A method and system for identifying risk of dysphagia based on doppler acoustic features

Abstract

The application discloses a kind of dysphagia risk identification method and system based on Doppler acoustic characteristics, for medical signal processing, the method comprises: collecting laryngeal region reflection signal;Extracting laryngeal movement coordination feature, muscle strength efficiency feature, resting period abnormal micro-motion feature and swallowing mode stability feature;The standardization processing is carried out to each feature extracted and is combined with corresponding preset weight coefficient, and the Doppler acoustic swallowing stability index is obtained;The Doppler acoustic swallowing stability index value of each swallowing event is obtained and is gradually fitted, and the individual baseline stability index reference interval is constructed;Doppler acoustic swallowing stability index is compared with preset group risk threshold and individual baseline stability index reference interval, and the risk level of swallowing event is determined according to comparison result.The present application has individualization adaptive capacity, and can meet the demand of long-term continuous monitoring in daily scene.

Description

Technical Field

[0001] This invention relates to medical signal processing, and more particularly to a method and system for identifying swallowing disorder risk based on Doppler acoustic features. Background Technology

[0002] Dysphagia refers to functional impairments in the process of transporting food from the mouth to the stomach due to nerve, muscle, or structural abnormalities. It is commonly seen in patients with stroke, Parkinson's disease, head and neck tumors, and degenerative diseases in the elderly. Mild cases may affect eating and nutrient intake, while severe cases may lead to aspiration pneumonia due to accidental aspiration, seriously threatening the patient's quality of life.

[0003] Currently, clinical scales rely on subjective judgment and lack accuracy; imaging examinations are radiation- or invasive and difficult to track over a long period; effective assessments can continuously capture subtle changes in swallowing function in daily life, enabling early warning and objective quantification, providing a basis for clinical intervention decisions, and avoiding serious complications caused by missed diagnoses or delayed interventions.

[0004] Existing swallowing disorder assessment technologies either rely on video capture, which raises concerns about privacy and sensitivity to light and posture, or require the attachment of electrodes and the collection of biological samples, leading to insufficient invasiveness and compliance. Furthermore, most of these technologies only capture surface movements or the quality of a single swallow, lacking means to assess the internal laryngeal exertion, abnormal micro-movements during resting periods, and the consistency of swallowing patterns across multiple swallowing events.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To overcome the above problems, this invention aims to propose a method and system for identifying swallowing disorders based on Doppler acoustic features. The purpose is to solve the problems of existing swallowing disorder assessment technologies that rely on video acquisition, which leads to privacy leaks and sensitivity to light and posture, or that require attaching electrodes and collecting biological samples, resulting in insufficient invasiveness and compliance.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows: A first aspect of the present invention provides a method for identifying swallowing disorder risk based on Doppler acoustic features, the method comprising: The laryngeal region reflection signal containing phase and amplitude information was acquired; adaptive bandpass filtering was performed on the laryngeal region reflection signal, and laryngeal motor coordination features, muscle efficacy features, resting period abnormal micro-movement features, and swallowing pattern stability features were extracted from the filtered signal. The extracted features are standardized and combined with the corresponding preset weight coefficients to obtain the Doppler acoustic swallowing stability index of the current swallowing event. The Doppler acoustic swallowing stability index values ​​of each swallowing event were obtained and stepwise fitted to construct an individual baseline stability index reference range. The Doppler acoustic swallowing stability index of the current swallowing event is compared with the preset group risk threshold and the reference range of the individual baseline stability index. The risk level of the swallowing event is determined based on the comparison results, and a differentiated message push strategy is implemented based on the risk level.

[0008] Alternatively, the method for adaptive bandpass filtering of the throat region reflection signal is as follows: Real-time spectral analysis of the phase information in the throat region reflection signal is performed to determine the frequency band distribution of the low-frequency displacement component; Real-time spectral analysis of the amplitude information in the throat region reflection signal is performed to determine the frequency band distribution of high-frequency vibration components and acoustic event components; The cutoff frequency of the first bandpass filter is dynamically adjusted based on the frequency band distribution of the low-frequency displacement component, the cutoff frequency of the second bandpass filter is dynamically adjusted based on the frequency band distribution of the high-frequency vibration component, and the cutoff frequency of the third bandpass filter is dynamically adjusted based on the frequency band distribution of the acoustic event component. The phase information is filtered using the adjusted first bandpass filter to separate the low-frequency displacement component; The amplitude information is filtered by the adjusted second bandpass filter to separate the high-frequency vibration component; The amplitude information is filtered by the adjusted third bandpass filter to separate the acoustic event components.

[0009] Optionally, features of intralarynx motor coordination, muscle efficacy, resting-phase abnormal micromovements, and swallowing pattern stability are extracted, including: The peak value of the pharyngeal initiation acceleration and the time asymmetry ratio of laryngeal elevation and fall were calculated from the low-frequency displacement components to obtain the characteristics of intralarynx movement coordination. The power spectrum integral and spectral centroid frequency of the high-frequency vibration components within the swallowing cycle are calculated from the high-frequency vibration components to obtain the muscle efficacy characteristics. Multi-scale sample entropy was calculated from the low-frequency displacement component and high-frequency vibration component in the signal segment during the non-swallowing period to obtain the abnormal micro-motion characteristics during the resting period. The similarity of the time spectrum diagrams of adjacent swallowing events is calculated from the time spectrum diagrams of low-frequency displacement components, high-frequency vibration components, and acoustic event components of multiple swallowing events, thus obtaining the swallowing pattern stability characteristics.

