Pet intelligent necklace integrated with sign collection

Abstract

The application discloses a pet intelligent necklace integrated with sign collection, and relates to the field of pet wearable monitoring devices, which comprises a necklace belt body, a sign collection component, a wearing state collection component, a posture collection component, a processing unit and a data output unit. The sign collection component collects a first deformation signal containing neck respiratory movement and necklace overall movement, a reference deformation signal representing necklace overall movement and a throat acoustic response signal. The processing unit determines an effective monitoring section according to contact pressure, pressure distribution, movement state and neck posture, separates a common mode deformation component by using the reference deformation signal, and obtains a neck respiratory movement representation signal; the neck respiratory movement feature and the throat acoustic response feature in the same respiratory cycle are paired, compared with the corresponding baseline of the individual, and a respiratory movement-acoustic response mismatch event is identified. The necklace can reduce the influence of wearing state and necklace overall movement on the monitoring result, and output the mismatch event and corresponding wearing state information.

Description

Technical Field

[0001] This invention relates to the field of wearable monitoring devices for pets, specifically to a smart pet collar that integrates vital sign collection. Background Technology

[0002] Pet collars are characterized by their relatively fixed wearing position, making them suitable for continuously collecting physiological signals from the neck. Existing pet collars typically utilize a single accelerometer, microphone, or strain sensor to obtain information on activity levels, respiratory-related vibrations, or sound. However, when pets turn their heads, roll over, walk, or scratch, the collar can stretch, bend, or slip, making it difficult for a single deformation channel to distinguish between localized respiratory movements of the neck and the overall movement of the collar. Open acoustic channels are also easily affected by ambient sound, while contact acoustic channels change their coupling gain depending on the pressure applied and the sensor position.

[0003] On the other hand, a collar that is too loose, partially suspended, has a misaligned pressure distribution, or changes in the pet's posture can all affect both the deformation signal and the throat acoustic response. If judgment is made directly based on amplitude changes without confirming the contact and movement status, changes in wearing conditions may be mistaken for changes in vital signs. Furthermore, individual differences exist among pets in neck circumference, fur thickness, normal respiratory amplitude, and throat acoustic characteristics; using a uniform single-channel threshold is insufficient to reliably reflect the current signal changes in the pet. Summary of the Invention

[0004] This invention provides a smart pet collar that integrates vital sign acquisition. The technical problem to be solved is that the neck vital sign signals of existing pet collars are easily affected by the overall movement of the collar and changes in contact status, and the single-channel universal threshold is difficult to adapt to individual differences in pets, resulting in insufficient stability of respiratory-related monitoring results.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by the present invention is as follows: 1. A smart pet collar with integrated vital sign acquisition, characterized in that it includes a collar body, a vital sign acquisition component, a wearing status acquisition component, a posture acquisition component, a processing unit, and a data output unit;

[0006] The vital signs acquisition component includes a first deformation sensor, a reference deformation sensor, and a pharyngeal acoustic sensor disposed on the collar. The first deformation sensor is used to acquire a first deformation signal including the pet's neck respiratory movement and the overall movement of the collar. The reference deformation sensor is used to acquire a reference deformation signal characterizing the overall movement of the collar. The pharyngeal acoustic sensor is used to acquire tissue sound transmission and airway vibration in the pet's pharyngeal region and form a pharyngeal acoustic response signal.

[0007] The wearing status acquisition component includes multiple contact pressure sensors for acquiring the contact pressure and pressure distribution between the collar strap and the pet's neck;

[0008] The posture acquisition component is used to acquire the pet's movement status and neck posture;

[0009] The processing unit is connected to the vital sign acquisition component, the wearing status acquisition component, the posture acquisition component, and the data output unit, and is used to: determine an effective monitoring segment based on the contact pressure, the pressure distribution, the movement state, and the neck posture; separate the common-mode deformation component related to the overall movement of the collar from the first deformation signal based on the first deformation signal and the reference deformation signal to obtain a neck respiratory movement characterization signal; divide the respiratory cycle within the effective monitoring segment and time-pair the neck respiratory movement characterization signal and the pharyngeal acoustic response signal within the same respiratory cycle; identify respiratory movement-acoustic response mismatch events based on the degree of deviation between the time-paired neck respiratory movement characteristics and pharyngeal acoustic response characteristics and the corresponding baseline of the individual pet; and output the respiratory movement-acoustic response mismatch event and the corresponding wearing status information through the data output unit.

[0010] Furthermore, the collar body includes a respiratory sensing segment that contacts the area corresponding to the ventral trachea on the pet's neck, and a common-mode reference segment spaced apart from the respiratory sensing segment;

[0011] The respiratory sensing segment is provided with a flexible coupling seat, the first deformation sensor and the pharyngeal acoustic sensor are disposed on the flexible coupling seat, and the reference deformation sensor is disposed on the common mode reference segment;

[0012] The first deformation sensor and the reference deformation sensor have the same deformation-sensitive direction, and the common-mode reference segment is offset from the corresponding area of ​​the ventral trachea of ​​the pet's neck, so that the local respiratory motion component in the reference deformation signal is smaller than the local respiratory motion component in the first deformation signal.

[0013] Furthermore, at least one of the contact pressure sensors is respectively provided on both sides of the flexible coupling seat;

[0014] When the contact pressure on both sides of the flexible coupling seat reaches the lower limit of the contact pressure, the difference between the contact pressures on both sides does not exceed the pressure difference threshold, the displacement of the pressure center does not exceed the displacement threshold, and each contact pressure does not exceed the upper limit of the pressure safety limit, the processing unit determines the corresponding collection section as the effective contact section.

[0015] The processing unit identifies the effective monitoring segment as the segment in the effective contact area where the pet's movement is below the movement threshold and the neck posture remains stable.

