11-16 | 2022-03-18 | 北京航天试验技术研究所 | Hydrogen leakage monitoring method |
| CN114209302A (en) * | 2021-12-08 | 2022-03-22 | 赵永源 | Cough detection method based on data uncertainty learning |
| CN114330454A (en) * | 2022-01-05 | 2022-04-12 | 东北农业大学 | Live pig cough sound identification method based on DS evidence theory fusion characteristics |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120265024A1 (en) * | 2010-10-05 | 2012-10-18 | University Of Florida Research Foundation, Incorporated | Systems and methods of screening for medical states using speech and other vocal behaviors |
| CN106847262A (en) * | 2016-12-28 | 2017-06-13 | 华中农业大学 | A kind of porcine respiratory disease automatic identification alarm method |
| CN108135485A (en) * | 2015-10-08 | 2018-06-08 | 康迪欧医疗有限公司 | Lung conditions are assessed by speech analysis |
| CN109431507A (en) * | 2018-10-26 | 2019-03-08 | 平安科技（深圳）有限公司 | Cough disease identification method and device based on deep learning |
| CN110946554A (en) * | 2019-11-27 | 2020-04-03 | 深圳和而泰家居在线网络科技有限公司 | Cough type identification method, device and system |
| CN111523470A (en) * | 2020-04-23 | 2020-08-11 | 苏州浪潮智能科技有限公司 | Feature fusion block, convolutional neural network, pedestrian re-identification method and related equipment |
| CN111625763A (en) * | 2020-05-27 | 2020-09-04 | 郑州航空工业管理学院 | Operation risk prediction method and prediction system based on mathematical model |
| US20200323484A1 (en) * | 2017-12-14 | 2020-10-15 | Vocalis Health | Method and system for screening for covid-19 with a vocal biomarker |
| CN111951824A (en) * | 2020-08-14 | 2020-11-17 | 苏州国岭技研智能科技有限公司 | Detection method for distinguishing depression based on sound |
-
2020
- 2020-11-18 CN CN202011298720.2A patent/CN112472065A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120265024A1 (en) * | 2010-10-05 | 2012-10-18 | University Of Florida Research Foundation, Incorporated | Systems and methods of screening for medical states using speech and other vocal behaviors |
| CN108135485A (en) * | 2015-10-08 | 2018-06-08 | 康迪欧医疗有限公司 | Lung conditions are assessed by speech analysis |
| CN106847262A (en) * | 2016-12-28 | 2017-06-13 | 华中农业大学 | A kind of porcine respiratory disease automatic identification alarm method |
| US20200323484A1 (en) * | 2017-12-14 | 2020-10-15 | Vocalis Health | Method and system for screening for covid-19 with a vocal biomarker |
| CN109431507A (en) * | 2018-10-26 | 2019-03-08 | 平安科技（深圳）有限公司 | Cough disease identification method and device based on deep learning |
| CN110946554A (en) * | 2019-11-27 | 2020-04-03 | 深圳和而泰家居在线网络科技有限公司 | Cough type identification method, device and system |
| CN111523470A (en) * | 2020-04-23 | 2020-08-11 | 苏州浪潮智能科技有限公司 | Feature fusion block, convolutional neural network, pedestrian re-identification method and related equipment |
| CN111625763A (en) * | 2020-05-27 | 2020-09-04 | 郑州航空工业管理学院 | Operation risk prediction method and prediction system based on mathematical model |
| CN111951824A (en) * | 2020-08-14 | 2020-11-17 | 苏州国岭技研智能科技有限公司 | Detection method for distinguishing depression based on sound |
Non-Patent Citations (2)
| Title |
|---|
| FILIPE BARATA,ET AL: "Towards Device-Agnostic Mobile Cough Detection", 《2019 IEEE INTERNATIONAL CONFERENCE ON HEALTHCARE INFORMATICS》 * |
| VIPIN BANSAL,ET AL: "Cough Classification for COVID-19 based on audio mfcc features using Convolutional Neural Networks", 《2020 IEEE INTERNATIONAL CONFERENCE ON COMPUTING, POWER AND COMMUNICATION TECHNOLOGIES》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113096691A (en) * | 2021-03-22 | 2021-07-09 | 深圳市安保科技有限公司 | Detection method, device, equipment and computer storage medium |
| CN113804767A (en) * | 2021-08-16 | 2021-12-17 | 东南大学 | Bolt failure detection method |
| CN113804767B (en) * | 2021-08-16 | 2022-11-04 | 东南大学 | Bolt failure detection method |