[0010] Optionally, the method for obtaining the characteristics of intralarynx motor coordination is as follows: The rate of change of displacement is calculated frame by frame for the low-frequency displacement component, and then the rate of change of displacement is calculated to obtain the acceleration signal of the throat motion. The first acceleration peak corresponding to the starting point of the swallowing action in the acceleration signal is detected, and the amplitude of this peak is extracted as the peak value of the pharyngeal initiation acceleration. Identify the start and end times of the throat lifting phase from the low-frequency displacement components, and calculate the duration of the lifting. Identify the start and end times of the throat fall-off phase from the low-frequency displacement components and calculate the fall-off duration. Divide the duration of the rise by the duration of the fall to obtain the time imbalance of the rise and fall of the throat. By combining the peak acceleration at the pharyngeal initiation phase with the time imbalance of larynx elevation and descent, the characteristics of intralarynx movement coordination are obtained.

[0011] Alternatively, the method for obtaining muscle strength performance characteristics is as follows: The high-frequency vibration components are segmented into swallowing cycles to determine the start and end times of the swallowing action, and high-frequency vibration signal segments within the swallowing cycle are extracted. The power spectral density of the extracted high-frequency vibration signal segment is calculated, and the power spectral density is integrated in the frequency domain to obtain the power spectral integral of the high-frequency vibration component within the swallowing cycle. Spectral analysis was performed on the extracted high-frequency vibration signal segments to calculate their centroid frequencies. By combining the power spectrum integral with the centroid frequency of the spectrum, the muscle strength performance characteristics are obtained. The expression for the calculation process is as follows: ;

[0012] In the formula, Represents the power spectrum integral; This indicates the upper frequency limit of the high-frequency vibration component; This indicates the lower limit of the frequency of the high-frequency vibration component; Indicates the frequency of high-frequency vibration signals. Power spectral density at; Represents a tiny increment in frequency; Indicates the centroid frequency of the spectrum; Indicates the characteristic value of muscle strength performance; Represents the power spectrum integral The corresponding preset weighting coefficients; Indicates the centroid frequency of the spectrum The corresponding preset weighting coefficients.

[0013] Optionally, the method for obtaining the abnormal micro-motion characteristics during the resting period is as follows: The signal segments that are not swallowed are extracted from the low-frequency displacement components, and the signal segments that are not swallowed are extracted from the high-frequency vibration components. By splicing low-frequency displacement signal segments with high-frequency vibration signal segments, a resting signal during the non-swallowing period is obtained. Multi-scale coarse-graining was performed on the non-swallowing resting signal to obtain coarse-grained sequences at multiple time scales; Calculate the sample entropy for the coarse-grained sequence at each time scale to obtain the sample entropy value corresponding to each scale; The average of the sample entropy values ​​at each scale is taken to obtain the multi-scale sample entropy, which is used as the resting period abnormal micro-motion characteristic. The expression for its calculation process is as follows: ;

[0014] In the formula, Indicates the first Coarse-grained sequence at the i-th time scale The value of each point; This indicates the resting signal during the non-swallowing period at the first... The amplitude of each sampling point; This represents the total number of sampling points for the resting signal during the non-swallowing period. Indicates the time scale factor; Indicates the maximum time scale; Indicates the first The length of coarse-grained sequences at each time scale; Indicates the embedding dimension; Indicates similarity tolerance; Indicates based on In the embedding dimension is The probability that two subsequences are similar when +1; Indicates based on In the embedding dimension is The probability that two subsequences are similar; Indicates the first Based on time scales Calculated sample entropy value; This represents the characteristic value of abnormal micro-movements during the resting period.

[0015] Optionally, the expression for obtaining the Doppler acoustic swallowing stability index of the current swallowing event is: ;

[0016] In the formula, The Doppler acoustic swallowing stability index represents the current swallowing event; The preset weighting coefficients represent the characteristics of intralarynx motor coordination. This indicates the peak acceleration at the start of the pharyngeal phase; This indicates the unevenness in the timing of the rise and fall of the larynx. The standard deviation of the intralarynx motor coordination characteristics in a healthy population sample; This represents the mean of intralarynx motor coordination characteristics in a healthy population sample; Represents the preset weighting coefficients corresponding to muscle strength performance characteristics; Indicates the characteristic value of muscle strength performance; This represents the mean of muscle strength efficacy characteristics in a healthy population sample; This represents the standard deviation of muscle strength efficacy characteristics in a healthy population sample. This represents the preset weighting coefficient corresponding to the abnormal micro-motion characteristics during the resting period; This represents the characteristic value of abnormal micro-movements during the resting period; This represents the mean value of abnormal micro-motion characteristics during the resting period in a healthy population sample; This represents the standard deviation of abnormal micromotion characteristics during the resting period in a healthy population sample. The preset weighting coefficients represent the stability characteristics of swallowing patterns; The temporal spectrum similarity of adjacent swallowing events; This represents the mean of swallowing pattern stability characteristics in a healthy population sample. This represents the standard deviation of swallowing pattern stability characteristics in a healthy population sample.

