Digital stethoscope system based on piezoelectric film

Abstract

The invention discloses a digital stethoscope system based on a piezoelectric film, and relates to the technical field of medical electronics and biomedical engineering. An existing digital stethoscope based on a piezoelectric film has the challenges of signal noise and interference, limitation of a signal processing algorithm, nonlinear response and time migration, insufficient diagnosis accuracy and automation and the like. A piezoelectric film and microphone dual audio acquisition mode is fused, various physiological signals such as low-frequency vibration and high-quality heart and lung sound are captured, and background noise is removed through a self-adaptive filter and a frequency spectrum subtraction technology; an MFCC feature extraction and dynamic time warping algorithm is utilized to carry out fine feature analysis and time axis alignment on the preprocessed audio signal, and high-precision matching with a storage template is realized; different physiological states are accurately distinguished through intelligent feature processing, real-time and visual diagnosis feedback is achieved through wireless transmission and state indication, and the reliability and efficiency of medical detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of medical electronics and biomedical engineering, and specifically to a digital stethoscope system based on piezoelectric thin films. Background Art

[0002] The digital stethoscope based on piezoelectric thin films utilizes the piezoelectric effect to convert the weak physiological vibration signals of the human body into electrical signals, and through high-precision signal amplification, filtering, and digital signal processing technologies, realizes the high-quality acquisition and analysis of physiological sounds such as heart sounds and lung sounds; this technology has the advantages of being portable, non-invasive, and highly sensitive, and can be widely applied to intelligent healthcare, remote monitoring, and artificial intelligence-assisted diagnosis;

[0003] However, it still faces challenges such as signal noise and interference problems, limitations of signal processing algorithms, nonlinear response and time offset problems, and insufficient diagnostic accuracy and automation; affecting its stability and accuracy in clinical applications. Summary of the Invention

[0004] The purpose of the present invention is to solve the above problems and propose a digital stethoscope system based on piezoelectric thin films.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A digital stethoscope system based on piezoelectric thin films, comprising: a smart wearable stethoscope, a collection and analysis module, and a feature processing module;

[0006] The smart wearable stethoscope is composed of a stethoscope watch (3), a chest strap (1), and a buckle (2); wherein, the stethoscope watch (3) includes an operation button (11), an operation button (12), a chest strap interface (13), an integration hole (14), a stethoscope battery compartment (15), and a piezoelectric thin film (16); wherein the chest strap (1) is connected to the chest strap interface (13), and the buckle (2) is installed on the chest strap (1) and is used to adjust the length of the chest strap; the chest strap interface (13), the integration hole (14), the operation button (11), the stethoscope battery compartment (15), and the piezoelectric thin film (16) are all installed on the stethoscope body (12);

[0007] The collection and analysis module obtains the audio data of the smart wearable stethoscope through a wireless module and differentiates it into microphone audio data and piezoelectric thin film audio data; the microphone audio data is used for heart sound and lung sound analysis to identify heart diseases or breathing problems; the piezoelectric thin film audio data is used for low-frequency signal analysis to identify heart vibrations and pulse fluctuations;

[0008] The microphone audio data is denoised using an adaptive filter, and the LMS algorithm is used to dynamically adjust the filter weights to optimize the noise removal effect; the piezoelectric film audio data is denoised using spectral subtraction. The noise spectrum is estimated through short-time Fourier transform and subtracted from the signal; the inverse STFT is used to restore the time-domain signal, and smoothing processing and signal enhancement are performed to optimize the denoising effect;

[0009] The feature processing module performs unified format conversion and feature analysis on the denoised microphone and piezoelectric film audio data to obtain the similarity XSD; if XSD is lower than the preset threshold, it is determined as different audio features, otherwise it is regarded as the same audio;

[0010] The feature processing module performs audio feature matching, compares the acquired data with the stored preset audio template, and calculates the matching degree PPD; when PPD reaches the threshold, it is judged as a successful match, and the corresponding physiological state or abnormal mode is output; the result is displayed through an indicator light.

[0011] As a preferred embodiment of the present invention, the specific process of distinguishing the microphone audio data and the piezoelectric film audio data is as follows:

[0012] The data file format of the acquired audio data is distinguished and matched with the format corresponding to the microphone audio data. If the two are consistent, the acquired audio data is the microphone audio data; similarly, the file format corresponding to the piezoelectric film audio data is matched with the data file format of the acquired audio data. If the two are consistent, the acquired audio data is the piezoelectric film audio data.

