Borborygmus characteristic parameter real-time extraction method based on intelligent algorithm

Abstract

The invention provides a borborygmus characteristic parameter real-time extraction method based on an intelligent algorithm, and the method comprises the following steps: signal collection: collecting a borborygmus signal through a high-sensitivity recording device; signal preprocessing: preprocessing the collected borborygmus signal, including noise suppression and signal filtering, so as to reduce interference of environmental noise; and feature extraction: performing spectral analysis by adopting fast Fourier transform and wavelet transform, extracting frequency domain and time domain feature parameters of the borborygmus signal, and calculating statistical features of the signal, including mean value, variance, kurtosis and skewness, and frequency spectrum entropy and power spectrum density based on the frequency spectrum.

Description

Technical Field

[0001] The present invention relates to the field of real-time extraction of bowel sound characteristic parameters, and in particular to a real-time extraction method of bowel sound characteristic parameters based on an intelligent algorithm. Background Art

[0002] With the development of biomedical engineering and signal processing technology, more and more attention has been paid to the monitoring of intestinal functional status. Bowel sounds, as physiological parameters reflecting intestinal peristalsis, are increasingly widely used in clinical diagnosis and treatment. Abnormal changes in bowel sounds are often associated with a variety of digestive system diseases, such as intestinal obstruction and gastrointestinal dysfunction. Therefore, accurate and rapid extraction of bowel sound characteristic parameters is of great significance for the early detection, diagnosis and treatment of diseases.

[0003] Bowel sound detection usually involves the acquisition, processing and analysis of sound signals. However, due to the nonlinearity, non-stationarity and complexity of bowel sound signals, as well as the interference of external environmental noise, traditional bowel sound feature parameter extraction methods face many challenges. In addition, the real-time and automated requirements for bowel sound monitoring in clinical practice are also increasing, which puts higher requirements on the existing bowel sound feature parameter extraction technology.

[0004] The existing bowel sound feature parameter extraction technologies mainly include the following solutions:

[0005] Extraction methods based on traditional signal processing: This method usually includes preprocessing (such as denoising, filtering), feature extraction (such as frequency domain analysis, time domain analysis) and parameter estimation. For example, Fourier transform is used to analyze the spectral characteristics of bowel sound signals, or wavelet transform is used for time-frequency analysis. These methods can extract the characteristic parameters of bowel sounds to a certain extent, but are limited by noise interference and signal non-stationarity, and their accuracy and stability need to be improved.

[0006] Extraction methods based on machine learning: In recent years, the application of machine learning technology in the extraction of bowel sound feature parameters has gradually increased. For example, support vector machines (SVM), random forests (RF) or deep learning models (such as convolutional neural networks (CNNs)) are used to classify or regress bowel sound signals. These methods have improved the accuracy of feature extraction to a certain extent, but they require a large amount of labeled data for training, and the generalization ability of the model is limited. Summary of the invention

[0007] In view of the deficiencies in the prior art, the present invention provides a real-time extraction method for bowel sound characteristic parameters based on an intelligent algorithm. By adopting advanced signal processing technology and adaptive filtering algorithm, noise interference is effectively suppressed, thereby improving the accuracy of bowel sound characteristic parameter extraction and ensuring the reliability of clinical diagnosis.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] A method for real-time extraction of bowel sound characteristic parameters based on an intelligent algorithm comprises the following steps:

[0010] Signal collection: Bowel sound signals are collected through high-sensitivity recording equipment;

[0011] Signal preprocessing: preprocess the collected bowel sound signals, including noise suppression and signal filtering, to reduce the interference of environmental noise;

[0012] Feature extraction: Fast Fourier transform and wavelet transform are used for spectrum analysis to extract the frequency domain and time domain characteristic parameters of bowel sound signals, and calculate the statistical characteristics of the signals, including mean, variance, kurtosis, skewness, as well as spectrum entropy and power spectrum density based on the spectrum;

[0013] Feature selection and optimization: Use intelligent algorithms, including random forests and deep neural networks, to optimize and select the extracted feature parameters;

[0014] Model training: Based on the optimized feature parameters, the classification model is trained and used to monitor the changes in bowel sounds in real time.

