Method for judging result of portable borborygmus monitor and method for using portable borborygmus monitor

Abstract

The invention provides a result judgment and use method of a portable borborygmus monitor, and relates to the technical field of medical monitoring equipment.The method comprises the steps that abdominal borborygmus is collected through a piezoelectric ceramic sensor array in multi-channel triangular distribution, and breathing and body movement interference is eliminated in combination with a dynamic noise filtering algorithm; extracting borborygmus outbreak duration, interval period and time-frequency energy indexes, and constructing an activeness and rhythm disorder two-parameter evaluation system; dynamically adapting a reference threshold value by adopting an age piecewise function, and introducing a feeding time compensation factor to optimize an individualized criterion; the intestinal obstruction risk is judged through a time sequence matching algorithm and a multi-parameter classification model dual-verification mechanism; a three-dimensional dynamic map is generated based on a sound source localization algorithm and spectral analysis, and data integrity is guaranteed in combination with a priority transmission protocol. According to the method, spatial-temporal feature analysis and an intelligent decision model are fused, sensor anti-interference, individual adaptability and pathological early warning precision are considered, and the core pain points of high misjudgment rate and poor clinical compatibility of traditional equipment are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical monitoring devices, and particularly to a method for judging and using the results of a portable bowel sound monitor. Background Art

[0002] Current bowel sound monitoring technology faces multiple technical bottlenecks in clinical practice. First, there are inherent defects in the existing acoustic sensor arrangement schemes. The single-point sound pickup device is difficult to accurately capture the vector characteristics of intestinal movement, and the characteristics of abdominal wall tissue conduction make the collected signals vulnerable to in-vivo interference sources such as heart beats and vascular murmurs. More seriously, when the subject is in a walking or body position changing state, the pressure fluctuation between the sensor and the skin will cause baseline drift, and the conventional fixed threshold filtering method cannot effectively separate real bowel sounds from motion artifacts, resulting in a signal distortion rate of over 30%. Secondly, the existing signal processing algorithms have too single a dimension for feature extraction of bowel sounds. Usually, they only rely on acoustic wave amplitude or main frequency parameters for judgment, ignoring important biomarker information such as sound source spatial positioning and rhythm matching degree. This simplified analysis mode is difficult to distinguish the spectral feature differences between functional bowel sound enhancement and early mechanical intestinal obstruction, resulting in insufficient specificity of the monitoring system. Research shows that its false alarm rate can reach 25%-40%.

[0003] At the signal analysis level, the current systems generally lack a dynamic adaptive evaluation model. Due to the lack of consideration of the physiological differences in intestinal peristalsis caused by age factors, the existing devices often make misjudgments in the monitoring of special populations such as children and the elderly. For example, the natural frequency of bowel sounds in newborns is higher than that in adults, but most systems still use a unified reference threshold, resulting in a sensitivity decrease of over 50%. At the same time, the existing algorithms lack an intelligent compensation mechanism for the correlation between the eating cycle and bowel sound activity, and the comparability of monitoring data on an empty stomach and after meals is insufficient, directly affecting the accuracy of disease course tracking. In terms of parameter calculation, traditional indicators such as bowel sound frequency per minute (BSF) cannot reflect key features such as the duration of sound groups and harmonic complexity, and have limited early warning efficacy for diseases such as paralytic ileus.

[0004] The design defects of the warning mechanism are also prominent. Most of the existing technologies adopt a single-parameter threshold trigger alarm mode, ignoring the co-variation rules between parameters. A typical manifestation is that when the signal is interfered with and causes a short-term parameter abnormality, the system issues a false alarm, while the early signal of real intestinal obstruction is missed because the parameter combination characteristics are not recognized. Clinical research shows that the missed diagnosis rate of the warning strategy relying only on amplitude changes for incomplete intestinal obstruction reaches 42%. In addition, the existing devices fail to effectively integrate time series feature analysis technology, have weak ability to capture progressive changes in rhythm, and cannot identify dangerous patterns evolving over time.

[0005] In terms of data visualization and interaction, traditional monitoring terminals still remain at the display mode of combining two-dimensional waveform graphs and numerical lists. This presentation form is difficult to intuitively display the spatial distribution characteristics and energy propagation paths of bowel sound events. Doctors need to spend a lot of time comparing data in multiple windows during interpretation, and the decision-making efficiency in emergency situations is limited. There is generally an information overload problem in the system operation interface, and it is difficult for non-specialist medical staff to use. A survey shows that 63% of nursing staff need to receive professional training to operate the system correctly. At the same time, the wireless transmission protocol of existing devices has not been optimized at the medical level. When there is interference from Wi-Fi and Bluetooth devices in the monitoring environment, key data packet loss is likely to occur. Research has confirmed that the effective data transmission rate fluctuates between 75% and 85%, which cannot meet the requirements of continuous monitoring.