[0016] Furthermore, the processing unit selects a common-mode calibration segment from the acquisition segment where the contact state is stable and the attitude acquisition component detects the overall movement of the collar strap;

[0017] The processing unit determines the common-mode deformation transfer coefficient by minimizing the common-mode separation residual based on the correspondence between the first deformation signal and the reference deformation signal within the common-mode calibration section, and subtracts the common-mode deformation component corresponding to the reference deformation signal from the first deformation signal based on the common-mode deformation transfer coefficient.

[0018] Furthermore, the attitude acquisition component includes an accelerometer and an angular velocity sensor disposed on the collar strap;

[0019] The processing unit identifies at least one action among rolling over, raising head, walking, scratching, licking, and swallowing based on changes in acceleration, angular velocity, and contact pressure, and excludes the segment in which the identified action occurs and the preset buffer zone before and after the action from the effective monitoring segment.

[0020] Furthermore, the processing unit is configured to: determine successive respiratory cycles based on the periodic boundaries of the neck respiratory motion representation signal, and extract the neck motion amplitude from the i-th respiratory cycle. ; Extract acoustic response energy from the pharyngeal acoustic response signal corresponding to the i-th respiratory cycle time. Acoustic response time difference relative to neck breathing movement and acoustic response continuity Select at least a preset number of effective respiratory cycles from the same contact pressure range and neck posture range as baseline respiratory cycles, and establish the corresponding baseline for the individual pet based on the baseline respiratory cycles; when the pet smart collar is re-worn or the pressure center of the flexible coupling seat continuously exceeds the baseline update threshold, re-establish or update the corresponding baseline for the individual pet.

[0021] Furthermore, for the i-th respiratory cycle, the processing unit, based on the cross-modal feature vector... Calculate the degree of mismatch :

[0022]

[0023]

[0024] In the formula, and These are the baseline values ​​for acoustic response energy and neck movement amplitude for the corresponding individual pet. Let be the duration of the i-th respiratory cycle. and , respectively, are the mean and standard deviation of the j-th cross-modal feature in the baseline for the corresponding pet individual. Let the weights of the j-th cross-modal feature satisfy the condition that the sum of all weights is 1 and Not less than 0, A constant that is greater than zero;

[0025] When the mismatch If the mismatch threshold is greater than the mismatch distribution determined based on the baseline respiratory cycle, the i-th respiratory cycle is identified as an abnormal mismatch cycle.

[0026] Furthermore, the processing unit counts the consecutive number and cumulative number of the abnormal mismatch cycles within the event statistics window;

[0027] When the number of consecutive abnormal mismatch cycles reaches the first threshold, or the cumulative number of abnormal mismatch cycles within the event statistics window reaches the second threshold, a first-level respiratory motion-acoustic response mismatch event is generated.

[0028] When a first-level respiratory motion-acoustic response mismatch event is followed by at least one of the following events: increased respiratory amplitude, change in respiratory rate, head lifting, turning over, or change from rest to activity, the first-level respiratory motion-acoustic response mismatch event is marked as a second-level respiratory motion-acoustic response mismatch event.

[0029] Furthermore, when any contact pressure is lower than the lower limit of the contact pressure, the processing unit marks the corresponding collection section as a contact failure section and outputs a position adjustment prompt;

[0030] When the contact pressure collected by at least two spaced contact pressure sensors is continuously higher than the pressure safety limit, the processing unit marks the corresponding collection section as a continuous pressure section, outputs a collar too tight warning, and prohibits the generation of the respiratory motion-acoustic response mismatch event using the continuous pressure section.

[0031] Furthermore, the vital signs acquisition component also includes at least one of a heartbeat vibration sensor and a neck temperature sensor;

[0032] The heartbeat vibration sensor is used to collect the heartbeat vibration signal of the pet, and the processing unit obtains the heart rate change before and after the respiratory motion-acoustic response mismatch event based on the heartbeat vibration signal.

[0033] The neck temperature sensor is used to collect the pet's neck temperature.

[0034] The data output unit is used to output at least one of the following: the occurrence time, duration, event level, wearing status, heart rate change, and neck temperature of the respiratory motion-acoustic response mismatch event, and to save the vital signs evidence data within a preset time period before and after the occurrence of the respiratory motion-acoustic response mismatch event.

[0035] The beneficial effects of this invention are as follows:

[0036] 1. Contact pressure, pressure distribution, motion state, and neck posture are used together to determine the effective monitoring section, so that subsequent calculations are limited to the section where the sensor fits reliably and the motion interference is low. This can reduce the impact of collar loosening, local suspension, and sudden posture changes on the vital sign monitoring results.

[0037] 2. The first deformation signal includes local neck respiratory motion and overall collar motion, and the reference deformation signal represents the overall collar motion. By separating the common mode deformation component based on the correspondence between the two deformation signals, the common interference caused by overall collar stretching, bending and slippage can be reduced, providing a relatively stable neck respiratory motion representation signal for respiratory cycle division.

[0038] 3. The neck respiratory movement characteristics and pharyngeal acoustic response characteristics within the same respiratory cycle are time-paired and compared with the individual baseline formed by the current pet in a similar wearing state; this processing utilizes the correspondence between two physical modes rather than the absolute amplitude of a single signal, which can reduce the influence caused by breed, neck circumference, hair thickness and sensor coupling differences;

[0039] 4. The output results simultaneously include respiratory movement-acoustic response mismatch events and wearing status information, enabling users to distinguish between deviations in the correspondence between vital signs and wearing abnormalities, facilitating review and adjustment of the collar; the above effects are all generated by the overall structure and processing flow of the independent technical solution of this invention. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the system composition of the pet smart collar according to an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of the circumferential structural arrangement of the collar strap body according to an embodiment of the present invention;

[0042] Figure 3 This is a cross-sectional schematic diagram of the respiratory sensing segment fitting against the pet's neck, according to an embodiment of the present invention.