[0013] As a preferred embodiment of the present invention, the specific process of denoising the microphone audio data using an adaptive filter is as follows:

[0014] The adaptive filter is retrieved for processing, and the adaptive filter automatically adjusts the coefficients of the filter according to the change of the input signal; the adaptive filter uses the LMS algorithm for calculation. The mixed audio signal x(n) containing the target signal and noise and the introduced noise reference signal d(n) are substituted into the adaptive filter to output the microphone filtered signal y(n); from the formula: Output the microphone filtered signal y(n), w k (n) is the weight coefficient of the filter, M is the order of the filter, and n is the current time; then calculate the error between the filter output y(n) and the actually acquired signal x(n). Through the formula: e(n) = x(n) - y(n), the error e(n) represents the noise part that the filter fails to effectively remove; the weight coefficient w of the filter is adjusted through the error e(n) k (n); the LMS algorithm uses the gradient descent method to update the weight: w k (n + 1) = w k(n) + μ·e(n)·d(n - k), where μ is the step factor that controls the amplitude of each update; (n - k) is the delayed version of the reference signal.

[0015] As a preferred embodiment of the present invention, the specific process of using spectral subtraction to denoise the piezoelectric thin film audio data is as follows:

[0016] Invoke spectral subtraction for denoising:

[0017] Step 1: When there is no target information in the signal, these time periods are regarded as pure noise segments; within the time periods without signal, calculate the spectral characteristics of the noise; use the short-time Fourier transform to extract the spectral information of the signal: through the formula: t ∈ segments without signal, is the noise spectral estimate at time t and frequency f; X(f,t) is the result of the Fourier transform of the signal; N is the number of segments without signal used for noise estimation;

[0018] Step 2: After the noise spectral estimate, subtract the noise spectrum from the spectrum of the original signal; use the short-time Fourier transform to decompose the noisy signal into spectra; the spectrum of the signal is represented as X(f,t), which contains the mixture of the target signal and noise; for each time-frequency point, subtract the estimated noise spectrum from the spectrum |X(f,t)| of the signal; introduce a threshold to obtain the formula:

[0019] is the spectrum of the denoised signal; the max function ensures that the spectral value will not be negative;

[0020] Step 3: Convert the processed spectrum back to the time-domain signal through the inverse short-time Fourier transform to obtain the denoised audio signal, by the formula: x c (t) is the time-domain signal after denoising, and ISTFT is the inverse short-time Fourier transform operation;

[0021] Step 4: After completing spectral subtraction, reduce the high-frequency residual noise through smoothing to maintain the smoothness of the signal; for the still weak target signal after denoising, enhance the signal strength through an enhancement algorithm.

[0022] As a preferred embodiment of the present invention, the specific process of obtaining the similarity XSD is as follows:

[0023] The audio data is framed according to a fixed duration and smoothed using a window function; MFCC is used for feature extraction, including pre-emphasis, FFT transformation, power spectrum calculation, Mel filter bank processing, logarithmic transformation, and DCT transformation, and finally an MFCC feature vector is obtained; to solve the non-linear offset on the time axis, the DTW algorithm is used to perform optimal alignment on the MFCC sequence, calculate the cumulative matching error, and normalize to obtain the similarity XSD.

[0024] As a preferred embodiment of the present invention, the specific process of framing the audio data according to a fixed duration and smoothing it using a window function is as follows:

[0025] The synchronized audio data is framed according to a fixed duration, and each frame contains N1 = sampling rate × 0.025 sampling points; frame overlap is set to smooth the signal transition; then the window function calculation is performed on the data of each frame through the formula: Output the windowed frame data x i,ω (n1).

[0026] As a preferred embodiment of the present invention, the specific process of MFCC for feature extraction is as follows:

[0027] For each frame of audio data, MFCC is used for feature extraction: pre-emphasis is performed on the signal to enhance the high-frequency components: x p (n1) = x i1,ω (n1) - α·x i1,ω (n1 - 1), α = 0.95; the pre-emphasized signal is subjected to a fast Fourier transform to convert the time-domain signal to the frequency domain to obtain the spectrum X i1 (k); based on this spectrum, the power spectrum of each frame is calculated through the formula: P(k) = |X i1 (k)| 2 ; then, according to the non-linear characteristics of the human ear's sensitivity to different frequencies, a Mel filter bank is constructed to decompose the power spectrum in several Mel frequency bands, and the energy output E of each filter is calculated m , specifically: for the m-th filter in the Mel filter bank, let its response function be H m (k), usually, the shape of H m (k) is triangular, defined within a certain frequency range, and the value outside this range is 0; for the m-th filter, the energy is calculated by weighted summing the power spectrum within the frequency range covered by the filter response: k s and k e are respectively the start and end bins of the frequency range corresponding to the filter m in the FFT; H m(k) is the weight of the filter at frequency bin k. To prevent the energy of some filters from approaching or being equal to zero, a very small constant ∈ is added to ensure that the value is always greater than zero, that is, E' m = E m + ε; Using the natural logarithm or the common logarithm form, perform a logarithmic transformation on the output energy E' of each filter m Perform logarithmic transformation, using the natural logarithm: logE m = ln(E' m ); Finally, input the logarithmic energy sequence into the discrete cosine transform to obtain a set of discrete coefficients, and select the first 12 to 13 coefficients as the MFCC feature vector of this frame.

[0028] As a preferred embodiment of the present invention, the specific process of obtaining the similarity XSD is as follows:

[0029] Use the dynamic time warping algorithm to perform optimal alignment on the two sequences, and the specific process is as follows:

[0030] Let the MFCC feature vector of the i2-th frame of the microphone be The piezoelectric film feature vector of the j2-th frame is The distance between them is defined as the Euclidean distance of the two vectors: D1 is the dimension of the MFCC feature; Let D(i2, j2) represent the minimum cumulative distance between the first i2 frames and the first j2 frames; Use the recursive update formula: D(i2, j2) = d(i2, j2)+min{D(i2 - 1, j2), D(i2, j2 - 1), D(i2 - 1, j2 - 1)}, boundary conditions: starting point: D(1, 1); The first row and the first column can be initialized by accumulation: D(i2, 1) = d(i2, 1)+min{D(i2 - 1, 1), D(i2 - 1, 1)}, D(1, j2) = d(1, j2)+min{D(1, j2), D(1, j2 - 1)}; Construct the entire DTW matrix, let the length of the optimal alignment path be L, then the normalized error is: N m and N p are the number of frames of the microphone and piezoelectric film sequences respectively; D n represents the overall matching error of the two sequences; Convert the normalized error into a similarity index between 0 and 1: Output the similarity XSD, where λ is a tuning parameter that controls the decay rate of the similarity.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] 1. The present invention adopts an integrated design, combines piezoelectric films with high-quality microphones, and realizes multimodal audio acquisition. Specifically, the piezoelectric films focus on capturing low-frequency signals such as heart vibrations and pulse fluctuations, while the microphones are used to capture high-fidelity cardiopulmonary sounds, breathing sounds, and friction and noise in the environment. During the acquisition process, the adaptive filter built into the system uses the LMS algorithm to adjust the filter weights according to the real-time changes of the input signal, effectively suppressing the noise components in the mixed signal. At the same time, for the audio data of the piezoelectric films, a spectral subtraction technique based on the short-time Fourier transform (STFT) is adopted. By extracting the spectral characteristics of the "pure noise" segment to estimate and subtract the noise spectrum, the clarity and signal-to-noise ratio of the signal are greatly improved. This multi-level and detailed noise suppression strategy provides a reliable data basis for subsequent feature extraction and diagnostic analysis, ensuring that the device can still obtain accurate and stable physiological signal data in complex environments.

[0033] 2. After signal preprocessing, the present invention converts to a unified file format, divides the frames with a fixed duration, and weights each frame of data with a window function to ensure time continuity and smooth transition. Subsequently, the system performs pre-emphasis processing on each frame of the signal, converts the time-domain signal to the frequency domain by means of the fast Fourier transform, and calculates the power spectrum of each frequency component. Combining the characteristics of the human ear's sensitivity to different frequencies, through weighted summation of the Mel filter bank, the energy output of each filter is obtained, and then through logarithmic conversion and discrete cosine transform, representative MFCC feature vectors are extracted. Since there may be a non-linear offset in time for the audio data obtained by the microphone and the piezoelectric film, the system further adopts the dynamic time warping algorithm for optimal sequence alignment. By calculating the normalized matching error and converting it into a similarity index XSD, the precise comparison of the two audio features is realized. Finally, the obtained matching degree is compared with the preset template features. When the matching degree reaches the threshold, the system can display the status in real time through the indicator light and transmit the diagnostic results to the intelligent device via the wireless module, providing accurate and intelligent support for the rapid identification of abnormal states such as heart diseases and breathing problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the drawings.