[0017] Optionally, the method for constructing the reference interval for individual baseline stability index is as follows: Obtain the Doppler acoustic swallowing stability index values ​​of each swallowing event of the current user and form a historical index sequence; Statistical analysis is performed on the historical index sequence to calculate the mean and standard deviation of the historical index sequence, and the mean and standard deviation are updated each time a new Doppler acoustic swallowing stability index value is obtained for a swallowing event. Based on the updated mean and standard deviation, a reference interval for the individual baseline stability index is constructed.

[0018] A second aspect of the present invention provides a swallowing disorder risk identification system based on Doppler acoustic features, the system comprising: The feature extraction module is used to acquire laryngeal region reflection signals containing phase and amplitude information; adaptive bandpass filtering is performed on the laryngeal region reflection signals, and laryngeal motor coordination features, muscle efficacy features, resting period abnormal micro-movement features, and swallowing pattern stability features are extracted from the filtered signals. The swallowing stability index generation module is used to standardize the extracted features and combine them with the corresponding preset weight coefficients to obtain the Doppler acoustic swallowing stability index of the current swallowing event. The individual baseline fitting module is used to obtain the Doppler acoustic swallowing stability index values ​​of each swallowing event and perform stepwise fitting to construct the individual baseline stability index reference range. The risk assessment and push module is used to compare the Doppler acoustic swallowing stability index of the current swallowing event with the preset group risk threshold and the individual baseline stability index reference range. Based on the comparison results, the risk level of the swallowing event is determined, and a differentiated message push strategy is executed based on the risk level.

[0019] Compared with existing technologies, this application has the following advantages: This application takes Doppler acoustic sensing as its core, and picks up the phase and amplitude information of the reflected signal in the laryngeal region in a non-contact manner. After adaptive bandpass filtering, three types of components are separated: low-frequency displacement, high-frequency vibration and acoustic events. Four features are extracted from these components: laryngeal movement coordination, muscle efficacy, resting period abnormal micro-movements and swallowing pattern stability. These features are then fused into a Doppler acoustic swallowing stability index. At the same time, a dual-reference risk assessment is performed by combining group thresholds and individual baseline reference intervals. Under the premise of protecting privacy, this application achieves comprehensive quantification of laryngeal internal movement, muscle exertion state and central control mode, and has individualized adaptive capabilities, which can meet the needs of long-term continuous monitoring in daily scenarios. Attached Figure Description

[0020] The above-mentioned features, characteristics, and advantages of the present invention, as well as their implementation methods, will become clearer and more readily understood in conjunction with the following description of the embodiments, which are illustrated in detail with reference to the accompanying drawings. Schematic diagrams are shown here: Figure 1 This is a flowchart of the swallowing disorder risk identification method based on Doppler acoustic features of the present invention; Figure 2 This is an extracted image of the Doppler acoustic features in this invention; Figure 3 This is a group risk level classification diagram in this invention; Figure 4 This is a comparison chart of the individual baseline and the current index in this invention; Figure 5 This is the comprehensive risk assessment diagram in this invention; Figure 6 This is a diagram illustrating the risk level-based differentiated message push strategy for the swallowing disorder risk identification method in this invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0022] This embodiment provides a method for identifying swallowing disorder risk based on Doppler acoustic features. While protecting privacy, it achieves comprehensive quantification of internal laryngeal movement, muscle exertion, and central control patterns, and possesses individualized adaptive capabilities, meeting the needs of long-term continuous monitoring in everyday scenarios. Figures 1-6 As shown, the method includes: The laryngeal region reflection signal containing phase and amplitude information was acquired; adaptive bandpass filtering was performed on the laryngeal region reflection signal, and laryngeal motor coordination features, muscle efficacy features, resting period abnormal micro-movement features, and swallowing pattern stability features were extracted from the filtered signal.

[0023] Preferably, the method for adaptive bandpass filtering of the throat region reflection signal is as follows: Real-time spectral analysis of the phase information in the throat region reflection signal is performed to determine the frequency band distribution of the low-frequency displacement component; Real-time spectral analysis of the amplitude information in the throat region reflection signal is performed to determine the frequency band distribution of high-frequency vibration components and acoustic event components; The cutoff frequency of the first bandpass filter is dynamically adjusted based on the frequency band distribution of the low-frequency displacement component, the cutoff frequency of the second bandpass filter is dynamically adjusted based on the frequency band distribution of the high-frequency vibration component, and the cutoff frequency of the third bandpass filter is dynamically adjusted based on the frequency band distribution of the acoustic event component. The phase information is filtered using the adjusted first bandpass filter to separate the low-frequency displacement component; The amplitude information is filtered by the adjusted second bandpass filter to separate the high-frequency vibration component; The amplitude information is filtered by the adjusted third bandpass filter to separate the acoustic event components.