[0043] Figure 4 This is a schematic diagram of the vital sign signal processing flow according to an embodiment of the present invention;

[0044] Figure 5 This is a schematic diagram of the abnormal mismatch cycle aggregation and event classification process in an embodiment of the present invention.

[0045] Reference numerals: 10. Collar strap; 11. Respiratory sensing segment; 12. Flexible coupling seat; 13. Common mode reference segment; 14. Electronic compartment; 20. Vital signs acquisition component; 21. First deformation sensor; 22. Reference deformation sensor; 23. Pharyngeal acoustic sensor; 24. Heartbeat vibration sensor; 25. Neck temperature sensor; 30. Wearing status acquisition component; 31. Contact pressure sensor; 40. Posture acquisition component; 41. Accelerometer; 42. Angular velocity sensor; 100. Processor; 110. Acquisition circuit; 120. Memory; 130. Clock module; 140. Power supply module; 200. Data output unit; 210. User terminal; 300. Pet neck. Detailed Implementation

[0046] To make the content of this invention easier to understand, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings.

[0047] This invention provides a smart pet collar with integrated vital sign acquisition, including a collar body, a vital sign acquisition component, a wearing status acquisition component, a posture acquisition component, a processing unit, and a data output unit. The vital sign acquisition component includes a first deformation sensor, a reference deformation sensor, and a pharyngeal acoustic sensor disposed on the collar body. The first deformation sensor is used to acquire a first deformation signal including the pet's neck respiratory movement and the overall movement of the collar body. The reference deformation sensor is used to acquire a reference deformation signal characterizing the overall movement of the collar body. The pharyngeal acoustic sensor is used to acquire tissue sound transmission and airway vibration in the pet's pharyngeal region and form a pharyngeal acoustic response signal.

[0048] The wearing status acquisition component includes multiple contact pressure sensors for acquiring the contact pressure and pressure distribution between the collar and the pet's neck; the posture acquisition component is used to acquire the pet's movement state and neck posture; the processing unit is connected to the vital signs acquisition component, the wearing status acquisition component, the posture acquisition component, and the data output unit, and is used for: determining the effective monitoring segment based on contact pressure, pressure distribution, movement state, and neck posture; separating the common-mode deformation component related to the overall movement of the collar from the first deformation signal based on the first deformation signal and the reference deformation signal to obtain the neck respiratory movement characterization signal; dividing the respiratory cycle within the effective monitoring segment and performing time pairing between the neck respiratory movement characterization signal and the pharyngeal acoustic response signal within the same respiratory cycle; identifying respiratory movement-acoustic response mismatch events based on the degree of deviation between the time-paired neck respiratory movement characteristics and pharyngeal acoustic response characteristics and the corresponding baseline of the individual pet; and outputting the respiratory movement-acoustic response mismatch event and the corresponding wearing status information through the data output unit.

[0049] Specifically, such as Figures 1-4As shown, the collar strap 10 constitutes a mechanical carrier. The vital sign acquisition component 20 and the wearing status acquisition component 30 are located on the ventral side and its adjacent position. The posture acquisition component 40, the processor 100, and the data output unit 200 can be located in the electronic compartment 14 on the back side. The processing unit consists of the processor 100, the acquisition circuit 110, the memory 120, and the clock module 130. The power supply module 140 supplies power to the processing unit, and each sampling channel is aligned using a unified clock or timestamp.

[0050] The first deformation sensor 21 converts the mixed deformation caused by localized neck breathing movements and the overall stretching and bending of the collar into a first deformation signal. The reference deformation sensor 22 is positioned where localized breathing has minimal impact, providing a reference value for the overall collar movement. The pharyngeal acoustic sensor 23 preferably employs a contact piezoelectric microphone or a contact accelerometer to obtain tracheal sounds and pharyngeal vibrations through neck tissue. The neck breathing movement characterization signal is not equivalent to actual breathing effort, and the pharyngeal acoustic response is not equivalent to actual airflow. Multiple contact pressure sensors 31 describe whether the sensors are in contact and whether the pressure distribution is stable, while the posture acquisition component 40 describes movement states such as head turning, rolling over, and scratching.

[0051] The processor 100 first determines the valid monitoring segment, then performs common-mode separation, period division, cross-modal pairing, and baseline comparison. The output of the previous stage constitutes the input of the next stage. The data output unit 200 outputs mismatch events and wearing status, enabling separate verification of changes in the correspondence between vital signs and wearing abnormalities. The collar strap 10 can adopt a fabric-covered elastic band, a flexible polymer band, or a combination structure of flexible segments and low-elongation segments, and is equipped with length adjustment components and buckles. For example, the deformation and contact pressure channels are sampled at 100 Hz to 200 Hz, the attitude channel at 50 Hz to 100 Hz, and the acoustic channel at 8 kHz to 16 kHz. Data from different sampling rates are aligned using timestamps from the clock module 130. Validity screening, common-mode separation, period division, cross-modal pairing, baseline comparison, and event output are executed sequentially; any invalid segment is not included in subsequent vital sign judgment.

[0052] Furthermore, the collar includes a respiratory sensing segment that contacts the area corresponding to the ventral trachea on the pet's neck, and a common-mode reference segment spaced apart from the respiratory sensing segment. The respiratory sensing segment is provided with a flexible coupling seat, on which a first deformation sensor and a pharyngeal acoustic sensor are disposed, and on which a reference deformation sensor is disposed. The first deformation sensor and the reference deformation sensor have the same deformation-sensitive direction, and the common-mode reference segment is offset from the area corresponding to the ventral trachea on the pet's neck, so that the local respiratory motion component in the reference deformation signal is smaller than the local respiratory motion component in the first deformation signal.