[0035] Figure 1 is the principle block diagram of the present invention;

[0036] Figure 2 is the schematic diagram of the intelligent wearable stethoscope of the present invention Figure 1 ;

[0037] Figure 3 is the schematic diagram of the intelligent wearable stethoscope of the present invention Figure 2 ;

[0038] Figure 4 Schematic diagram of the stethoscope of the present invention Figure 1 ;

[0039] Figure 5 Schematic diagram of the stethoscope of the present invention Figure 2 。

[0040] Description of the drawings: 1. Chest strap, 2. Buckle, 3. Stethoscope table, A. First surface of the stethoscope, B. Second surface of the stethoscope, 11. Operation button, 12. Operation button, 13. Chest strap interface, 14. Integration hole, 15. Stethoscope battery compartment, 16. Piezoelectric film. Detailed implementation manners

[0041] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0042] It should be understood that the terms "including" and "comprising" used in the specification and claims of this disclosure indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0043] It should also be understood that the terms used in this disclosure specification are only for the purpose of describing specific embodiments and are not intended to limit this disclosure. As used in this disclosure specification and claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms. It should be further understood that the term " / 1or" used in this disclosure specification and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0044] Please refer to Figure 1 as shown, the digital stethoscope system based on piezoelectric film includes: an intelligent wearable stethoscope, an acquisition and analysis module, and a feature processing module;

[0045] The intelligent wearable stethoscope consists of a stethoscope watch 3, a chest strap 1, and a buckle 2; among them, the stethoscope watch 3 includes an operation button 11, an operation button 12, a chest strap interface 13, an integration hole 14, a stethoscope battery compartment 15, and a piezoelectric film 16; the chest strap 1 is connected to the chest strap interface 13, and the buckle 3 is installed on the chest strap 1 and is used to adjust the length of the chest strap; the chest strap interface 13, the integration hole 14, the operation button 11, the stethoscope battery compartment 15, and the piezoelectric film 16 are all installed on the stethoscope body 12; it should be noted that: the intelligent wearable stethoscope internally integrates a wireless module (such as Bluetooth, Wi-Fi, or other low-power wireless technologies), a microphone, an indicator light, and a storage device; the wireless module is used to transmit the data of the stethoscope to other devices, the microphone part is responsible for capturing high-quality internal and external sounds of the human body, breathing sounds, friction sounds, noisy sounds, etc.; the indicator light is used to provide feedback on the device status, usually used to display information such as battery power, whether the device is connected, and whether it is working properly; the storage device is used to save the audio data and other diagnostic information recorded by the stethoscope;

[0046] The acquisition and analysis module obtains and analyzes the audio data recorded by the stethoscope in the storage device of the intelligent wearable stethoscope through the wireless module;

[0047] The audio data is differentiated into microphone audio data and piezoelectric film audio data. The specific differentiation method is as follows: obtain the data file format of the audio data for differentiation and match it with the format corresponding to the microphone audio data. If the two are consistent, the obtained audio data is microphone audio data; similarly, match the file format corresponding to the piezoelectric film audio data with the data file format of the obtained audio data. If the two are consistent, the obtained audio data is piezoelectric film audio data; it should be noted that: the microphone audio data is used for further analysis of heart sounds and lung sounds, and feature extraction to identify heart diseases or respiratory problems; the piezoelectric film audio data focuses on the analysis of low-frequency signals, such as the vibration of the heart and the fluctuation of the pulse, etc.;

[0048] Denoising processing is performed on the microphone audio data. An adaptive filter is adjusted for processing. The key of the adaptive filter is to be able to automatically adjust the coefficients of the filter according to the change of the input signal; the LMS algorithm is used for calculation through the adaptive filter. Specifically:

[0049] The mixed audio signal x(n) containing the target signal and noise and the introduced noise reference signal d(n) (a signal obtained from the source of environmental noise or other sensors of the microphone receiver; the reference signal is used to help the filter distinguish noise from the target signal) are substituted into the adaptive filter to output the microphone filtered signal y(n); by the formula: Output the microphone filtered signal y(n), where w k(n) is the weight coefficient of the filter, M is the order of the filter, and n is the current time; then calculate the error between the filter output y(n) and the actually acquired signal x(n) (i.e., the residual of the signal), which reflects the gap between the filter output and the actual signal. Through the formula: e(n) = x(n) - y(n), the error e(n) represents the part of the noise that the filter fails to effectively remove; adjust the weight coefficient w of the filter through the error e(n). k (n) so that the filter can remove noise more accurately in the next step; the LMS algorithm uses the gradient descent method to update the weights: w k (n + 1) = w k (n) + μ·e(n)·d(n - k), where μ is the step size factor that controls the amplitude of each update; (n - k) is the delayed version of the reference signal; by continuously updating the weights of the filter, the LMS algorithm can adapt to signal changes in real time, thereby effectively removing noise;

[0050] Perform denoising on the piezoelectric thin - film audio data, and retrieve the spectral subtraction method for denoising (by estimating the noise spectrum and subtracting it from the spectrum of the signal containing noise, thereby reducing the influence of noise and obtaining a clearer signal);

[0051] Step 1: When there is no target information in the signal (for example, when the patient does not make any sound); regard these time periods as "pure noise" segments (the spectra in these time periods are mainly composed of environmental noise and do not contain any useful physiological signals); within the time periods without signals, calculate the spectral characteristics of the noise; use the short - time Fourier transform (STFT) to extract the spectral information of the signal: Through the formula: t ∈ the segment without signal, where, is the noise spectrum estimate at time t and frequency f; X(f, t) is the result of the Fourier transform of the signal; N is the number of signal - free segments used for noise estimation;

[0052] Step 2: After estimating the noise spectrum, subtract the noise spectrum from the spectrum of the original signal; use the short - time Fourier transform (STFT) to decompose the noisy signal into spectra; the spectrum of the signal is represented as X(f, t), which contains the mixture of the target signal and noise; for each time - frequency point, subtract the estimated noise spectrum from the spectrum of the signal |X(f, t)|; to avoid negative spectrum values, a threshold also needs to be introduced to ensure that the subtracted noise does not cause the spectrum values to become unrealistic, obtaining the formula: where, is the spectrum of the denoised signal; the max function ensures that the spectrum values will not be negative, avoiding spectrum errors caused by subtracting noise; Note: Through this process, the noise spectrum part is removed, and only the spectrum of the target signal is retained;

[0053] Step 3: Convert the processed spectrum back to the time-domain signal through the inverse short-time Fourier transform (ISTFT), thereby obtaining the denoised audio signal. The formula is: where x c (t) is the denoised time-domain signal, and ISTFT is the inverse short-time Fourier transform operation. Note: Through this step, an audio signal with background noise removed is obtained, and the main components of the target signal are restored;

[0054] Step 4: After completing the spectral subtraction, reduce the high-frequency residual noise through smoothing processing (such as moving average or low-pass filtering) to maintain the smoothness of the signal; for the still-weak target signal after denoising, enhance the signal strength through an enhancement algorithm (such as gain control);

[0055] The feature processing module is used to perform feature processing on the denoised data, specifically:

[0056] Convert the processed microphone audio data and piezoelectric film audio data into the same file format for the same feature analysis;

[0057] Frame the synchronized audio data at a fixed duration (e.g., 25 milliseconds). Each frame contains N1 = sampling rate × 0.025 sampling points; set the frame overlap (e.g., 50% overlap) to smooth the signal conversion; then perform window function calculation on the data of each frame through the formula: Output the windowed frame data x i,ω (n1), and n1 = 1, 2, 3,..., N1 - 1; for each frame of audio data, use MFCC for feature extraction: pre-emphasize the signal to enhance the high-frequency components: x p (n1) = x i1,ω (n1) - α·x i1,ω (n1 - 1), where α is usually taken as approximately 0.95 to enhance the high-frequency components and compensate for the attenuation of high-frequency signals during acquisition; next, perform a fast Fourier transform (FFT) on the pre-emphasized signal to convert the time-domain signal to the frequency domain, obtaining the spectrum X i1 (k); based on this spectrum, calculate the power spectrum of each frame through the formula: P(k) = |X i1 (k)| 2 , which reflects the energy distribution of each frequency component; then, according to the non-linear characteristics of the human ear's sensitivity to different frequencies, construct a Mel filter bank to decompose the power spectrum into several Mel frequency bands and calculate the energy output E m of each filter, specifically: for the m-th filter in the Mel filter bank, let its response function be H m (k). Usually, H m(k) is triangular in shape and is defined within a certain frequency range (e.g., the filter starts at f m-1 , reaches a peak at f m , and ends at f m+1 ), and values outside this range are 0; for the m-th filter, its energy is calculated by weighted summation of the power spectrum within the frequency range covered by the filter response: where k s and k e are the starting and ending bins in the FFT of the frequency range corresponding to filter m respectively; H m (k) is the weight of the filter at frequency bin k, reflecting the contribution of this frequency component to the filter energy; to prevent the energy of some filters from being close to or equal to zero, a very small constant ∈ (such as 10 -10 ) is usually added to ensure that the value is always greater than zero, that is, E' m = E m + ε; further, the output energy E' m of each filter is logarithmically transformed using the natural logarithm or the common logarithm form: logE m = ln(E' m ); finally, the logarithmic energy sequence is input into the discrete cosine transform (DCT) to obtain a set of discrete coefficients, and the first 12 to 13 coefficients are selected as the MFCC feature vectors of this frame;