[0024] Preferably, the following features are extracted: laryngeal motor coordination characteristics, muscle strength efficiency characteristics, resting-phase abnormal micromovement characteristics, and swallowing pattern stability characteristics, including: The peak value of the pharyngeal initiation acceleration and the time asymmetry ratio of laryngeal elevation and fall were calculated from the low-frequency displacement components to obtain the characteristics of intralarynx movement coordination. The power spectrum integral and spectral centroid frequency of the high-frequency vibration components within the swallowing cycle are calculated from the high-frequency vibration components to obtain the muscle efficacy characteristics. Multi-scale sample entropy was calculated from the low-frequency displacement component and high-frequency vibration component in the signal segment during the non-swallowing period to obtain the abnormal micro-motion characteristics during the resting period. The similarity of the time spectrum diagrams of adjacent swallowing events is calculated from the time spectrum diagrams of low-frequency displacement components, high-frequency vibration components, and acoustic event components of multiple swallowing events, thus obtaining the swallowing pattern stability characteristics.

[0025] Preferably, the method for obtaining the characteristics of intralarynx motor coordination is as follows: The rate of change of displacement is calculated frame by frame for the low-frequency displacement component, and then the rate of change of displacement is calculated to obtain the acceleration signal of the throat motion. The first acceleration peak corresponding to the starting point of the swallowing action in the acceleration signal is detected, and the amplitude of this peak is extracted as the peak value of the pharyngeal initiation acceleration. Identify the start and end times of the throat lifting phase from the low-frequency displacement components, and calculate the duration of the lifting. Identify the start and end times of the throat fall-off phase from the low-frequency displacement components and calculate the fall-off duration. Divide the duration of the rise by the duration of the fall to obtain the time imbalance of the rise and fall of the throat. By combining the peak acceleration at the pharyngeal initiation phase with the time imbalance of larynx elevation and descent, the characteristics of intralarynx movement coordination are obtained.

[0026] Preferably, the method for obtaining muscle strength performance characteristics is as follows: The high-frequency vibration components are segmented into swallowing cycles to determine the start and end times of the swallowing action, and high-frequency vibration signal segments within the swallowing cycle are extracted. The power spectral density of the extracted high-frequency vibration signal segment is calculated, and the power spectral density is integrated in the frequency domain to obtain the power spectral integral of the high-frequency vibration component within the swallowing cycle. Spectral analysis was performed on the extracted high-frequency vibration signal segments to calculate their centroid frequencies. By combining the power spectrum integral with the centroid frequency of the spectrum, the muscle strength performance characteristics are obtained. The expression for the calculation process is as follows: ;

[0027] In the formula, Represents the power spectrum integral; This indicates the upper frequency limit of the high-frequency vibration component; This indicates the lower limit of the frequency of the high-frequency vibration component; Indicates the frequency of high-frequency vibration signals. Power spectral density at; Represents a tiny increment in frequency; Indicates the centroid frequency of the spectrum; Indicates the characteristic value of muscle strength performance; Represents the power spectrum integral The corresponding preset weighting coefficients; Indicates the centroid frequency of the spectrum The corresponding preset weighting coefficients.

[0028] Preferably, the method for obtaining abnormal micro-motion characteristics during the resting period is as follows: The signal segments that are not swallowed are extracted from the low-frequency displacement components, and the signal segments that are not swallowed are extracted from the high-frequency vibration components. By splicing low-frequency displacement signal segments with high-frequency vibration signal segments, a resting signal during the non-swallowing period is obtained. Multi-scale coarse-graining was performed on the non-swallowing resting signal to obtain coarse-grained sequences at multiple time scales; Calculate the sample entropy for the coarse-grained sequence at each time scale to obtain the sample entropy value corresponding to each scale; The average of the sample entropy values ​​at each scale is taken to obtain the multi-scale sample entropy, which is used as the resting period abnormal micro-motion characteristic. The expression for its calculation process is as follows: ;

[0029] In the formula, Indicates the first Coarse-grained sequence at the i-th time scale The value of each point; This indicates the resting signal during the non-swallowing period at the first... The amplitude of each sampling point; This represents the total number of sampling points for the resting signal during the non-swallowing period. Indicates the time scale factor; Indicates the maximum time scale; Indicates the first The length of coarse-grained sequences at each time scale; Indicates the embedding dimension; Indicates similarity tolerance; Indicates based on In the embedding dimension is The probability that two subsequences are similar when +1; Indicates based on In the embedding dimension is The probability that two subsequences are similar; Indicates the first Based on time scales Calculated sample entropy value; This represents the characteristic value of abnormal micro-movements during the resting period.

[0030] It should be explained that the process of combining the peak pharyngeal initiation acceleration with the time imbalance of larynx rise and fall is as follows: the two indicators are treated as two sub-indicators of the intralarynx movement coordination characteristics, and are retained in the form of two-dimensional vectors without being merged into a single value; in the subsequent standardization process, these two sub-indicators are standardized separately, and then combined with other features into a weighted summation.