[0053] Specifically, such as Figure 2 and Figure 3 As shown, the respiratory sensing segment 11 is located on the ventral side of the pet's neck 300°. The flexible coupling seat 12 keeps the first deformation sensor 21 and the pharyngeal acoustic sensor 23 in an elastic fit against the fur and skin surface. The common-mode reference segment 13 is offset from the corresponding area of ​​the trachea, and can still respond to the overall expansion and contraction of the collar, but its local respiratory component is lower than that of the respiratory sensing segment 11. The two deformation sensors are of the same model or have been sensitively calibrated, and their sensitive axes are both along the circumference of the collar, thereby making the two signals comparable in response to the overall movement of the collar.

[0054] This arrangement forms a local respiratory sensing channel, a collar motion reference channel, and a tissue acoustic transmission channel, providing physically corresponding inputs for subsequent common-mode separation and cross-modal comparison. The flexible coupling seat 12 can be formed of silicone, thermoplastic elastomer, or elastic foam, and its outer surface is adapted to the curved surface of the neck. The first deformation sensor 21 is positioned circumferentially around the collar, and the pressure direction of the pharyngeal acoustic sensor 23 faces the neck tissue. The common-mode reference segment 13 can be located on the side or back of the collar, circumferentially separated from the respiratory sensing segment 11 by one-quarter to one-half of the effective circumference. The reference segment is allowed to contain a weak respiratory component; the algorithm only requires that its local respiratory sensitivity be lower than that of the respiratory sensing segment, without requiring the reference signal to be absolutely free of respiratory components.

[0055] Furthermore, at least one contact pressure sensor is installed on each side of the flexible coupling seat; when the contact pressure on both sides of the flexible coupling seat reaches the lower limit of contact pressure, the difference between the contact pressures on both sides does not exceed the pressure difference threshold, the displacement of the pressure center does not exceed the displacement threshold, and each contact pressure does not exceed the upper limit of pressure safety, the corresponding acquisition section is determined as the effective contact section; the processing unit determines the section in the effective contact section where the pet's movement is lower than the movement threshold and the neck posture remains stable as the effective monitoring section.

[0056] Specifically, such as Figure 3 and Figure 4 As shown, the contact pressure sensors 31 on both sides of the flexible coupling seat 12 are preferably symmetrically arranged. When the pressure on both sides reaches the lower limit, it indicates that basic contact is maintained; when the pressure difference and pressure center displacement do not exceed the threshold, it indicates that the coupling seat has no significant tilting or slippage; and when each pressure does not exceed the safety upper limit, it indicates that there is no continuous pressure. Let the position of the nth contact pressure sensor 31 be... Pressure is The pressure center is calculated using the following formula: The lower limit of contact pressure and the pressure difference threshold can be determined by the stable pressure distribution during the first wear, the displacement threshold can be determined based on the width of the flexible coupling seat, and the upper limit of pressure safety is verified through product safety testing.

[0057] After effective contact, the processor 100 continues to use motion and neck posture to exclude obvious movement segments, thus limiting the effective monitoring segment by contact conditions, motion conditions, and posture conditions. The motion can be calculated based on acceleration and angular velocity signals collected by the posture acquisition component 40, for example, by taking the root mean square or peak value of the sum of acceleration or angular velocity within a preset time window as the motion. The motion threshold can be calibrated by combining the distribution of motion in a quiet state and the product's sensitivity requirements. During initial calibration, the median and normal fluctuation range of each contact pressure in a quiet state can be recorded. The lower limit of contact pressure is determined by a certain proportion of the median of stable pressure, the pressure difference threshold is determined based on the normal distribution of the left-right pressure difference, and the displacement threshold is determined based on the width of the flexible coupling seat 12. The upper limit of pressure safety cannot be estimated solely by the signal amplitude; it should be verified through a wearing safety test, taking into account the strap width, sensor contact area, and duration. Thresholds are grouped and stored according to neck circumference and hair type; calibration values ​​for a single pet are not directly used for all pets.

[0058] Furthermore, the processing unit selects a common-mode calibration section from the acquisition section where the contact state is stable and the attitude acquisition component detects the overall movement of the collar strap; based on the correspondence between the first deformation signal and the reference deformation signal within the common-mode calibration section, the processing unit determines the common-mode deformation transfer coefficient by minimizing the common-mode separation residual, and subtracts the common-mode deformation component corresponding to the reference deformation signal from the first deformation signal based on the common-mode deformation transfer coefficient.

[0059] Specifically, the processor 100 selects a common-mode calibration window from the segment where contact is stable and the collar exhibits overall movement. Let the first deformation signal be... The reference deformation signal is The calibration window is The common-mode strain transfer coefficient is determined according to the following formula:

[0060]

[0061] The neck respiratory movement signal is obtained according to the following formula: In the formula, and The mean value is the window value, and δ is a positive number to prevent the denominator from being zero. This calculation reduces the overall deformation of the collar that changes synchronously with the reference signal, preserving more local periodic movements of the neck. The transfer coefficients can be saved separately according to the contact pressure range and the posture range, and updated after a new effective calibration window appears. The common-mode calibration window needs to simultaneously meet the following conditions: stable contact state, sufficient variation amplitude of the reference deformation signal, and no local movements such as swallowing, licking, or scratching. If the two signals are close to constant within the window, the transfer coefficients are easily affected by noise, and this window is not used for updating. Common-mode separation does not require the elimination of all motion effects; the processor 100 can also filter the output signal in the breathing frequency band and combine it with the motion segment to eliminate residual interference. When re-wearing or when the contact pressure group changes, the transfer coefficients of the corresponding group are used or the calibration is re-performed.

[0062] Furthermore, the posture acquisition component includes an accelerometer and an angular velocity sensor installed on the collar strap; the processing unit identifies at least one action among rolling over, head raising, walking, scratching, licking, and swallowing based on changes in acceleration, angular velocity, and contact pressure, and excludes the segment in which the identified action occurs and the preset buffer segment before and after the action from the effective monitoring segment.