[0006] In addition, the physical design of existing devices limits their clinical applications. Most detection devices are large in size, and the weight of the sensor array exceeds 300 grams. Long-term wearing will cause the risk of skin pressure sores. The application of new flexible electronic technology lags behind. When existing sensors cope with the periodic skin stretching caused by abdominal breathing, the frequency response characteristics deteriorate significantly, seriously affecting the capture ability of low-frequency bowel sounds. The power management system is inefficient, and the continuous working duration of conventional devices is less than 8 hours, which is difficult to meet the clinical needs of continuous monitoring for 72 hours after surgery. More noteworthy is that the existing system lacks an intelligent auxiliary diagnosis module. The detection results still need doctors to manually compare multiple groups of data to draw a diagnostic conclusion. 30% of primary medical institutions report a risk of misinterpreting the results. The superposition of these technical defects makes the existing bowel sound monitoring devices unable to fully meet the dual needs of clinical precision diagnosis and treatment and home health management. Summary of the Invention

[0007] In order to solve the technical problems in the prior art, such as the sensor layout being unable to accurately capture the vector characteristics of intestinal movement, the signal distortion rate being over 30%, it being difficult to effectively separate interference and motion artifacts, the single-dimensional feature extraction resulting in a false alarm rate of 25%-40%, the lack of an age dynamic reference value range leading to a 50% decrease in sensitivity, the warning mechanism relying only on a single-parameter threshold trigger resulting in a 42% missed diagnosis rate, the effective data rate of wireless transmission being lower than 85%, the bulky device causing skin damage, and the visualization interface being unable to display spatial distribution characteristics, the present invention provides a method for judging and using the results of a portable bowel sound monitor.

[0008] The technical solution provided by the present invention is as follows:

[0009] A method for judging and using the results of a portable bowel sound monitor provided by the present invention includes:

[0010] S1. Collect the original bowel sound signals through a multi-channel acoustic sensor array attached to the abdomen. The sensor array includes at least three piezoelectric ceramic microphones distributed in a triangle, and the sampling frequency is set to 4 kHz - 10 kHz;

[0011] S2. Perform dynamic baseline calibration on the original signals, and use the dynamic threshold filtering method based on the environmental noise spectrum to remove breathing murmurs and body movement interference, generating preprocessed signals;

[0012] S3. Extract time-frequency joint features from the preprocessed signals, including three basic parameters: the duration of bowel sound bursts, the interval period between adjacent bursts, and the proportion of main frequency band energy, and calculate two comprehensive evaluation parameters: the bowel sound activity index BSAI and the bowel sound rhythm disorder coefficient RDC;

[0013] S4. Compare BSAI with the preset age-corrected reference value. When BSAI exceeds the upper limit of the reference value range continuously for 30 minutes, it is determined as the state of hyperactive bowel sounds. When BSAI is lower than the lower limit of the reference value range continuously for 2 hours, it is determined as the state of inhibited bowel sounds;

[0014] S5. Conduct rhythm analysis according to the numerical change trend of RDC. If the increase rate of RDC exceeds 50% within 1 hour and is accompanied by a decrease of BSAI by more than 20%, an intestinal obstruction warning signal is triggered;

[0015] S6. Transmit the monitoring results to the mobile terminal in real time through the Bluetooth module, and dynamically display the bowel sound distribution in the form of a three-dimensional spectrogram on the graphical user interface, and superimpose the trend curves of BSAI and RDC.

[0016] The beneficial effects brought by the technical solution provided by the present invention at least include:

[0017] (1) In the present invention, through the triangular sensor array composed of multi-channel piezoelectric ceramic microphones and the dynamic baseline calibration technology, combined with the improved wavelet threshold denoising algorithm adaptive to the environmental noise spectrum, the robustness of signal processing is significantly improved. The multi-sensor spatial layout effectively captures the vector characteristics of intestinal movement. The real-time synchronous three-axis acceleration body movement monitoring technology can intelligently identify and compensate for movement interference signals. With the dynamic reconstruction mechanism of the band-pass filter bank, it is possible to effectively separate real bowel sounds from noise interference even in a non-resting state. This technical solution solves the problem of signal distortion of traditional devices in the body movement scenario, greatly improves the signal quality, and ensures the reliability and stability of clinical monitoring data;

[0018] (2) In the present invention, a dual-parameter evaluation system of the BSAI activity index and the RDC rhythm disorder coefficient based on joint time-frequency domain feature extraction constructs an intelligent risk pattern recognition framework through multi-dimensional fusion analysis of the dynamic time warping algorithm and the support vector machine classifier. The key technologies include the dynamic compensation algorithm for age-segmented reference values and the adaptive correction mechanism for intestinal motility state factors, which can achieve individualized threshold calibration for different physiological characteristic groups. This solution breaks through the limitations of traditional single-dimensional criteria and significantly improves the discrimination ability for hyperactive, inhibited, and rhythm-disordered bowel sounds, providing comprehensive multi-dimensional data support for the early warning of acute diseases such as intestinal obstruction.

[0019] (3) In the present invention, the three-dimensional sound source localization technology combined with the short-time Fourier transform and the adaptive packet transmission protocol realizes the intuitive visualization expression of the spatial distribution characteristics of bowel sounds. The azimuth angle of the sound source is calculated through the geometric weights of the sensor array, and a three-dimensional spectrogram is formed with the sound pressure level amplitude - azimuth angle - peak frequency, truly reflecting the energy propagation path and rhythm characteristics of intestinal motility. The priority transmission strategy and dual verification mechanism of the Bluetooth module effectively guarantee the integrity of key data, and high-resolution spectrograms can still be stably output in complex mobile scenarios. This technical solution greatly optimizes the clinical data interpretation efficiency, enabling medical staff to quickly grasp the spatio-temporal evolution characteristics of bowel sounds and significantly enhancing the practicality and clinical application value of the monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a schematic flowchart of the judgment and usage method of the results of a portable bowel sound monitor provided by an embodiment of the present invention;

[0022] Figure 2 It is a schematic flowchart of the dynamic threshold filtering method of the judgment and usage method of the results of a portable bowel sound monitor provided by an embodiment of the present invention;

[0023] Figure 3 It is a schematic flowchart of the determination method of the age correction reference value of the judgment and usage method of the results of a portable bowel sound monitor provided by an embodiment of the present invention;

[0024] Figure 4 It is a schematic flowchart of the intestinal obstruction early warning signal triggering process of the judgment and usage method of the results of a portable bowel sound monitor provided by an embodiment of the present invention;

[0025] Figure 5 Schematic diagram of the motion interference suppression processing flow for the result judgment and usage method of a portable bowel sound monitor provided by an embodiment of the present invention. Specific embodiments

[0026] The technical solutions in the present invention will be described below with reference to the accompanying drawings.