[0031] The non-swallowing period refers to the time window between two swallowing events; by detecting the start and end times of the swallowing action (which has been located in the extraction of intralarynx motor coordination features), the signal segment between adjacent swallowing events is the non-swallowing period signal segment; The definition of the non-swallowing period is as follows: The non-swallowing period refers to the time window between the end of the current swallowing event and the start of the next swallowing event; the start and end times of the swallowing event have been determined by the swallowing cycle segmentation of the low-frequency displacement component, and the signal segment between two adjacent swallowing events is the non-swallowing period signal segment; if there is continuous swallowing, a minimum interval threshold can be set, and the interval less than the threshold is considered as a continuation of a swallowing event.

[0032] The extracted features are standardized and combined with the corresponding preset weight coefficients to obtain the Doppler acoustic swallowing stability index of the current swallowing event.

[0033] Preferably, the expression for obtaining the Doppler acoustic swallowing stability index of the current swallowing event is as follows: ;

[0034] In the formula, The Doppler acoustic swallowing stability index represents the current swallowing event; The preset weighting coefficients represent the characteristics of intralarynx motor coordination. This indicates the peak acceleration at the start of the pharyngeal phase; This indicates the unevenness in the timing of the rise and fall of the larynx. The standard deviation of the intralarynx motor coordination characteristics in a healthy population sample; This represents the mean of intralarynx motor coordination characteristics in a healthy population sample; Represents the preset weighting coefficients corresponding to muscle strength performance characteristics; Indicates the characteristic value of muscle strength performance; This represents the mean of muscle strength efficacy characteristics in a healthy population sample; This represents the standard deviation of muscle strength efficacy characteristics in a healthy population sample. This represents the preset weighting coefficient corresponding to the abnormal micro-motion characteristics during the resting period; This represents the characteristic value of abnormal micro-movements during the resting period; This represents the mean value of abnormal micro-motion characteristics during the resting period in a healthy population sample; This represents the standard deviation of abnormal micromotion characteristics during the resting period in a healthy population sample. The preset weighting coefficients represent the stability characteristics of swallowing patterns; The temporal spectrum similarity of adjacent swallowing events; This represents the mean of swallowing pattern stability characteristics in a healthy population sample. This represents the standard deviation of swallowing pattern stability characteristics in a healthy population sample.

[0035] The Doppler acoustic swallowing stability index values ​​of each swallowing event were obtained and stepwise fitted to construct an individual baseline stability index reference range.

[0036] Preferably, the method for constructing the reference interval for the individual baseline stability index is as follows: Obtain the Doppler acoustic swallowing stability index values ​​of each swallowing event of the current user and form a historical index sequence; Statistical analysis is performed on the historical index sequence to calculate the mean and standard deviation of the historical index sequence, and the mean and standard deviation are updated each time a new Doppler acoustic swallowing stability index value is obtained for a swallowing event. Based on the updated mean and standard deviation, a reference interval for the individual baseline stability index is constructed.

[0037] The application scenarios for the examples are as follows: A community health service center conducted an early screening program for dysphagia in the elderly. A non-invasive swallowing function assessment was performed on a 65-year-old male subject (who complained of occasional choking when drinking water and had no clear history of neurological diagnosis). The test used a Doppler acoustic detection device. The subject was seated and the probe was placed on the surface of the thyroid cartilage in the neck. The laryngeal reflex signals were collected after three consecutive swallowings of 5ml of warm water. The signal sampling rate was 1000Hz. The raw signal is acquired, which includes both phase information (reflecting macroscopic displacement of the laryngeal tissue) and amplitude information (reflecting microscopic vibration and acoustic events of the laryngeal tissue). Real-time spectrum analysis of phase information is performed to identify low-frequency shift components concentrated in the range of 0.5 to 8 Hz. The cutoff frequency of the first bandpass filter is dynamically adjusted to 0.5 Hz to 8 Hz, and the low-frequency shift components are separated after filtering. Real-time spectrum analysis of amplitude information is performed to identify high-frequency vibration components concentrated in 100~500Hz and acoustic event components concentrated in 500~2000Hz; the second bandpass filter is adjusted to 100~500Hz and the third bandpass filter is adjusted to 500~2000Hz respectively, and the high-frequency vibration component and acoustic event component are separated after filtering. The corresponding solution logic is as follows: calculate the ratio of acceleration to duration from the low-frequency displacement components, retain it as a two-dimensional vector, and do not fuse it into a single value; The second-order difference of the low-frequency displacement component is calculated frame by frame to obtain the laryngeal acceleration signal. The first acceleration peak at the swallowing initiation point is identified and measured. =12.5cm / s 2 ; Identify the laryngeal elevation phase: initial 0.12s, final 0.38s, elevation duration = 0.38 - 0.12 = 0.26s; Identify the larynx's descent phase: beginning 0.42s, ending 0.75s, descent duration = 0.75 - 0.42 = 0.33s; Latinal rise and fall time imbalance: ; The final laryngeal motor coordination characteristics are represented by a two-dimensional vector: ; Parameter and numerical settings: Frequency range of high-frequency vibration components: , ; The power spectral density is calculated for the high-frequency vibration segment within the swallowing cycle, and the power spectral integral is obtained by integration. (Unit of energy); Calculate the centroid frequency of the spectrum ; Preset weights: , ; Substitute muscle strength performance characteristic value The calculation is as follows: =80×2.35×10⁻³+0.02×268=0.188+5.36=5.548; The parameter setting is: the total number of sampling points for the resting signal during the non-swallowing period (between two swallowing sessions). (Corresponding to a 5-second resting period); Maximum time scale Embedded dimension Similarity tolerance ; Entropy calculation results for samples at various scales: hour ; hour ; hour ; hour ; hour ; Substitute the resting period abnormal micro-motion characteristic values The calculation is as follows:

[0038] ; Calculate the structural similarity between two adjacent swallowing events from the time-spectrum graphs of the three swallowing events, and remove the mean. The first and second similarity scores were 0.86; the second and third similarity scores were 0.89. Therefore, the temporal spectrum similarity of adjacent swallowing events was... U It is 0.875; The reference parameters and weightings for healthy individuals are shown in Table 1 below: Table 1 Parameter and Weight Setting Table

[0039] Based on the content of Table 1, acceleration normalization = ; Imbalance Standardization = ; Muscle strength efficiency standardization = ; Sample entropy standardization = ; Pattern similarity standardization = ; The final calculated Doppler acoustic swallowing stability index for the current swallowing event is D = 0.18 × (-1.0) + 0.12 × 0.756 + 0.3 × (-0.673) + 0.2 × 1.4 + 0.2 × (-0.9) ≈ -0.191.

[0040] The Doppler acoustic swallowing stability index of the current swallowing event is compared with the preset group risk threshold and the reference range of the individual baseline stability index. The risk level of the swallowing event is determined based on the comparison results, and a differentiated message push strategy is implemented based on the risk level.

[0041] It should be explained that, based on a large sample of testing data from healthy elderly individuals aged 60-75, the distribution characteristics of the swallowing stability index D were statistically analyzed, and a four-level risk classification standard was pre-set: Normal (Level 0): When the index D≥0, it means that the swallowing function is at or above the average level of healthy people, and there is no risk of swallowing disorders. Mild risk (Level 1): When the index satisfies -1.0≤D<0, it indicates a mild decline in swallowing function, slightly below the average level of healthy people, with occasional choking risk and no clear pathological damage; Moderate risk (Level 2): ​​When the index meets -2.0≤D<-1.0, it represents moderate impairment of swallowing function, with clear abnormalities in motor coordination or muscle strength, and a significantly increased risk of aspiration, requiring further clinical evaluation; Severe risk (Level 3): When the index D < -2.0, it indicates severe impairment of swallowing function, highly suggestive of occult aspiration or organic lesions of the swallowing organs, and a high risk of aspiration pneumonia.

[0042] Using the baseline reference interval constructed from individual historical tests as a longitudinal reference, the group assessment results are graded: if the current D value falls within the baseline reference interval, it indicates that there is no significant fluctuation in individual function, and the risk level is based on the group threshold assessment result; if the current D value is lower than the lower limit of the baseline interval, it indicates that the individual's swallowing function has a statistically significant decline, and the risk level is upgraded by one level based on the group assessment; if the D value is lower than the lower limit of the baseline interval for two consecutive tests, it is judged as progressive swallowing function decline, the risk level is upgraded by one level based on the group assessment and labeled as progressive decline, and the individual is included in the key follow-up list.

[0043] Group-wide assessment: The subject's swallowing stability index D = -0.191, meeting the requirement of -1.0 ≤ D < 0, corresponding to a mild risk (Level I) within the group threshold; Individual longitudinal assessment: The subject's individual baseline reference range is [-0.264, 0.100], and the current D = -0.191 is within this range, not significantly lower than the baseline, therefore no upward adjustment of the risk level is necessary; Final comprehensive assessment result: Mild risk of dysphagia (Grade 1), functional status was not significantly different from their baseline, but slightly lower than the average level of healthy people of the same age. They were included in the 3-month follow-up, which was completely consistent with the clinical manifestations of occasional choking when drinking water reported by the subjects. Based on the final risk level, differentiated content is automatically pushed to three types of terminals (the subject's personal health terminal, the family member's linked account, and the community family doctor's work platform). The push for mild risk in this example is executed as follows, and the full-level rules are supplemented to fully present the solution logic.

[0044] This embodiment also provides a swallowing disorder risk identification system based on Doppler acoustic features, the system comprising: The feature extraction module is used to acquire laryngeal region reflection signals containing phase and amplitude information; adaptive bandpass filtering is performed on the laryngeal region reflection signals, and laryngeal motor coordination features, muscle efficacy features, resting period abnormal micro-movement features, and swallowing pattern stability features are extracted from the filtered signals. The swallowing stability index generation module is used to standardize the extracted features and combine them with the corresponding preset weight coefficients to obtain the Doppler acoustic swallowing stability index of the current swallowing event. The individual baseline fitting module is used to obtain the Doppler acoustic swallowing stability index values ​​of each swallowing event and perform stepwise fitting to construct the individual baseline stability index reference range. The risk assessment and push module is used to compare the Doppler acoustic swallowing stability index of the current swallowing event with the preset group risk threshold and the individual baseline stability index reference range. Based on the comparison results, the risk level of the swallowing event is determined, and a differentiated message push strategy is executed based on the risk level.