[0063] Specifically, the attitude acquisition component 40 includes an accelerometer 41 and an angular velocity sensor 42. Rolling over is typically accompanied by changes in attitude angle and center of pressure; head lifting manifests as a change in pitch angle; walking manifests as periodic acceleration; scratching manifests as high-frequency impacts and rapid pressure fluctuations; and licking and swallowing may manifest as ventral deformation or short acoustic pulses. The processor 100 forms candidate action segments based on the sum of acceleration, sum of angular velocity, changes in attitude angle, and displacement of the center of pressure, and then confirms the action type based on duration and signal morphology. Action segments and their preceding and following buffer zones do not participate in the vital sign mismatch judgment. Only after the contact pressure and attitude stabilize again after the buffer zone ends does the data re-enter the effective monitoring segment.

[0064] Candidate action segments can be triggered by at least one of the following exceeding a threshold: short-term acceleration resultant, angular velocity resultant, attitude angle change, and pressure center displacement. A shorter buffer time is used for normal head tilting, and a longer buffer time for vigorous scratching. Swallowing and licking can be confirmed by isolated broadband pulses in the throat acoustic signal, but this acoustic segment is only used for action exclusion. The expiration of the buffer time does not mean automatic recovery; the processor 100 still needs to confirm that the contact pressure and attitude continuously meet the stability conditions to prevent premature entry into monitoring before the collar rebounds or slips have ended.

[0065] Furthermore, the processing unit is used to: determine successive respiratory cycles based on the periodic boundaries of the neck respiratory motion representation signal, and extract the neck motion amplitude from the i-th respiratory cycle. ; Extract acoustic response energy from the pharyngeal acoustic response signal corresponding to the i-th respiratory cycle time. Acoustic response time difference relative to neck respiratory movement and acoustic response continuity Select at least a preset number of effective respiratory cycles from the same contact pressure range and neck posture range as the baseline respiratory cycle, and establish a corresponding baseline for each pet based on the baseline respiratory cycle; when the pet smart collar is re-worn or the pressure center of the flexible coupling seat continuously exceeds the baseline update threshold, re-establish or update the corresponding baseline for each pet.

[0066] Specifically, the processor 100 performs respiratory frequency filtering on the neck respiratory motion representation signal and determines the period boundary based on adjacent peaks, troughs, or zero crossings. Neck motion amplitude The peak-to-valley difference of the i-th period can be taken as the acoustic response energy. The sum of squares or root mean square of the corresponding acoustic segments can be taken, and the time difference can be taken. The continuity can be determined by the time difference between the main motion peak and the main acoustic envelope peak. It is the ratio of the effective acoustic response duration to the corresponding respiratory phase duration.

[0067] Individual baselines are established separately for contact pressure and stable posture intervals. Each group can collect 30 to 100 effective respiratory cycles. After removing movement and significant outlier cycles, the mean, standard deviation, or quantiles of the features are calculated. For initial wear and when the number of baselines is insufficient, only data quality and trends are output, not mismatch events. Baselines are re-established upon re-wearing or when the pressure center continues to shift. Only cycles with stable contact and mismatch levels not exceeding the update gating threshold are used for baseline updates to prevent persistent anomalies from being absorbed into the normal baseline. Cycles marked as invalid are those with movement markers near cycle boundaries, cycle durations exceeding the target pet's reasonable range, or movement amplitudes below the noise threshold. Individual baselines can divide contact pressure into low, medium, and high effective intervals, and stable postures into left lateral recumbent, right lateral recumbent, prone, and other postures. Baseline values ​​and dispersion are calculated independently for each group. Baseline updates use a limited rate, absorbing only data from multiple consecutive cycles without mismatch events and with stable contact status, allowing the baseline to adapt to normal, slow changes without incorporating short-term anomalies.

[0068] Furthermore, for the i-th respiratory cycle, the processing unit calculates the cross-modal feature vector... Calculate the degree of mismatch :

[0069]

[0070]

[0071] In the formula, and These are the baseline values ​​for acoustic response energy and neck movement amplitude for the corresponding individual pet. Let be the duration of the i-th respiratory cycle. and , respectively, are the mean and standard deviation of the j-th cross-modal feature in the baseline for the corresponding pet individual. Let the weights of the j-th cross-modal feature satisfy the condition that the sum of all weights is 1 and Not less than 0, It is a constant greater than zero.

[0072] When mismatch If the mismatch threshold is greater than the mismatch distribution determined based on the baseline respiratory cycle, the i-th respiratory cycle is identified as an abnormal mismatch cycle.

[0073] Specifically, the cross-modal eigenvector simultaneously characterizes the amplitude, temporal, and continuity of the acoustic response relative to neck movement. and These represent the changes in current acoustic energy and neck movement amplitude relative to an individual baseline, respectively. The time difference was normalized according to the duration of the respiratory cycle. It indicates the degree of discontinuity in the acoustic response.

[0074] Baseline Mean and standard deviation From baseline groups corresponding to the current contact pressure and attitude, Indicate the contribution of each feature. To prevent the denominator from approaching zero, the mismatch threshold can be taken as the 95th percentile of the baseline mismatch, or adjusted based on validation data. Exceeding the threshold only indicates that the correspondence between the two measurable signals deviates from the individual baseline, and does not directly prove that airflow has stopped or that a specific disease exists. All three cross-modal features are dimensionless and can be processed within the same standardization framework. Example weights can be set to amplitude relation 0.5, time relation 0.3, and continuity 0.2, with the sum of all weights remaining at 1; optimization based on calibration samples is also possible, but the sample composition, optimization objective, and validation method must be preserved. Constants Positive numbers within the range of 10 to the power of -6 to 10 to the power of -3 can be used. If the goal is to reduce false alarms, the baseline quantile threshold can be increased; if the goal is to increase screening sensitivity, the quantile can be decreased, but changes in false alarms should be evaluated using independent validation data.