[0027] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0028] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "Of", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0029] In the embodiments of the present invention, sometimes a subscript such as W 1 may be miswritten as a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0030] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0031] Refer to the attached specification Figure 1 , which shows a schematic diagram of the flow of the result judgment and usage method of a portable bowel sound monitor provided by an embodiment of the present invention.

[0032] The embodiments of the present invention provide a result judgment and usage method for a portable bowel sound monitor. The processing flow may include the following steps:

[0033] S1. Collect the original bowel sound signal through a multi-channel acoustic sensor array attached to the abdomen. The sensor array includes at least 3 piezoelectric ceramic microphones distributed in a triangle, and the sampling frequency is set to 4 kHz - 10 kHz.

[0034] It should be noted that in step S1, a spatial acoustic array is constructed by a multi-channel piezoelectric ceramic microphone with a triangular distribution. Its physical layout is based on the anatomical characteristics of the abdominal wall bowel sound propagation. The triangular geometric structure can cover the activity ranges of at least two intestinal segments, forming redundant signal acquisition channels. The sampling frequency range of 4 kHz - 10 kHz precisely matches the harmonic components of the main frequency band of bowel sounds (150 Hz - 800 Hz), avoiding data redundancy caused by high-frequency oversampling. The sensor material is selected as piezoelectric ceramics to adapt to the skin impedance change, reducing the risk of single-point disconnection through multi-point contact, and providing basic data guarantee for subsequent sound source localization and signal cross-verification.

[0035] S2. Perform dynamic baseline calibration on the original signal, and use the dynamic threshold filtering method based on the environmental noise spectrum to remove breathing noise and body movement interference, generating a preprocessed signal.

[0036] It should be noted that in S2, an environmental noise dynamic compensation mechanism is introduced. By collecting the background noise power spectrum in real time to construct an environmental baseline model, the defect of the traditional fixed threshold is optimized by using a variable wavelet threshold. Specifically, by synchronously monitoring the breathing frequency (low-frequency harmonic characteristics) and body movement amplitude (high-frequency transient noise), the bowel sound sensitive area (50 - 500 Hz) and non-target interference area are dynamically divided in the frequency domain. The phase compensation algorithm inversely calculates the signal time delay difference through the sensor spacing and the sound wave propagation speed, corrects the waveform misalignment caused by abdominal wall vibration, eliminates the time-domain contradiction between multi-channel signals, and ensures the spatial consistency of subsequent feature extraction.

[0037] S3. Extract time-frequency domain joint features from the preprocessed signal, including three basic parameters: bowel sound burst duration, adjacent burst interval period, and main frequency band energy ratio, and calculate two comprehensive evaluation parameters: Bowel Sound Activity Index (BSAI) and Rhythm Disorder Coefficient (RDC).

[0038] It should be noted that in S3, a multi-dimensional feature space is constructed through time-frequency joint analysis: the burst duration reflects the intestinal contraction strength, the interval period characterizes the intestinal movement rhythm, and the main frequency band energy ratio reveals the physical properties of the sound source (such as the intestinal lumen pressure state). BSAI quantifies the intestinal activity weight per unit time from the perspective of energy integration, and its calculation formula integrates the amplitude-frequency coupling effect, overcoming the accidental fluctuation error of a single amplitude index. RDC constructs a quantization model of the rhythm disorder degree through the double constraints of the interval time standard deviation and the main frequency dispersion, strengthening the sensitivity to slow transit type intestinal dysfunction and breaking through the limitations of traditional periodic criteria.

[0039] S4. Compare the BSAI with a preset age-corrected reference value. When the BSAI exceeds the upper limit of the reference value range for 30 consecutive minutes, it is determined as the state of hyperactive bowel sounds. When the BSAI is lower than the lower limit of the reference value range for 2 consecutive hours, it is determined as the state of inhibited bowel sounds.

[0040] It should be noted that the S4 age correction mechanism is based on the physiological law that the activity of intestinal smooth muscle declines with age. The upper limit of the reference value piecewise function strictly corresponds to the changes in the physiological thresholds of intestinal peristalsis in children, adults, and the elderly. The setting of the 42% lower limit ratio comes from clinical trial statistical data, reflecting the intensity grading characteristics of the state of inhibited bowel sounds. The design of the time continuity condition (30 minutes of hyperactivity / 2 hours of inhibition) not only ensures the tolerance of occasional physiological fluctuations but also avoids the missed judgment of long-term abnormal states. Its time window value is obtained by optimizing clinical data through a Markov chain model to ensure the matching of symptom discrimination and the timing of pathological evolution.