[0045] It should be noted that the calculation formulas and all parameters involved in the calculations in this application have been dimensionless beforehand. The process of dimensionless processing is well known in the industry and will not be described here.

[0046] Although the present invention has been disclosed above with reference to preferred embodiments, the embodiments are merely examples for illustrative purposes and are not intended to limit the present invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention should be determined by the claims.

Claims

1. A method for identifying swallowing disorder risk based on Doppler acoustic features, characterized in that, The method includes: The laryngeal region reflection signal containing phase and amplitude information was acquired; adaptive bandpass filtering was performed on the laryngeal region reflection signal, and laryngeal motor coordination features, muscle efficacy features, resting period abnormal micro-movement features, and swallowing pattern stability features were extracted from the filtered signal. The extracted features are standardized and combined with the corresponding preset weight coefficients to obtain the Doppler acoustic swallowing stability index of the current swallowing event. The Doppler acoustic swallowing stability index values ​​of each swallowing event were obtained and stepwise fitted to construct an individual baseline stability index reference range. The Doppler acoustic swallowing stability index of the current swallowing event is compared with the preset group risk threshold and the reference range of the individual baseline stability index. The risk level of the swallowing event is determined based on the comparison results, and a differentiated message push strategy is implemented based on the risk level.

2. The method for identifying swallowing disorder risk according to claim 1, characterized in that, The method for adaptive bandpass filtering of the throat region reflection signal is as follows: Real-time spectral analysis of the phase information in the throat region reflection signal is performed to determine the frequency band distribution of the low-frequency displacement component; Real-time spectral analysis of the amplitude information in the throat region reflection signal is performed to determine the frequency band distribution of high-frequency vibration components and acoustic event components; The cutoff frequency of the first bandpass filter is dynamically adjusted based on the frequency band distribution of the low-frequency displacement component, the cutoff frequency of the second bandpass filter is dynamically adjusted based on the frequency band distribution of the high-frequency vibration component, and the cutoff frequency of the third bandpass filter is dynamically adjusted based on the frequency band distribution of the acoustic event component. The phase information is filtered using the adjusted first bandpass filter to separate the low-frequency displacement component; The amplitude information is filtered by the adjusted second bandpass filter to separate the high-frequency vibration component; The amplitude information is filtered by the adjusted third bandpass filter to separate the acoustic event components.

3. The method for identifying swallowing disorder risk according to claim 2, characterized in that, The extracted laryngeal motor coordination characteristics, muscle strength efficiency characteristics, resting-phase abnormal micromovement characteristics, and swallowing pattern stability characteristics include: The peak value of the pharyngeal initiation acceleration and the time asymmetry ratio of laryngeal elevation and fall were calculated from the low-frequency displacement components to obtain the characteristics of intralarynx movement coordination. The power spectrum integral and spectral centroid frequency of the high-frequency vibration components within the swallowing cycle are calculated from the high-frequency vibration components to obtain the muscle efficacy characteristics. Multi-scale sample entropy was calculated from the low-frequency displacement component and high-frequency vibration component in the signal segment during the non-swallowing period to obtain the abnormal micro-motion characteristics during the resting period. The similarity of the time spectrum of adjacent swallowing events is calculated from the time spectrum of low-frequency displacement components, high-frequency vibration components and acoustic event components of multiple swallowing events to obtain the swallowing pattern stability characteristics.

4. The method for identifying swallowing disorder risk according to claim 3, characterized in that, The method for obtaining the characteristics of intralarynx motor coordination is as follows: The rate of change of displacement is calculated frame by frame for the low-frequency displacement component, and then the rate of change of displacement is calculated to obtain the acceleration signal of the throat motion. The first acceleration peak corresponding to the starting point of the swallowing action in the acceleration signal is detected, and the amplitude of this peak is extracted as the peak value of the pharyngeal initiation acceleration. Identify the start and end times of the throat lifting phase from the low-frequency displacement components, and calculate the duration of the lifting. Identify the start and end times of the throat fall-off phase from the low-frequency displacement components and calculate the fall-off duration. Divide the duration of the rise by the duration of the fall to obtain the time imbalance of the rise and fall of the throat. By combining the peak acceleration at the pharyngeal initiation phase with the time imbalance of larynx elevation and descent, the characteristics of intralarynx movement coordination are obtained.

5. The method for identifying swallowing disorder risk according to claim 3, characterized in that, The method for obtaining muscle strength performance characteristics is as follows: The high-frequency vibration components are segmented into swallowing cycles to determine the start and end times of the swallowing action, and high-frequency vibration signal segments within the swallowing cycle are extracted. The power spectral density of the extracted high-frequency vibration signal segment is calculated, and the power spectral density is integrated in the frequency domain to obtain the power spectral integral of the high-frequency vibration component within the swallowing cycle. Spectral analysis was performed on the extracted high-frequency vibration signal segments to calculate their centroid frequencies. By combining the power spectrum integral with the centroid frequency of the spectrum, the muscle strength performance characteristics are obtained. The expression for the calculation process is as follows: ； In the formula, Represents the power spectrum integral; This indicates the upper frequency limit of the high-frequency vibration component; This indicates the lower limit of the frequency of the high-frequency vibration component; Indicates the frequency of high-frequency vibration signals. Power spectral density at; Represents a tiny increment in frequency; Indicates the centroid frequency of the spectrum; Indicates the characteristic value of muscle strength performance; Represents the power spectrum integral The corresponding preset weighting coefficients; Indicates the centroid frequency of the spectrum The corresponding preset weighting coefficients.