[0015] Preferably, the high-sensitivity recording device is a microphone with a wide frequency response, which is suitable for accurately capturing low-frequency bowel sound signals.

[0016] Preferably, the noise suppression adopts a denoising algorithm based on wavelet transform to effectively remove environmental noise in the signal through multi-scale analysis.

[0017] Preferably, the signal filtering includes bandpass filtering, and the frequency range of the filter is set to 0-150 Hz to retain the main frequency components of bowel sounds.

[0018] Preferably, the intelligent algorithm includes: using correlation analysis to analyze the correlation between the extracted features and remove highly correlated redundant features; using a random forest algorithm to select the extracted feature parameters; using a genetic algorithm to optimize the feature parameters; so as to remove redundant features and optimize the performance of the classification model.

[0019] Preferably, the classification model is a deep neural network or a support vector machine, which can achieve accurate classification of bowel sound features by training on a large amount of labeled data.

[0020] Preferably, the real-time monitoring includes inputting the real-time collected bowel sound signals into a trained classification model and outputting the classification results of the bowel sound characteristic parameters.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] 1. The present invention aims to improve the accuracy of bowel sound feature parameter extraction, thereby providing more reliable diagnostic information;

[0023] 2. The present invention strives to achieve rapid extraction of bowel sound characteristic parameters by optimizing algorithms and data processing procedures to meet the needs of clinical real-time monitoring;

[0024] 3. The present invention reduces manual intervention and improves the automation and efficiency of bowel sound feature parameter extraction by introducing intelligent algorithms and automated processes;

[0025] 4. The present invention takes individual differences into consideration and provides a personalized feature parameter extraction scheme to improve adaptability to different populations and disease states;

[0026] 5. The present invention aims to reduce the overall cost of bowel sound feature parameter extraction by optimizing the design and using cost-effective components, making it easier to popularize and apply.

[0027] 6. The technical solution of the present invention has good scalability, can adapt to future technological development, and integrate more advanced acoustic analysis and machine learning algorithms BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1 It is a schematic diagram of the overall framework flow of the present invention. DETAILED DESCRIPTION

[0030] The present invention is described in further detail below with reference to the accompanying drawings.

[0031] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be used for other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0032] Those skilled in the art should understand that, in the disclosure of the present invention, the directions or positions indicated by the terms "longitudinal", "lateral", "up", "down", "left", "right", "front", "back", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc. are based on the directions or positional relationships shown in the accompanying drawings, which are only simplified descriptions for the convenience of describing the present invention, and do not indicate or imply that the device or component referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, the above terms should not be understood as limitations on the present invention.

[0033] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0034] Example:

[0035] See also Figure 1 The present invention provides a method for real-time extraction of bowel sound characteristic parameters based on an intelligent algorithm, comprising the following steps:

[0036] S1. Signal acquisition: Bowel sound signals are collected through high-sensitivity recording equipment;

[0037] S2. Signal preprocessing: preprocessing the collected bowel sound signals, including noise suppression and signal filtering, to reduce the interference of environmental noise;

[0038] S3. Feature extraction: Fast Fourier transform and wavelet transform are used for spectrum analysis to extract the frequency domain and time domain characteristic parameters of bowel sound signals, and calculate the statistical characteristics of the signals, including mean, variance, kurtosis, skewness, as well as spectrum entropy and power spectrum density based on the spectrum;

[0039] S4, Feature selection and optimization: Use intelligent algorithms, including random forests and deep neural networks, to optimize and select the extracted feature parameters;

[0040] S5. Model training: Based on the optimized feature parameters, the classification model is trained and used to monitor the changes in bowel sounds in real time.