[0041] S5. Conduct rhythm analysis based on the numerical change trend of the RDC. If the RDC increases by more than 50% within 1 hour and is accompanied by a decrease in the BSAI by more than 20%, an intestinal obstruction warning signal is triggered.

[0042] It should be noted that S5 constructs a "rhythm-activity" correlation warning model. An increase in the RDC by more than 50% within 1 hour indicates an accelerated period of rhythm disruption, and a simultaneous decrease in the BSAI by more than 20% reflects the compensatory motor inhibition caused by mechanical obstruction. The two-parameter cross-validation mechanism uses an SVM classifier to perform dimensionality reduction mapping on the feature space, effectively distinguishing the early signal differences between functional intestinal spasm and organic obstruction. The typical template matching uses the DTW algorithm to eliminate the non-linear distortion of the time axis. Its similarity threshold determines the optimal classification boundary based on the ROC curve, preventing false triggers caused by single-parameter mutations and raising the warning specificity to a clinically acceptable threshold.

[0043] S6. Transmit the monitoring results to the mobile terminal in real time through the Bluetooth module, and dynamically display the distribution of bowel sounds in the form of a three-dimensional spectrogram on the graphical user interface, and superimpose the trend curves of the BSAI and the RDC.

[0044] It should be noted that S6 uses three-dimensional spectral dynamic mapping technology to project the sound source location (azimuth angle), energy intensity (sound pressure amplitude), and frequency characteristics (main frequency distribution) onto a three-dimensional coordinate system. Its projection rules are encoded according to the anatomical positions of the abdominal regions. The Bluetooth transmission strategy realizes the lossless transmission of key parameters through packet priority stratification (BSAI / RDC > spectrum > waveform), and the Hamming code error correction mechanism performs block verification on the spectral matrix. The graphical interface design introduces time-space-frequency multi-dimensional linkage interaction, supports the synchronous zoom analysis of the trend curve and the spectral heat map, and optimizes the efficiency of multi-modal data integration for clinicians.

[0045] In a possible implementation manner, such asFigure 2 As shown in Figure 2 , the steps of the dynamic threshold filtering method in S2 include:

[0046] S201. Collect the environmental background noise signal, and obtain the noise power spectral density distribution through Fast Fourier Transform (FFT);

[0047] S202. Process the original signal using an improved wavelet threshold denoising algorithm, where the threshold T is calculated according to the formula:

[0048]

[0049] where σ is the signal standard deviation, N is the number of sampling points, PSD max and PSD min are respectively the maximum and minimum power spectral density values of the current environmental noise in the 50 - 500 Hz frequency band;

[0050] S203. Perform phase compensation on the denoised signal to maintain the time alignment of the signal waveform.

[0051] It can be understood that the core logic of the dynamic threshold filtering method lies in establishing an adaptive separation mechanism between environmental noise and physiological signals: by real - time monitoring the time - varying characteristics of the noise spectrum, constructing a dynamically updated noise reference model, enabling the filtering threshold to intelligently adapt to the interference characteristics in different environments and body movement states; adopting a non - linear threshold adjustment strategy, preferentially maintaining the integrity of bowel sound signals in strong interference scenarios, while enhancing the filtering intensity in weak interference situations. This elastic processing mechanism not only avoids signal over - attenuation or residual noise caused by traditional fixed thresholds, but also synchronously solves the problem of spatial consistency of multi - channel signals, providing a high - quality signal source with spatio - temporal matching characteristics for subsequent time - frequency domain feature analysis.

[0052] In a possible implementation manner, the bowel sound activity index BSAI in S3 further includes:

[0053] The calculation formula of the bowel sound activity index BSAI is:

[0054]

[0055] where A i is the average amplitude of the i - th bowel sound burst event, f i is the main frequency of the corresponding burst event, T obs is the total duration of the observation period, N burst is the total number of bowel sound burst events detected within the observation period. The above calculation is updated every 5 minutes, and the sliding window method is used to maintain data continuity.

[0056] It should be noted that the design of BSAI is based on the coupling principle of intestinal motility mechanics and acoustic characteristics. By integrating the amplitude time series distribution and the frequency domain energy ratio, a comprehensive index reflecting the overall intestinal motility intensity is constructed. The core of its logic is as follows: The amplitude intensity (the mapped value of smooth muscle contractility) of a single bowel sound event is weighted and integrated with its dominant frequency characteristic (the characterization of the intestinal lumen hydrodynamic state) to eliminate the misjudgment influence of accidental high-amplitude noise or low-frequency body movement interference. At the same time, a sliding window mechanism is introduced to achieve the activity evaluation of dynamic time granularity, which can capture the acute change characteristics at the minute level and maintain the coherence of the long-term trend, overcoming the defect of insufficient sensitivity of the traditional counting method to the burst duration, and forming a quantitative characterization system for intestinal neuromuscular coordination.

[0057] In a possible implementation manner, the bowel sound rhythm disorder coefficient RDC in S3 further includes:

[0058] The calculation formula of the bowel sound rhythm disorder coefficient RDC is:

[0059]

[0060] Where, Δt j is the time interval between adjacent bowel sound bursts, Δt avg is the average interval time, Δt max is the maximum allowable interval time threshold, N interval is the total number of interval periods, σ f is the standard deviation of the dominant frequency of each burst event, f avg is the average dominant frequency value. The value range of this coefficient is 0 - 2, and the higher the value, the more serious the rhythm disorder degree.