[0058] Since there may be non - linear offsets or length differences between two MFCC feature sequences on the time axis, it is difficult to directly compare frame - by - frame to reflect their true similarity; then the dynamic time warping (DTW) algorithm is used to perform the best alignment on the two sequences, and the specific process is as follows:

[0059] Let the MFCC feature vector of the i2 - th frame of the microphone be and the piezoelectric film feature vector of the j2 - th frame be . The distance between them is defined as the Euclidean distance of the two vectors: Among them, D1 is the dimension of the MFCC feature (for example, 12 or 13, and may be higher if differential features are added); let D(i2,j2) represent the minimum cumulative distance between the first i2 frames (microphone) and the first j2 frames (piezoelectric); the recursive update formula is used: D(i2,j2) = d(i2,j2) + min{D(i2 - 1,j2), D(i2,j2 - 1), D(i2 - 1,j2 - 1)}, and the boundary conditions are: starting point: D(1,1); the first row and the first column can be initialized by accumulation: D(i2,1) = d(i2,1) + min{D(i2 - 1,1), D(i2 - 1,1)}, D(1,j2) = d(1,j2) + min{D(1,j2), D(1,j2 - 1)}; through this process, the entire DTW matrix is constructed, where each element reflects the cumulative matching error of the two sequences under different frame alignments; for easy comparison, the cumulative distance is normalized; assuming the length of the optimal alignment path is L, the normalized error is: where N m and N p are the number of frames of the microphone and piezoelectric film sequences respectively; this value represents the overall matching error of the two sequences; in order to make the result more intuitive, the normalized error is converted into a similarity index between 0 and 1: Output the similarity XSD, where λ is a tuning parameter that controls the attenuation rate of the similarity; the closer the value is to 1, the more similar the two sequences are, and vice versa;

[0060] Compare the similarity XSD with a preset similarity threshold. If the similarity XSD is less than the preset similarity threshold, it is determined to be two audio features; on the contrary, if the similarity XSD is greater than or equal to the preset similarity threshold, it is determined to be the same audio feature;

[0061] The feature processing module is also used for audio data feature matching, including microphone audio data and piezoelectric film audio data; match the obtained data with the preset audio data features in the data storage unit (a large number of feature templates of preset audio data are pre-stored in the data storage unit, and these templates correspond to different physiological states or diagnostic criteria, such as standard heart sounds, normal breathing sounds, and possible abnormal states); similarly, the similarity XSD is calculated and recorded as the matching degree PPD. When the matching degree PPD reaches the preset threshold, it can be judged as "matching successful", and the corresponding physiological state or abnormal mode is confirmed; and the device status (such as normal, abnormal, connection status, etc.) is displayed through an indicator light, and the matching result is transmitted to other devices (such as smartphones, monitoring platforms) through a wireless module for further diagnosis.