| CN113724731A (en) * | 2021-08-30 | 2021-11-30 | 中国科学院声学研究所 | Method and device for audio discrimination by using audio discrimination model |
| CN113724731B (en) * | 2021-08-30 | 2024-01-05 | 中国科学院声学研究所 | Method and device for carrying out audio discrimination by utilizing audio discrimination model |
| CN114202892A (en) * | 2021-11-16 | 2022-03-18 | 北京航天试验技术研究所 | Hydrogen leakage monitoring method |
| CN114209302A (en) * | 2021-12-08 | 2022-03-22 | 赵永源 | Cough detection method based on data uncertainty learning |
| CN114209302B (en) * | 2021-12-08 | 2024-03-26 | 赵永源 | Cough detection method based on data uncertainty learning |
| CN114330454A (en) * | 2022-01-05 | 2022-04-12 | 东北农业大学 | Live pig cough sound identification method based on DS evidence theory fusion characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112472065A (en) | Disease detection method based on cough sound recognition and related equipment thereof | |
| CN108648748B (en) | Acoustic event detection method under hospital noise environment | |
| Belkacem et al. | End-to-end AI-based point-of-care diagnosis system for classifying respiratory illnesses and early detection of COVID-19: a theoretical framework | |
| WO2019104890A1 (en) | Fraud identification method and device combining audio analysis and video analysis and storage medium | |
| TWI473080B (en) | The use of phonological emotions or excitement to assist in resolving the gender or age of speech signals | |
| Hsu et al. | Benchmarking of eight recurrent neural network variants for breath phase and adventitious sound detection on a self-developed open-access lung sound database—HF_Lung_V1 | |
| Liu et al. | Detection of adventitious respiratory sounds based on convolutional neural network | |
| AU2013239327A1 (en) | A method and apparatus for processing patient sounds | |
| WO2021114761A1 (en) | Lung rale artificial intelligence real-time classification method, system and device of electronic stethoscope, and readable storage medium | |
| WO2021054742A1 (en) | Method, system, and non-transitory computer-readable recording media for analyzing breathing-related sound | |
| CN107818366A (en) | A kind of lungs sound sorting technique, system and purposes based on convolutional neural networks | |
| CN112294253A (en) | Disease diagnosis system based on user voice change and household intelligent robot | |
| CN112382302A (en) | Baby cry identification method and terminal equipment | |
| CN113053410A (en) | Voice recognition method, voice recognition device, computer equipment and storage medium | |
| Usman et al. | Speech as A Biomarker for COVID-19 detection using machine learning | |
| CN116019467A (en) | Feature extraction method of facial nerve electromyographic signals | |
| Dam et al. | e-Breath: breath detection and monitoring using frequency cepstral feature fusion | |
| Mendes et al. | Subvocal speech recognition based on EMG signal using independent component analysis and neural network MLP | |
| CN112634871B (en) | Lie detection method and system based on voice and radar dual sensors | |
| Rashid et al. | CoughNet-V2: A scalable multimodal DNN framework for point-of-care edge devices to detect symptomatic COVID-19 cough | |
| CN114520062A (en) | Medical cloud communication system based on AI and letter creation | |
| CN113823291A (en) | Voiceprint recognition method and system applied to power operation | |
| WO2022205400A1 (en) | Voice recognition-based safety alerting method and apparatus, and terminal device | |
| Khan et al. | Classification of pulmonary crackles and pleural friction rubs using MFCC statistical parameters | |
| Jayadi et al. | Embedded Flu Detection System based Cough Sound using MFCC and kNN Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210312 |