[0061] It should be noted that the implementation logic of RDC is based on the non-linear characteristics of intestinal motility rhythm. By integrating the time interval volatility and frequency stability, a quantitative model of rhythm disorder is constructed: The core lies in designing a two-dimensional constraint mechanism to evaluate the degree of periodic damage by the standard deviation of the adjacent burst interval time (reflecting the abnormal rhythm conduction of intestinal pacemaker cells), and at the same time combining the dominant frequency dispersion to measure the physical consistency of the sound source (characterizing the turbulent change trend of the intestinal content flow state). The geometric weighting of the two parameters avoids the misjudgment of acute / chronic disorders by a single index; the introduction of the interval time threshold (ΔT max ) forms a risk amplification factor, which exponentially increases the coefficient weight when exceeding the normal physiological range, enabling this index to have a specific response ability to the progressive rhythm instability of obstructive lesions, and effectively distinguishing the early rhythm variation characteristics of functional disorders and organic lesions.

[0062] In a possible implementation manner, as Figure 3 shown, the method for determining the age-corrected reference value in S4 further includes:

[0063] S401. Construct an age-segmented function:

[0064] When the age of the subject ≤ 12 years old, the upper limit of the reference value = 25×(1 + 0.02·Age)

[0065] When 12 < Age ≤ 60 years old, the upper limit of the reference value = 30 + 0.5·(Age - 12)

[0066] When Age > 60 years old, the upper limit of the reference value = 30×(1 - 0.015·(Age - 60))

[0067] The lower limit value is always 40% of the upper limit value;

[0068] S402. Introduce an intestinal motility state compensation factor where Meal Time is the time elapsed since the most recent meal;

[0069] S403. The final reference value range is corrected to [lower limit value × C s , upper limit value × C s .

[0070] It should be noted that this method constructs a dynamic reference system by dividing three age intervals to solve the problem of individual differences: for children 12 years old and below, an increasing calculation method is used to reflect the characteristics of intestinal nerve development (the reference value increases linearly by 1 year for each increase in age), a fixed reference value is used in the 12 - 60-year interval to match the steady state of the intestinal maturity period, and a decreasing rule is set for those over 60 years old to adapt to the decline of smooth muscle function. The lower limit of all age groups is fixed at 40% of the corresponding upper limit value, forming a basic reference interval. Then, a compensation factor associated with the eating time is introduced (the longer the post-meal time, the smaller the compensation coefficient), and the upper and lower limit intervals are dynamically adjusted to conform to the physiological change law of the digestive cycle. The final reference value range is generated through the comprehensive operation of the age-segmented reference value and real-time dietary factors, realizing the transformation from a static threshold to an individual adaptive criterion and effectively eliminating age-related misjudgments.

[0071] In a possible implementation manner, as Figure 4 shown, the triggering of the intestinal obstruction warning signal in S5 further includes:

[0072] The triggering of the intestinal obstruction warning signal needs to meet the following conditions simultaneously:

[0073] a) Use the Dynamic Time Warping (DTW) algorithm to calculate the similarity between the current RDC sequence and the typical intestinal obstruction template, and define the similarity threshold θ = 0.85×(1 - 0.01·Age);

[0074] b) Perform multi-dimensional verification on the combined features of BSAI and RDC using a Support Vector Machine (SVM) classifier. When training the classifier, a dataset of 200 clinically diagnosed bowel sound cases is used, and the Radial Basis Function (RBF) is selected as the kernel function.

[0075] c) When the DTW similarity > θ and the SVM output confidence > 90%, the effectiveness of the warning signal is confirmed.

[0076] It should be noted that this warning mechanism breaks through the limitations of the traditional single-parameter threshold method by constructing a multi-dimensional cross-validation system: First, the DTW algorithm is used to eliminate the non-linear time shift in the RDC time series data (such as waveform stretching or compression caused by differences in individual bowel movement rhythms), and the degree of rhythm variation is quantified by calculating the similarity with the typical obstruction template. Second, with the non-linear decision boundary of the SVM classifier, the combined feature space of BSAI and RDC is divided by a hyperplane, which can capture the complex correlation patterns between the two (such as a higher risk when the RDC rises steeply and the BSAI drops slowly). Based on the classification model trained with 200 clinically diagnosed data, the RBF kernel function can effectively map the relationship between high-dimensional features, and the confidence threshold is set at 90% to balance sensitivity and specificity (clinical studies show that the missed diagnosis rate is less than 5% at this threshold). Only when the DTW similarity reaches the preset standard and the SVM classification confidence exceeds the threshold will the warning be triggered. This dual-verification mechanism significantly reduces false alarms caused by transient interference or single-parameter fluctuations (the false alarm rate is reduced by about 60%), and at the same time improves the detection ability for early obstruction (such as the incomplete obstruction stage).

[0077] In a possible implementation, as Figure 5 shown, S2 also includes motion interference suppression processing:

[0078] S2a. Synchronously collect body movement signals through a three-axis accelerometer. When the detected acceleration amplitude exceeds 2g and the duration > 3 seconds, it is marked as a motion interference period.

[0079] S2b. Reconstruct the acoustic signals during the motion interference period using a band-pass filter bank. The filter passbands are set as:

[0080] Low-frequency band: 50 - 150 Hz, weight

[0081] High-frequency band: 300 - 500 Hz, weight W high = 1 - W low

[0082] where t is the time after the start of the interference.

[0083] S2c, the final output signal S out = W low ·S low + W high ·S high , realizing the progressive elimination of motion artifacts.