[0062] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments only. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A digital stethoscope system based on a piezoelectric thin film, comprising: Intelligent wearable stethoscope, acquisition and analysis module, and feature processing module; characterized in that: The intelligent wearable stethoscope consists of a stethoscope watch (3), a chest strap (1), and a buckle (2); wherein, the stethoscope watch (3) includes an operation button (11), an operation button (12), a chest strap interface (13), an integration hole (14), a stethoscope battery compartment (15), and a piezoelectric film (16); the chest strap (1) is connected to the chest strap interface (13), the buckle (2) is installed on the chest strap (1) and is used to adjust the length of the chest strap; the chest strap interface (13), the integration hole (14), the operation button (11), the stethoscope battery compartment (15), and the piezoelectric film (16) are all installed on the stethoscope body (12); The acquisition and analysis module obtains the audio data of the intelligent wearable stethoscope through a wireless module and differentiates it into microphone audio data and piezoelectric film audio data; the microphone audio data is used for heart sound and lung sound analysis to identify heart diseases or breathing problems; the piezoelectric film audio data is used for low-frequency signal analysis to identify heart vibrations and pulse fluctuations; The microphone audio data is denoised using an adaptive filter, and the LMS algorithm is used to dynamically adjust the filter weights to optimize the noise removal effect; the piezoelectric film audio data is denoised using spectral subtraction, the noise spectrum is estimated through short-time Fourier transform, and subtracted from the signal; the inverse STFT is used to restore the time-domain signal, and smoothing processing and signal enhancement are performed to optimize the noise removal effect; The feature processing module performs unified format conversion and feature analysis on the denoised microphone and piezoelectric film audio data to obtain a similarity XSD; if XSD is lower than a preset threshold, it is determined as different audio features, otherwise it is regarded as the same audio; The feature processing module performs audio feature matching, compares the acquired data with a stored preset audio template, and calculates a matching degree PPD; when PPD reaches the threshold, it is determined as a successful match, and the corresponding physiological state or abnormal mode is output; the result is displayed through an indicator light.

2. The digital stethoscope system based on a piezoelectric thin film according to claim 1, wherein The specific process of differentiating into microphone audio data and piezoelectric film audio data is as follows: The data file format of the acquired audio data is differentiated, and matched with the format corresponding to the microphone audio data. If the two are consistent, the acquired audio data is microphone audio data; similarly, the file format corresponding to the piezoelectric film audio data is matched with the data file format of the acquired audio data. If the two are consistent, the acquired audio data is piezoelectric film audio data.

3. The digital stethoscope system based on a piezoelectric thin film according to claim 2, wherein The specific process of denoising the microphone audio data using an adaptive filter is as follows: The adaptive filter is retrieved for processing. The adaptive filter automatically adjusts the filter coefficients according to the change of the input signal; the adaptive filter uses the LMS algorithm for calculation, and substitutes the mixed audio signal x(n) containing the target signal and noise and the introduced noise reference signal d(n) into the adaptive filter to output the microphone filtered signal y(n); from the formula: Output the microphone-filtered signal y(n), where w k (n) are the weight coefficients of the filter, M is the order of the filter, and n is the current time; then calculate the error between the filter output y(n) and the actually collected signal x(n) through the formula: e(n) = x(n) - y(n), where the error e(n) represents the noise part that the filter fails to effectively remove; adjust the weight coefficient w k (n) of the filter; the LMS algorithm uses the gradient descent method to update the weights: w k (n + 1) = w k (n) + μ·e(n)·d(n - k), where μ is the step factor that controls the amplitude of each update; (n - k) is the delayed version of the reference signal.

4. The digital stethoscope system based on a piezoelectric thin film according to claim 3, characterized in that, The specific process of denoising the piezoelectric film audio data using spectral subtraction is as follows: Retrieve spectral subtraction for denoising: Step 1: There is no target information in the signal, and these time periods are regarded as pure noise segments; within the time periods without signal, calculate the spectral characteristics of the noise; use the short-time Fourier transform to extract the spectral information of the signal: through the formula: t belongs to the signal-free segment, is the noise spectrum estimation at time t and frequency f; X(f, t) is the Fourier transform result of the signal; N is the number of signal-free segments used for noise estimation; Step 2: After estimating the noise spectrum, subtract the noise spectrum from the spectrum of the original signal; use the short-time Fourier transform to decompose the noisy signal into spectra; the spectrum of the signal is represented as X(f,t), which contains the mixture of the target signal and the noise; For each time-frequency point, subtract the estimated noise spectrum from the spectrum of the signal |X(f,t)|; introduce a threshold to obtain the formula: is the signal spectrum after denoising; the max function ensures that the spectrum values are not negative; Step 3: Convert the processed spectrum back to the time-domain signal through the inverse short-time Fourier transform to obtain the denoised audio signal, by the formula: x c (t) is the time-domain signal after denoising, and ISTFT is the inverse short-time Fourier transform operation; Step 4: After completing the spectral subtraction, reduce the high-frequency residual noise through smoothing to maintain the smoothness of the signal; for the target signal that is still weak after denoising, enhance the signal strength through an enhancement algorithm.