[0075] Furthermore, the processing unit counts the consecutive and cumulative number of abnormal mismatch cycles within the event statistics window; when the number of consecutive abnormal mismatch cycles reaches a first threshold, or the number of cumulative abnormal mismatch cycles within the event statistics window reaches a second threshold, a first-level respiratory motion-acoustic response mismatch event is generated; when a first-level respiratory motion-acoustic response mismatch event is followed by at least one of the following events: increased respiratory amplitude, change in respiratory rate, head lifting, turning over, or change from rest to activity, the first-level respiratory motion-acoustic response mismatch event is marked as a second-level respiratory motion-acoustic response mismatch event.

[0076] Specifically, such as Figure 5 As shown, a single abnormal cycle does not directly generate an event. The processor 100 simultaneously counts consecutive abnormal cycles and cumulative abnormal cycles within a sliding statistical window. When either quantity reaches the corresponding threshold, a Level 1 mismatch event is generated to reduce false alarms caused by occasional noise or short-term interference that has not been completely eliminated. After a Level 1 event, the response observation window checks for increased respiratory amplitude, changes in respiratory rate, head lifting, turning over, or a change from rest to activity. The occurrence of at least one response is marked as a Level 2 event. Post-event responses serve only as reinforcing evidence and do not constitute a mismatch event independently. The output may include start and end times, the number of abnormal cycles, the maximum mismatch degree, and the wearing status.

[0077] The event statistics window can be defined by time or the number of respiratory cycles, for example, using a sliding window of 60 to 300 seconds. The first threshold is set at 3 to 5 consecutive abnormal cycles, and the second threshold is set at 5 to 10 cumulative abnormal cycles within the window. When the interval between adjacent events is less than the merging threshold, they can be merged into the same event to avoid splitting the continuous process. The response observation window for secondary events starts from the end of the primary event. The event record saves the start time, end time, number of consecutive and cumulative abnormal cycles, maximum mismatch, and whether there is a post-event response, facilitating review by the user terminal 210.

[0078] Furthermore, when any contact pressure is lower than the lower limit of contact pressure, the processing unit marks the corresponding acquisition section as a contact failure section and outputs a position adjustment prompt; when the contact pressure acquired by at least two spaced contact pressure sensors is continuously higher than the upper limit of pressure safety, the processing unit marks the corresponding acquisition section as a continuous pressure section, outputs a collar too tight prompt, and prohibits the generation of respiratory motion-acoustic response mismatch events using the continuous pressure section.

[0079] Specifically, if the contact pressure at any critical location falls below the lower limit, the flexible coupling seat 12 may be partially suspended, and the processor 100 will stop mismatch calculations and output a position adjustment prompt. If at least two spaced contact pressure sensors 31 continuously exceed the safety upper limit, it indicates that the collar is at risk of being too tight overall, and the processor 100 will output an over-tightness prompt and prevent the generation of mismatch events in the corresponding section. Continuous pressure and effective contact have different meanings: effective contact requires sufficient pressure to maintain coupling, while continuous pressure indicates that multiple locations exceed the safety upper limit for an extended period.

[0080] This results in three processing branches: contact failure, contact success, and continuous pressure, corresponding to position adjustment, normal monitoring, and collar loosening, respectively. After contact failure, position adjustment prompts may include realigning the collar with the ventral side of the neck, checking for hair trapped under the flexible coupling seat 12, and readjusting the collar length. Continuous pressure detection uses at least two interval sensors simultaneously exceeding their upper limits, reducing false alarms caused by single-point protrusions or instantaneous collisions. After contact is restored, the processor 100 waits for the pressure center, left-right pressure difference, and posture to continuously stabilize and reach the recovery time before restarting mismatch calculations. Abnormal pressure sections can still be recorded as wearing status, but cannot be used as evidence of vital sign mismatch.

[0081] Furthermore, the vital signs acquisition component also includes at least one of a heartbeat vibration sensor and a neck temperature sensor; the heartbeat vibration sensor is used to acquire the heartbeat vibration signal of the pet, and the processing unit obtains the heart rate change before and after the respiratory motion-acoustic response mismatch event based on the heartbeat vibration signal; the neck temperature sensor is used to acquire the neck temperature of the pet; the data output unit is used to output at least one of the occurrence time, duration, event level, wearing status, heart rate change and neck temperature of the respiratory motion-acoustic response mismatch event, and save the vital signs evidence data within a preset time period before and after the respiratory motion-acoustic response mismatch event.

[0082] Specifically, the heartbeat vibration sensor 24 can share a piezoelectric sensitive element with the first deformation sensor 21 and obtain the heartbeat component through frequency band separation, or it can use an independent mechanical heartbeat sensor. The processor 100 calculates the heart rate change through peak detection when the signal quality meets the threshold. The neck temperature sensor 25 is located away from the heat source of the electronic compartment 14. The neck temperature output is a trend information information affected by ambient temperature, hair, and contact pressure, and is not equivalent to the core body temperature.

[0083] The memory 120 can use a circular buffer to store short-term vital sign evidence data before and after the event. Heart rate changes and neck temperature are additional outputs and do not participate in the formation or grading of respiratory motion-acoustic response mismatch events. The cardiac vibration signal can be obtained by bandpass filtering and peak detection to obtain the interval between adjacent heartbeats, and the heart rate change before and after the event can be taken as the difference between the average heart rates of two observation windows. If the signal quality is insufficient due to exercise, the processor 100 outputs an invalid heart rate marker instead of forcing calculation. The neck temperature sensor 25 can use a digital temperature element or a thermistor, and simultaneously record the ambient temperature outside the collar to provide background information. Event evidence data can cover 30 to 120 seconds before the event, during the event, and 30 to 120 seconds after the event. During continuous monitoring, only the circular buffered data is retained to balance power consumption, storage size, and data minimization.