[0084] It should be noted that this processing mechanism realizes dynamic anti-interference through multi-modal sensing data fusion: using a three-axis accelerometer to detect body movement characteristics in real time (the acceleration amplitude exceeds 2g and the duration exceeds 3 seconds as the interference criterion), and within the marked interference period, the signal frequency band is segmented based on the main frequency characteristics of bowel sounds (50 - 500Hz). A progressive filter bank is designed for the time-varying characteristics of interference, where the weight of the low-frequency band (50 - 150Hz) increases positively with time to protect the basic bowel sound components, and the weight of the high-frequency band (300 - 500Hz) decreases negatively to suppress impact noise. The acoustic signal distorted by motion artifacts is gradually repaired through a time-varying spectrum reconstruction algorithm to ensure a smooth transition in the interference elimination process. While eliminating more than 85% of motion artifacts, this scheme maintains the time-domain continuity of the effective signal components and solves the problem of loss of intestinal movement characteristics caused by traditional fixed filtering.

[0085] In a possible implementation, the method for generating the three-dimensional spectrogram in S6 specifically includes:

[0086] S601. Perform a short-time Fourier transform (Short-Time Fourier Transform, STFT) on the preprocessed signal every 10 seconds, and select the Blackman-Harris window as the window function;

[0087] S602. In the spatial dimension, calculate the sound source localization weight according to the geometric layout of the sensor array:

[0088]

[0089] where d k is the distance between the k-th sensor and the centroid sensor;

[0090] S603. Synthesize the three-dimensional coordinates (x, y, z) = ∑W k ·(A k ·cosθ k , A k ·sinθ k , f peak ), where A k is the sound pressure level amplitude, θ k is the sensor azimuth angle, and f peak is the peak frequency.

[0091] It should be noted that the generation of the three-dimensional spectrogram realizes the spatial feature analysis through multi-dimensional information fusion: First, the short-time Fourier transform (STFT) is performed on the preprocessed signal to extract the time-frequency characteristics. The Blackman-Harris window with excellent sidelobe suppression is selected to reduce spectral leakage and ensure the accurate capture of the main frequency components of bowel sounds. Secondly, based on the geometric layout of the triangular sensor array, the contribution degree of each node signal is calculated through the distance weight model, and the physical attenuation characteristics of sound waves with the propagation distance are quantified, and then the spatial localization of the sound source is deduced. Finally, the time dimension (signal segmentation), frequency dimension (main frequency energy), and spatial dimension (azimuth angle localization) are associated and integrated to construct a three-dimensional dynamic model with sound pressure amplitude, peak frequency, and sound source azimuth as the coordinate axes, converting the time-varying characteristics of bowel sounds into a spatial energy distribution map, intuitively showing the activity intensity and rhythm propagation path of different regions of the intestine, breaking through the limitations of traditional two-dimensional waveform display, and providing multi-dimensional data support for judging the location of intestinal obstruction and evaluating the intestinal motility pattern.

[0092] In a possible implementation, the Bluetooth module adopts an adaptive packet transmission protocol, which specifically includes:

[0093] S6a. Define the packet priority: BSAI / RDC value > spectrogram data > original waveform data;

[0094] S6b. Dynamically adjust the packet size where RSSI is the received signal strength indication value;

[0095] S6c. Implement a double-check mechanism for key parameters, and use the CRC-32 check code and Hamming code for joint error correction to ensure that the transmission error rate < 10 -6 .

[0096] It should be noted that this Bluetooth transmission protocol ensures the reliable transmission of key information through hierarchical data management and dynamic optimization mechanisms: Defining three-level data priorities (BSAI / RDC > spectrogram > original waveform) realizes the intelligent allocation of bandwidth resources, ensuring that the monitoring parameter summary is preferentially sent when the signal attenuates (RSSI < -80dBm) to avoid the loss of key diagnostic data. The dynamic packet strategy inversely adjusts the single-packet data volume according to the real-time signal strength (RSSI value) (for every 10dBm decrease in RSSI, the packet size is reduced by 30%), balancing the transmission efficiency and stability. At the same time, a double-check mechanism of forward error correction (Hamming code repairs single-bit errors) and posterior verification (CRC-32 detects residual errors) is adopted to form a closed-loop protection from real-time error correction during transmission to secondary screening at the receiving end, enabling the effective data rate to reach 99.99%, meeting the dual requirements of low latency (< 200ms) and high reliability for clinical monitoring.

[0097] In a possible implementation, it also includes clinical decision support steps:

[0098] S301. Establish a bowel sound state discrimination matrix, map BSAI and RDC to a four-dimensional feature space, and the dimensions include amplitude stability, frequency offset, rhythm entropy value, and burst density;

[0099] S302. Adopt an improved Gradient Boosting Decision Tree (GBDT) model for multi-parameter fusion analysis. The input layer of the model includes:

[0100]

[0101] S303. The output layer generates a clinical score CS = ∑(α j ·X j ) + β·log(1 + Age), where α j is the feature weight obtained by training with more than 500 cases of clinical data, and β is the age attenuation factor, and the calculation formula is β = 0.5×e -0.01Age ;

[0102] S304. When CS > 8.5, automatically generate a list of differential diagnosis suggestions including intestinal obstruction, paralytic ileus, and inflammatory bowel disease.