5. The digital stethoscope system based on piezoelectric thin films according to claim 1, wherein The specific process of obtaining the similarity XSD is as follows: The audio data is framed at a fixed duration and smoothed using a window function; MFCC is used for feature extraction, including pre-emphasis, FFT transformation, power spectrum calculation, Mel filter bank processing, logarithmic transformation, and DCT transformation, and finally the MFCC feature vector is obtained; to solve the non-linear offset on the time axis, the DTW algorithm is used to perform the best alignment on the MFCC sequence, calculate the cumulative matching error, and normalize to obtain the similarity XSD.

6. The digital stethoscope system based on a piezoelectric thin film according to claim 5, wherein The specific process of framing the audio data at a fixed duration and smoothing it using a window function is as follows: The synchronized audio data is framed at a fixed duration, and each frame contains N1 = sampling rate × 0.025 sampling points; set the frame overlap to smooth the signal transition; then perform window function calculation on the data of each frame, through the formula: Output the windowed frame data x i,ω (n1).

7. The digital stethoscope system based on piezoelectric thin films according to claim 5, characterized in that, The specific process of performing feature extraction using MFCC is as follows: For each frame of audio data, MFCC is used for feature extraction: pre-emphasize the signal to enhance the high-frequency components: x p (n1) = x i1,ω (n1) - α·x i1,ω (n1 - 1), where α = 0.95; perform a fast Fourier transform on the pre-emphasized signal to convert the time-domain signal to the frequency domain, obtaining the spectrum X i1 (k); based on this spectrum, calculate the power spectrum of each frame. Through the formula: P(k) = |X i1 (k)| 2 ; then, according to the non-linear characteristics of the human ear's sensitivity to different frequencies, construct a Mel filter bank, decompose the power spectrum into several Mel frequency bands, and calculate the energy output E of each filter m , specifically: for the m-th filter in the Mel filter bank, let its response function be H m (k). Usually, the shape of H m (k) is triangular, defined within a certain frequency range, and the values outside this range are 0; for the m-th filter, calculate the energy by weighted summing the power spectrum within the frequency range covered by the filter's response: k s and k e are respectively the start and end bins in the FFT of the frequency range corresponding to filter m; H m (k) is the weight of the filter at frequency bin k. To prevent the energy of some filters from approaching or being equal to zero, add a very small constant ∈ to ensure that the value is always greater than zero, that is, E' m = E m + ε; use the natural logarithm or the common logarithm form to perform a logarithmic transformation on the energy E' output by each filter m , using the natural logarithm: logE m = ln(E' m ); finally, input the logarithmic energy sequence into the discrete cosine transform to obtain a set of discrete coefficients, and select the first 12 to 13 coefficients as the MFCC feature vector of this frame.

8. The digital stethoscope system based on a piezoelectric thin film according to claim 5, characterized in that The specific process of obtaining the similarity XSD is as follows: Use the dynamic time warping algorithm to perform the best alignment on two sequences. The specific process is as follows: Let the microphone MFCC feature vector of the i2-th frame be and the piezoelectric film feature vector of the j2-th frame be The distance between them is defined as the Euclidean distance of the two vectors: D1 is the dimension of the MFCC feature; Let \(D(i2,j2)\) denote the minimum cumulative distance between the first \(i2\) frames and the first \(j2\) frames; use the recursive update formula: \(D(i2,j2)=d(i2,j2)+\min\{D(i2 - 1,j2),D(i2,j2 - 1),D(i2 - 1,j2 - 1)\}\), with boundary conditions: starting point: \(D(1,1)\); the first row and the first column can be initialized by accumulation: \(D(i2,1)=d(i2,1)+\min\{D(i2 - 1,1),D(i2 - 1,1)\}\), \(D(1,j2)=d(1,j2)+\min\{D(1,j2),D(1,j2 - 1)\}\); construct the entire DTW matrix. Let the length of the optimal alignment path be \(L\), then the normalized error is: N m and \(N\) p are the number of frames of the microphone and piezoelectric film sequences respectively; \(D\) n represents the overall matching error between the two sequences; convert the normalized error into a similarity metric between 0 and 1: Output the similarity \(XSD\), where \(\lambda\) is a tuning parameter that controls the decay rate of the similarity.

Images