[0041] In the present invention, high-quality, high-sensitivity recording equipment is used to collect bowel sound signals, including Levitt LCT440PURE condenser microphone or West Saint Mike wireless microphone and other equipment that can capture the most subtle sound details.

[0042] 1) Selection of auscultation position: Auscultation of bowel sounds is mainly performed in the abdomen, usually starting from the right lower abdomen around the umbilicus, followed by the area parallel to the umbilicus and close to the waist, the area below the umbilicus, and the area parallel to the area below the umbilicus and close to the waist1.

[0043] 2) Microphone placement: Fix the microphone to the stethoscope head to ensure that the microphone can accurately capture bowel sounds. Avoid placing the microphone in a position that may cause interference, such as in the radiation area of ​​the speaker2.

[0044] 3) Adjustment and testing: After fixing the microphone, make necessary adjustments to ensure that the microphone can clearly capture bowel sounds. The position and effect of the microphone can be verified by test recording.

[0045] 4) Signal acquisition: Place the stethoscope head on the patient's abdomen, starting from the right lower abdomen around the umbilicus, followed by the area parallel to the umbilicus and close to the waist, the area below the umbilicus, and the area parallel to the area below the umbilicus and close to the waist, to capture bowel sounds in the patient's abdomen;

[0046] In the present invention, the collected bowel sounds are preprocessed by the following steps:

[0047] The captured analog signal is converted into a digital signal. When the captured bowel sound signal is subjected to noise removal, a denoising algorithm based on wavelet transform is used. The denoising algorithm based on wavelet transform is more suitable for noise removal in non-stationary signals such as bowel sounds, including multi-scale analysis using db4 (Daubechies 4) wavelet basis, multi-scale wavelet decomposition of the bowel sound signal, and obtaining detail coefficients (high frequency part) and approximate coefficients (low frequency part) at different scales (frequencies);

[0048] Signal filtering includes bandpass filtering, and the filter frequency range is set to 0-150 Hz to retain the main frequency components of bowel sounds while suppressing or removing noise and unnecessary interference signals in other frequency ranges, thereby improving the quality of the signal;

[0049] Then, wavelet transform and bandpass filtering are implemented in MATLAB or Python to preprocess the signal, and then the preprocessed signal is used for feature extraction.

[0050] The preprocessed signal in the present invention is subjected to signal feature extraction by the following steps:

[0051] Performing analysis and preprocessing on the extracted signals, using fast Fourier transform and wavelet transform to analyze the preprocessed signals;

[0052] Extract frequency domain and time domain characteristic parameters of bowel sound signals, and calculate the statistical characteristics of the signals, including mean, variance, kurtosis, skewness, as well as spectrum entropy and power spectrum density based on spectrum;

[0053] Use existing signal processing libraries, including scipy, to implement spectrum analysis and statistical feature calculations.

[0054] Use intelligent algorithms to select and optimize the extracted signal features:

[0055] 1. Correlation analysis:

[0056] Data preparation: Prepare the preprocessed bowel sound feature parameter dataset and construct the feature matrix X and target variable y;

[0057] Calculate the correlation coefficient: Use methods such as Pearson correlation coefficient, Spearman correlation coefficient, or Keldobalier coefficient to calculate the correlation coefficient between each feature and the target variable.

[0058]

[0059] Feature selection: Based on the absolute value of the correlation coefficient, features that are highly correlated with the target variable are selected for subsequent analysis and modeling, and highly correlated redundant features are removed.

[0060] 2. Feature selection algorithm: Use random forest or other machine learning algorithms to select the features that contribute most to classification. The steps are:

[0061] Build multiple decision trees: Use the training data set to build multiple decision trees, each of which is generated by random sampling;

[0062] Feature importance evaluation: When building each tree, evaluate the contribution of each feature to reducing impurity (such as Gini impurity or entropy) when splitting the node;

[0063] Calculate feature importance: Average the results of all trees to get a comprehensive importance score for each feature;

[0064] Feature selection: Based on the feature importance score, select the features that contribute most to model prediction for model optimization.