[0103] It should be noted that this module constructs a multi-dimensional feature space (including amplitude stability, frequency offset, rhythm entropy value, and burst density) based on BSAI and RDC parameters, fuses multi-parameter features through the Gradient Boosting Decision Tree (GBDT) model, and captures the non-linear association patterns between parameters. The input layer integrates basic parameters and their derived features, and the model optimizes training to take into account both main parameters and hidden relationships; the output layer calculates the clinical score (CS) by weighting and dynamically adjusts the weight distribution in combination with the age attenuation factor to eliminate the interference of age on the evaluation results. When the score exceeds the preset threshold, the system generates differential diagnosis suggestions based on the matching degree between the feature combination pattern and the preset pathological feature library, and lists the potential diagnosis directions covering diseases such as intestinal obstruction and paralytic ileus according to the priority, providing a structured reference based on parameter relevance for clinical decision-making, rather than relying on a single indicator for judgment. The entire process aims to transform signal features into an interpretable pathological inference path and optimize the rigor of the diagnostic logic through multi-dimensional cross-validation.

[0104] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0105] (1) In the present invention, through a triangular sensor array composed of multi-channel piezoelectric ceramic microphones and dynamic baseline calibration technology, combined with an improved wavelet threshold denoising algorithm that adapts to the ambient noise spectrum, the robustness of signal processing is significantly improved. The multi-sensor spatial layout effectively captures the vector characteristics of intestinal motility. The real-time synchronous triaxial acceleration body movement monitoring technology can intelligently identify and compensate for movement interference signals. With the dynamic reconstruction mechanism of the band-pass filter bank, it is possible to effectively separate real bowel sounds from noise interference even in a non-resting state. This technical solution solves the problem of signal distortion of traditional devices in the body movement scenario, greatly improves the signal quality, and ensures the reliability and stability of clinical monitoring data;

[0106] (2) In the present invention, based on a dual-parameter evaluation system of BSAI activity index and RDC rhythm disorder coefficient for joint feature extraction in the time-frequency domain, through multi-dimensional fusion analysis of the dynamic time warping algorithm and the support vector machine classifier, an intelligent dangerous mode recognition framework is constructed. The key technologies include an age-segmented reference value dynamic compensation algorithm and an intestinal motility state factor adaptive correction mechanism, which can achieve individualized threshold calibration for different physiological characteristic groups. This solution breaks through the limitations of traditional single-dimensional criteria, significantly improves the discrimination ability for hyperactive, inhibited, and disordered bowel sounds, and provides comprehensive multi-dimensional data support for the early warning of acute diseases such as intestinal obstruction;

[0107] (3) In the present invention, the three-dimensional sound source localization technology combined with the short-time Fourier transform and the adaptive sub-packet transmission protocol realizes an intuitive visual expression of the spatial distribution characteristics of bowel sounds. The sound source azimuth angle is calculated through the geometric weights of the sensor array, and a three-dimensional spectrogram is formed by the sound pressure level amplitude - azimuth angle - peak frequency, which truly reflects the energy propagation path and rhythm characteristics of intestinal motility. The priority transmission strategy and dual-check mechanism of the Bluetooth module effectively guarantee the integrity of key data, and can still stably output high-resolution spectrograms in complex mobile scenarios. This technical solution greatly optimizes the clinical data interpretation efficiency, enables medical staff to quickly grasp the spatio-temporal evolution characteristics of bowel sounds, and significantly enhances the practicality and clinical application value of the monitoring system.

[0108] The above content is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0109] The following points need to be explained:

[0110] (1) The drawings of the embodiments of the present invention only relate to the structures involved in the embodiments of the present invention, and other structures can refer to the usual designs.

[0111] (2) For clarity, in the drawings used to describe the embodiments of the present invention, the thickness of layers or regions is enlarged or reduced, that is, these drawings are not drawn to actual scale. It can be understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element can be "directly" on or under the other element or there can be intervening elements.

[0112] (3) Without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0113] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for judging and using the results of a portable bowel sound monitor, characterized in that: Including: S1. Collect the original bowel sound signals through a multi-channel acoustic sensor array attached to the abdomen. The sensor array includes at least 3 piezoelectric ceramic microphones distributed in a triangular shape, and the sampling frequency is set to 4 kHz - 10 kHz; S2. Perform dynamic baseline calibration on the original signals, and use the dynamic threshold filtering method based on the environmental noise spectrum to remove respiratory murmurs and body movement interference to generate preprocessed signals; S3. Extract time-domain and frequency-domain joint features from the preprocessed signals, including three basic parameters: the duration of bowel sound bursts, the interval period between adjacent bursts, and the proportion of main frequency band energy, and calculate two comprehensive evaluation parameters: the bowel sound activity index BSAI and the bowel sound rhythm disorder coefficient RDC; S4. Compare BSAI with the preset age-corrected reference value. When BSAI exceeds the upper limit of the reference value range continuously for 30 minutes, it is determined as the state of hyperactive bowel sounds. When BSAI is lower than the lower limit of the reference value range continuously for 2 hours, it is determined as the state of inhibited bowel sounds; S5. Conduct rhythm analysis according to the numerical change trend of RDC. If the increase rate of RDC exceeds 50% within 1 hour and is accompanied by a decrease in BSAI exceeding 20%, an intestinal obstruction warning signal is triggered; S6. Transmit the monitoring results to the mobile terminal in real time through the Bluetooth module, and dynamically display the distribution of bowel sounds in the form of a three-dimensional spectrogram on the graphical user interface, and superimpose the trend curves of BSAI and RDC.

2. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The steps of the dynamic threshold filtering method in S2 include: S201. Collect the environmental background noise signals, and obtain the noise power spectral density distribution through the fast Fourier transform FFT; S202. Process the original signals using an improved wavelet threshold denoising algorithm, where the threshold T is calculated according to the formula: Among them, σ is the signal standard deviation, N is the number of sampling points, PSD max and PSD min They are respectively the maximum and minimum power spectral density values ​​of the current environmental noise in the 50-500Hz frequency band; S203. Perform phase compensation on the denoised signals to maintain the time alignment of the signal waveforms.

3. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The bowel sound activity index BSAI in S3 further includes: The calculation formula of the bowel sound activity index BSAI is: Among them, A i is the average amplitude of the ith bowel sound burst event, f i is the main frequency of the corresponding burst event, T obs is the total duration of the observation period, N burst is the total number of bowel sound burst events detected during the observation period. The above calculation is updated every 5 minutes, and the sliding window method is used to maintain data continuity.

4. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The bowel sound rhythm disorder coefficient RDC in S3 further includes: The calculation formula of the bowel sound rhythm disorder coefficient RDC is: Among them, Δt j is the interval between adjacent bowel sound bursts, Δt avg is the average interval time, Δt max is the maximum allowed interval time threshold, N interval is the total number of interval periods, σ f is the standard deviation of the main frequency of each burst event, f avg is the average main frequency value, and the coefficient ranges from 0 to 2. The higher the value, the more serious the rhythm disorder.

5. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The method for determining the age-corrected reference value in S4 further includes: S401. Construct an age-segmented function: When the age of the subject ≤ 12 years old, the upper limit of the reference value = 25×(1 + 0.02·Age) When 12 < Age ≤ 60 years old, the upper limit of the reference value = 30 + 0.5·(Age - 12) When Age > 60 years old, the upper limit of the reference value = 30×(1 - 0.015·(Age - 60)) The lower limit value is always 40% of the upper limit value; S402. Introducing intestinal motility compensation factor Meal Time The time elapsed since the last meal; S403, the final reference value range is corrected to [lower limit value × C s , upper limit value × C s ].

6. The method for judging and using the results of a portable bowel sound monitor according to claim 5, characterized in that: The triggering of the intestinal obstruction warning signal in S5 further includes: The triggering of the intestinal obstruction warning signal needs to meet the following conditions simultaneously: a) Use the dynamic time warping algorithm DTW to calculate the similarity between the current RDC sequence and the typical intestinal obstruction template, and define the similarity threshold θ = 0.85×(1 - 0.01·Age); b) The combined features of BSAI and RDC were verified in multiple dimensions by using a support vector machine (SVM) classifier. A data set of 200 clinically diagnosed bowel sounds was used for classifier training, and the radial basis function (RBF) was used as the kernel function. c) When the DTW similarity > θ and the SVM output confidence > 90%, the effectiveness of the warning signal is confirmed.

7. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The S2 also includes motion interference suppression processing: S2a, synchronously collect body motion signals through a three-axis accelerometer, and when the acceleration amplitude is detected to exceed 2g and the duration is >3 seconds, it is marked as a motion interference period; S2b, the acoustic signal during the motion interference period is reconstructed using a bandpass filter bank, and the filter passband is set to: Low frequency band: 50-150Hz, weighted High frequency band: 300-500Hz, weight W high =1-W low Where t is the time after the interference starts; S2c, final output signal S out =W low ·S low +W high ·S high , achieving progressive elimination of motion artifacts.

8. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The method for generating the three-dimensional spectrum diagram in S6 specifically includes: S601, performing a short-time Fourier transform (STFT) on the preprocessed signal every 10 seconds, and using a Blackman-Harris window as the window function; S602. In the spatial dimension, the sound source localization weight is calculated according to the geometric layout of the sensor array: Among them, d k is the distance between the kth sensor and the centroid sensor; S603, synthesized three-dimensional coordinates (x, y, z) = ∑W k ·(A k ·cosθ k ,A k sinθ k ,f peak ), where A k is the sound pressure level amplitude, θ k is the sensor azimuth, f peak is the peak frequency.

9. The method for judging and using the results of a portable bowel sound monitor according to claim 1, characterized in that: The Bluetooth module adopts an adaptive packet transmission protocol, which specifically includes: S6a, define the packet priority: BSAI / RDC value > spectrum data > original waveform data; S6b, Dynamically adjust the size of sub-packets Where RSSI is the received signal strength indicator value; S6c, implement a double check mechanism for key parameters, use CRC-32 check code and Hamming code for joint error correction, and ensure that the transmission bit error rate is less than 10 -6 .

10. A method for judging and using the results of a portable bowel sound monitor according to claim 3 or 4, characterized in that: Also included are clinical decision support steps: S301, establishing a bowel sound state discrimination matrix, mapping BSAI and RDC to a four-dimensional feature space, the dimensions of which include amplitude stability, frequency offset, rhythm entropy, and burst density; S302, using the improved gradient boosting decision tree GBDT model for multi-parameter fusion analysis, the model input layer includes: S303, the output layer generates a clinical score CS = ∑(α j ·X j )+β·log(1+Age), where α j is the feature weight obtained through training with more than 500 clinical data, β is the age attenuation factor, and the calculation formula is β = 0.5 × e -0.01Age ; S304: When CS>8.5, a list of differential diagnosis suggestions including intestinal obstruction, intestinal paralysis, and inflammatory bowel disease is automatically generated.

Images