[0084] Specific application examples:

[0085] Example 1, Structural Assembly and Initial Calibration: (e.g.) Figures 1 to 3 As shown, the collar strap 10 is made of a low-elongation fabric strap with a width of 25 mm, and the flexible coupling seat 12 is made of elastic silicone, with external dimensions of 30 mm × 22 mm × 4 mm. A first deformation sensor 21 is positioned circumferentially around the collar, a throat acoustic sensor 23 is located in the center of the flexible coupling seat 12, and two contact pressure sensors 31 are positioned on either side of it. A reference deformation sensor 22 is positioned on the side of the collar strap 10, with its sensitive axis aligned with that of the first deformation sensor 21. The first deformation signal and contact pressure are sampled at 200 Hz, posture data at 100 Hz, and throat acoustic response at 12 kHz. During initial wear, 60 seconds of stable pressure are recorded. The lower limit of contact pressure is set to 0.60 times the median of the corresponding pressure, the pressure difference threshold is set to 0.35 times the average reference pressure on both sides, and the pressure center displacement threshold is set to 4 mm. The upper limit of pressure safety is determined by a wearing safety test. After reaching the contact condition and maintaining stability for 5 seconds, the processor 100 begins collecting individual baseline data.

[0086] Example 2: Baseline Establishment and Processing Branch Verification: The processor 100 selects 20 seconds of stable contact data with slight overall collar movement to calculate the common-mode deformation transfer coefficient. A bandpass filter of 0.1 Hz to 3 Hz is applied to the neck respiratory motion characterization signal, and a bandpass filter of 80 Hz to 2000 Hz is applied to the pharyngeal acoustic response. Sixty valid cycles are collected for the same contact pressure and posture group. After removing obvious outlier cycles, a baseline is established, with the 95th percentile of the baseline mismatch used as the initial threshold. Three consecutive abnormal cycles or five cumulative abnormal cycles within 60 seconds constitute a Level 1 event. In the contact failure verification, lifting one side of the flexible coupling seat 12 to lower the corresponding pressure below the lower limit causes the processor 100 to stop mismatch calculation and output a position adjustment prompt. In the motion verification, the head-turning and scratching sections, as well as their preceding and following buffer zones, are excluded. In the continuous pressure verification, when both contact pressure sensors 31 exceed the safety limit for 30 consecutive seconds, an over-tightening prompt is output. The above embodiments are used to demonstrate that contact screening, common-mode separation, baseline comparison, and safety branching can be implemented according to the recorded input-output relationships, and are not intended as proof of medical diagnostic performance.

[0087] Example 3: Bench verification of common-mode separation and cross-modal link: An adjustable-diameter elastic neck model was used to simulate a pet's neck 300. A periodic micro-displacement actuator was placed on the ventral side of the model, and a low-frequency stretching actuator was placed circumferentially on the collar strap 10. The periodic micro-displacement simulated local respiratory movements of the neck, and the low-frequency stretching simulated the overall movement of the collar. The first deformation sensor 21 responded to both actuators simultaneously, while the reference deformation sensor 22 mainly responded to the low-frequency stretching. The processor 100 calculated the common-mode deformation component based on the reference deformation signal. The bench executed three operating conditions sequentially. In the first operating condition, only the low-frequency stretching was activated to check whether the residual deformation after common-mode separation was lower than the signal before separation. In the second operating condition, both periodic micro-displacement and low-frequency stretching were activated simultaneously to check whether the periodic boundary after separation was consistent with the micro-displacement drive period. In the third operating condition, the micro-displacement amplitude was kept constant and the acoustic excitation was reduced to check whether the cross-modal characteristics and mismatch were increased relative to the baseline. This verification was used to confirm the engineering causal relationship between the reference channel, common-mode separation, cycle-by-cycle pairing, and mismatch calculation, and was not used to prove medical diagnostic performance.

[0088] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart pet collar integrating vital sign data collection, characterized in that, It includes a collar strap, and a vital sign acquisition component, a wearing status acquisition component, a posture acquisition component, a processing unit, and a data output unit set on the collar strap; The vital signs acquisition component includes a first deformation sensor, a reference deformation sensor, and a pharyngeal acoustic sensor disposed on the collar. The first deformation sensor is used to acquire a first deformation signal including the pet's neck respiratory movement and the overall movement of the collar. The reference deformation sensor is used to acquire a reference deformation signal characterizing the overall movement of the collar. The pharyngeal acoustic sensor is used to acquire tissue sound transmission and airway vibration in the pet's pharyngeal region and form a pharyngeal acoustic response signal. The wearing status acquisition component includes multiple contact pressure sensors for acquiring the contact pressure and pressure distribution between the collar strap and the pet's neck; The posture acquisition component is used to acquire the pet's movement status and neck posture; The processing unit is connected to the vital sign acquisition component, the wearing status acquisition component, the posture acquisition component, and the data output unit, respectively, and is used for: The effective monitoring segment is determined based on the contact pressure, pressure distribution, motion state, and neck posture. Based on the first deformation signal and the reference deformation signal, a common-mode deformation component related to the overall motion of the collar is separated from the first deformation signal to obtain a neck respiratory motion characterization signal. Respiratory cycles are divided within the effective monitoring segment, and the neck respiratory motion characterization signal within the same respiratory cycle is time-paired with the pharyngeal acoustic response signal. Based on the degree of deviation between the time-paired neck respiratory motion characteristics and pharyngeal acoustic response characteristics and the corresponding baseline of the individual pet, a respiratory motion-acoustic response mismatch event is identified. The respiratory motion-acoustic response mismatch event and corresponding wearing status information are output through the data output unit.