[0065] 3. Genetic algorithm optimization: Use genetic algorithm to optimize feature parameters to improve classification performance. The steps are as follows:

[0066] Initialize the population: Generate a set of random feature subsets as the initial population, each subset represents a feature combination;

[0067] Fitness evaluation: Use classifiers including support vector machines or decision trees to evaluate the fitness of each feature subset through the classifier. The fitness function can be based on the prediction accuracy or AUC value of the model.

[0068] Selection: Select a subset of traits with higher fitness from the population as parents for the next generation;

[0069] Crossover and mutation: Perform a crossover operation on the selected parent feature subsets to generate new feature subsets; at the same time, mutate some subsets to randomly change the features in them;

[0070] Iterative optimization: Repeat the process of fitness evaluation, selection, crossover and mutation until the algorithm converges to the optimal feature subset.

[0071] 4. Technical implementation: Combine machine learning libraries including scikit-learn and genetic algorithm libraries including deap to achieve feature selection and optimization. The steps are as follows:

[0072] Data preparation and preprocessing: Prepare the required data sets and perform necessary preprocessing, such as missing value processing, normalization, etc.

[0073] Define the evaluation function for feature subsets: Use scikit-learn models to evaluate the performance of different feature subsets;

[0074] Genetic algorithm for feature selection: Use the DEAP library to define a genetic algorithm to optimize feature selection;

[0075] Validate the results: Retrain the model using the best feature set obtained and evaluate its performance on the test set or validation set.

[0076] Through training on a large amount of labeled data, accurate classification of bowel sound features is achieved:

[0077] Classification model selection: Choose support vector machine or deep neural network when using;

[0078] Real-time classification: Input the features of the real-time collected bowel sound signals into the trained model and output the classification results;

[0079] Technical implementation: Use deep learning frameworks including TensorFlow or PyTorch to implement model training and classification.

[0080] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. A method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm, characterized in that: The following steps are involved: Signal collection: Bowel sound signals are collected through high-sensitivity recording equipment; Signal preprocessing: preprocess the collected bowel sound signals, including noise suppression and signal filtering, to reduce the interference of environmental noise; Feature extraction: Fast Fourier transform and wavelet transform are used for spectrum analysis to extract the frequency domain and time domain characteristic parameters of bowel sound signals, and calculate the statistical characteristics of the signals, including mean, variance, kurtosis, skewness, as well as spectrum entropy and power spectrum density based on the spectrum; Feature selection and optimization: Use intelligent algorithms, including random forests and deep neural networks, to optimize and select the extracted feature parameters; Model training: Based on the optimized feature parameters, the classification model is trained and used to monitor the changes in bowel sounds in real time.

2. The method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm according to claim 1, characterized in that: The high-sensitivity recording device is a microphone with a wide frequency response, which is suitable for accurately capturing low-frequency bowel sound signals.

3. The method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm according to claim 1, characterized in that: The noise suppression adopts a denoising algorithm based on wavelet transform, and effectively removes environmental noise in the signal through multi-scale analysis.

4. The method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm according to claim 1, characterized in that: The signal filtering includes bandpass filtering, and the frequency range of the filter is set to 0-150 Hz to retain the main frequency components of bowel sounds.

5. The method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm according to claim 1, characterized in that: The intelligent algorithm includes: using correlation analysis to analyze the correlation between extracted features and remove highly correlated redundant features; using a random forest algorithm to select extracted feature parameters; using a genetic algorithm to optimize feature parameters; so as to remove redundant features and optimize the performance of the classification model.

6. The method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm according to claim 1, characterized in that: The classification model is a deep neural network or a support vector machine, which can achieve accurate classification of bowel sound features by training a large amount of labeled data.

7. The method for real-time extraction of bowel sound characteristic parameters based on intelligent algorithm according to claim 1, characterized in that: The real-time monitoring includes inputting the real-time collected bowel sound signals into a trained classification model and outputting the classification results of the bowel sound characteristic parameters.

Images