6. The method for identifying swallowing disorder risk according to claim 3, characterized in that, The method for obtaining abnormal micro-motion characteristics during the resting period is as follows: The signal segments that are not swallowed are extracted from the low-frequency displacement components, and the signal segments that are not swallowed are extracted from the high-frequency vibration components. By splicing low-frequency displacement signal segments with high-frequency vibration signal segments, a resting signal during the non-swallowing period is obtained. Multi-scale coarse-graining was performed on the non-swallowing resting signal to obtain coarse-grained sequences at multiple time scales; Calculate the sample entropy for the coarse-grained sequence at each time scale to obtain the sample entropy value corresponding to each scale; The average of the sample entropy values ​​at each scale is taken to obtain the multi-scale sample entropy, which is used as the resting period abnormal micro-motion characteristic. The expression for its calculation process is as follows: ； In the formula, Indicates the first Coarse-grained sequence at the i-th time scale The value of each point; This indicates the resting signal during the non-swallowing period at the first... The amplitude of each sampling point; This represents the total number of sampling points for the resting signal during the non-swallowing period. Indicates the time scale factor; Indicates the maximum time scale; Indicates the first The length of coarse-grained sequences at each time scale; Indicates the embedding dimension; Indicates similarity tolerance; Indicates based on In the embedding dimension is The probability that two subsequences are similar when +1; Indicates based on In the embedding dimension is The probability that two subsequences are similar; Indicates the first Based on time scales Calculated sample entropy value; This represents the characteristic value of abnormal micro-movements during the resting period.

7. The method for identifying swallowing disorder risk according to claim 6, characterized in that, The expression for obtaining the Doppler acoustic swallowing stability index of the current swallowing event is as follows: ； In the formula, The Doppler acoustic swallowing stability index represents the current swallowing event; The preset weighting coefficients represent the characteristics of intralarynx motor coordination. This indicates the peak acceleration at the start of the pharyngeal phase; This indicates the unevenness in the timing of the rise and fall of the larynx. The standard deviation of the intralarynx motor coordination characteristics in a healthy population sample; This represents the mean of intralarynx motor coordination characteristics in a healthy population sample; Represents the preset weighting coefficients corresponding to muscle strength performance characteristics; Indicates the characteristic value of muscle strength performance; This represents the mean of muscle strength efficacy characteristics in a healthy population sample; This represents the standard deviation of muscle strength efficacy characteristics in a healthy population sample. This represents the preset weighting coefficient corresponding to the abnormal micro-motion characteristics during the resting period; This represents the characteristic value of abnormal micro-movements during the resting period; This represents the mean value of abnormal micro-motion characteristics during the resting period in a healthy population sample; This represents the standard deviation of abnormal micromotion characteristics during the resting period in a healthy population sample. The preset weighting coefficients represent the stability characteristics of swallowing patterns; The temporal spectrum similarity of adjacent swallowing events; This represents the mean of swallowing pattern stability characteristics in a healthy population sample. This represents the standard deviation of swallowing pattern stability characteristics in a healthy population sample.

8. The method for identifying swallowing disorder risk according to claim 1, characterized in that, The method for constructing the reference interval for the individual baseline stability index is as follows: Obtain the Doppler acoustic swallowing stability index values ​​of each swallowing event of the current user and form a historical index sequence; Statistical analysis is performed on the historical index sequence to calculate the mean and standard deviation of the historical index sequence, and the mean and standard deviation are updated each time a new Doppler acoustic swallowing stability index value is obtained for a swallowing event. Based on the updated mean and standard deviation, a reference interval for the individual baseline stability index is constructed.

9. A swallowing disorder risk identification system based on Doppler acoustic features, used to implement the swallowing disorder risk identification method according to any one of claims 1-8, characterized in that, The system includes: The feature extraction module is used to acquire laryngeal region reflection signals containing phase and amplitude information; adaptive bandpass filtering is performed on the laryngeal region reflection signals, and laryngeal motor coordination features, muscle efficacy features, resting period abnormal micro-movement features, and swallowing pattern stability features are extracted from the filtered signals. The swallowing stability index generation module is used to standardize the extracted features and combine them with the corresponding preset weight coefficients to obtain the Doppler acoustic swallowing stability index of the current swallowing event. The individual baseline fitting module is used to obtain the Doppler acoustic swallowing stability index values ​​of each swallowing event and perform stepwise fitting to construct the individual baseline stability index reference range. The risk assessment and push module is used to compare the Doppler acoustic swallowing stability index of the current swallowing event with the preset group risk threshold and the individual baseline stability index reference range. Based on the comparison results, the risk level of the swallowing event is determined, and a differentiated message push strategy is executed based on the risk level.

Images