2. The pet smart collar with integrated vital sign collection according to claim 1, characterized in that, The collar includes a respiratory sensing segment that contacts the area of ​​the ventral trachea on the pet's neck and a common-mode reference segment that is spaced apart from the respiratory sensing segment. The respiratory sensing segment is provided with a flexible coupling seat, the first deformation sensor and the pharyngeal acoustic sensor are disposed on the flexible coupling seat, and the reference deformation sensor is disposed on the common mode reference segment; The first deformation sensor and the reference deformation sensor have the same deformation-sensitive direction, and the common-mode reference segment is offset from the corresponding area of ​​the ventral trachea of ​​the pet's neck, so that the local respiratory motion component in the reference deformation signal is smaller than the local respiratory motion component in the first deformation signal.

3. The pet smart collar with integrated vital sign collection according to claim 2, characterized in that, At least one contact pressure sensor is provided on each side of the flexible coupling seat; When the contact pressure on both sides of the flexible coupling seat reaches the lower limit of the contact pressure, the difference between the contact pressures on both sides does not exceed the pressure difference threshold, the displacement of the pressure center does not exceed the displacement threshold, and each contact pressure does not exceed the upper limit of the pressure safety limit, the processing unit determines the corresponding collection section as the effective contact section. The processing unit identifies the effective monitoring segment as the segment in the effective contact area where the pet's movement is below the movement threshold and the neck posture remains stable.

4. The pet smart collar with integrated vital sign collection according to claim 1, characterized in that, The processing unit selects a common-mode calibration section from the acquisition section where the contact state is stable and the attitude acquisition component detects the overall movement of the collar strap; The processing unit determines the common-mode deformation transfer coefficient by minimizing the common-mode separation residual based on the correspondence between the first deformation signal and the reference deformation signal within the common-mode calibration section, and subtracts the common-mode deformation component corresponding to the reference deformation signal from the first deformation signal based on the common-mode deformation transfer coefficient.

5. The pet smart collar with integrated vital sign collection according to claim 1, characterized in that, The attitude acquisition component includes an accelerometer and an angular velocity sensor disposed on the collar strap; The processing unit identifies at least one action among rolling over, raising head, walking, scratching, licking, and swallowing based on changes in acceleration, angular velocity, and contact pressure, and excludes the segment in which the identified action occurs and the preset buffer zone before and after the action from the effective monitoring segment.

6. The smart pet collar with integrated vital sign collection according to claim 1, characterized in that, The processing unit is configured to: determine successive respiratory cycles based on the periodic boundaries of the neck respiratory motion representation signal, and extract the neck motion amplitude from the i-th respiratory cycle. ; Extract acoustic response energy from the pharyngeal acoustic response signal corresponding to the i-th respiratory cycle time. Acoustic response time difference relative to neck respiratory movement and acoustic response continuity Select at least a preset number of effective respiratory cycles from the same contact pressure range and neck posture range as baseline respiratory cycles, and establish the corresponding baseline for the individual pet based on the baseline respiratory cycles; when the pet smart collar is re-worn or the pressure center of the flexible coupling seat continuously exceeds the baseline update threshold, re-establish or update the corresponding baseline for the individual pet.

7. The smart pet collar with integrated vital sign collection according to claim 6, characterized in that, For the i-th respiratory cycle, the processing unit uses the cross-modal feature vector Calculate the degree of mismatch : In the formula, and These are the baseline values ​​for acoustic response energy and neck movement amplitude for the corresponding individual pet. Let be the duration of the i-th respiratory cycle. and , respectively, are the mean and standard deviation of the j-th cross-modal feature in the baseline for the corresponding pet individual. Let the weights of the j-th cross-modal feature satisfy the condition that the sum of all weights is 1 and Not less than 0, A constant that is greater than zero; When the mismatch If the mismatch threshold is greater than the mismatch distribution determined based on the baseline respiratory cycle, the i-th respiratory cycle is identified as an abnormal mismatch cycle.

8. The smart pet collar with integrated vital sign collection according to claim 7, characterized in that, The processing unit counts the consecutive and cumulative number of abnormal mismatch cycles within the event statistics window; When the number of consecutive abnormal mismatch cycles reaches the first threshold, or the cumulative number of abnormal mismatch cycles within the event statistics window reaches the second threshold, a first-level respiratory motion-acoustic response mismatch event is generated. When a first-level respiratory motion-acoustic response mismatch event is followed by at least one of the following events: increased respiratory amplitude, change in respiratory rate, head lifting, turning over, or change from rest to activity, the first-level respiratory motion-acoustic response mismatch event is marked as a second-level respiratory motion-acoustic response mismatch event.

9. The smart pet collar with integrated vital sign collection according to claim 3, characterized in that, When any contact pressure is lower than the lower limit of the contact pressure, the processing unit marks the corresponding collection section as a contact failure section and outputs a position adjustment prompt; When the contact pressure collected by at least two spaced contact pressure sensors is continuously higher than the pressure safety limit, the processing unit marks the corresponding collection section as a continuous pressure section, outputs a collar too tight warning, and prohibits the generation of the respiratory motion-acoustic response mismatch event using the continuous pressure section.

10. The smart pet collar with integrated vital sign collection according to claim 1, characterized in that, The vital signs acquisition component also includes at least one of a heart rate vibration sensor and a neck temperature sensor disposed in the collar body; The heartbeat vibration sensor is used to collect the heartbeat vibration signal of the pet, and the processing unit obtains the heart rate change before and after the respiratory motion-acoustic response mismatch event based on the heartbeat vibration signal. The neck temperature sensor is used to collect the pet's neck temperature. The data output unit is used to output at least one of the following: the occurrence time, duration, event level, wearing status, heart rate change, and neck temperature of the respiratory motion-acoustic response mismatch event, and to save the vital signs evidence data within a preset time period before and after the occurrence of the respiratory motion-acoustic response mismatch event.

Images