State-gated pet vital sign collection and prompting method and system

Abstract

This invention discloses a state-gated method and system for collecting and alerting pet vital signs. The method acquires motion state data collected by an inertial measurement module, determines whether the pet has entered a low-motion candidate state and whether the stability of the monitoring host meets the requirements, and determines the positioning scene state. When the low-motion candidate state persists for a preset time and the device is worn stably, the millimeter-wave radar module is controlled to collect micro-motion signals of the chest cavity in the area under the pet's neck or chest, obtaining respiratory rate, heart rate, or their trend data. The system scores the acquisition window based on the motion state data and radar signal quality parameters during the acquisition period, using only acquisition windows that meet the quality requirements to update the pet's individualized baseline or compare with the baseline, and outputs non-diagnostic health alerts in conjunction with the positioning scene state, thereby reducing power consumption and motion artifacts.

Description

Technical Field

[0001] This invention relates to the fields of pet wearable devices, millimeter-wave radar vital sign acquisition, low-power sensor scheduling, location scene recognition, edge computing, and health alerts. In particular, it relates to a method, system, and collar for state gating based on an inertial measurement module and for waking up the millimeter-wave radar module on demand to collect pet vital sign data. Background Technology

[0002] Smart pet collars are limited by size, weight, battery capacity, wearing comfort, and the pet's activity status when worn for extended periods. Existing pet collars primarily function as location tracking, electronic fences, activity logs, or basic reminders, making it difficult to reliably acquire vital signs data related to health status, such as resting respiratory rate and resting heart rate, under daily, long-term wear conditions.

[0003] Millimeter-wave radar can detect subtle chest movements non-contactly and is less affected by fur and skin color compared to optical heart rate sensors. However, pets produce noticeable motion artifacts when running, jumping, rolling over, scratching, or bumping into things with their collars. If all radar data is continuously collected or used directly without distinguishing the pet's state, it is easy to obtain results with low reliability.

[0004] In addition, both millimeter-wave radar and GNSS positioning modules increase power consumption. If the radar is continuously on for extended periods and the GNSS is continuously sampling at high frequencies, the battery life of small pet collars will be significantly shortened.

[0005] Therefore, there is a need for a method that utilizes a low-power inertial measurement module to first determine low motion and wearing stability window, then wakes up millimeter-wave radar as needed, and outputs non-diagnostic health prompts by combining radar signal quality, pet individualized baseline and positioning scenario. Summary of the Invention

[0006] This invention provides a non-contact method for collecting vital signs and providing health alerts for pets based on state gating, applied to a monitoring host worn on a pet collar. The monitoring host may include an inertial measurement module, a millimeter-wave radar module, a GNSS positioning module, a main control module, a power supply module, and a wireless communication module.

[0007] This method acquires motion state data collected by the inertial measurement module, determines whether the pet has entered a low-motion candidate state, and further distinguishes the low-motion candidate state into a roughly stationary window during the day or in daily scenarios, and a higher stability window during nighttime sleep or in continuously stable scenarios. Simultaneously, it determines whether the stability of the monitoring host meets preset conditions. The system can also acquire or determine the pet's location-related scenario status, such as daily status, outing status, lost status, or unknown status.

[0008] When the low-motion candidate state continues for a first preset time and the wearing stability meets the preset conditions, the main control module controls the millimeter-wave radar module to enter the acquisition state from the low-power state, and acquires the chest cavity micro-motion signal under the pet's neck or chest area within a second preset time, and simultaneously records the motion state data within the acquisition window.

[0009] In one embodiment, the first preset time can be from 30 seconds to 120 seconds, and the second preset time can be from 20 seconds to 90 seconds; the specific values ​​can be configured according to the pet's size, activity habits, radar module response time, and endurance target.

[0010] The main control module or the processing unit connected to it obtains respiratory rate, heart rate, or their trend data based on the chest cavity micromotion signal, and at least scores the quality of the acquisition window based on the motion state data and the signal quality parameters of the chest cavity micromotion signal within the acquisition window. Acquisition windows with a quality score below the threshold are marked as invalid or low confidence.

[0011] The system only uses data from collection windows that meet quality requirements to update the pet's individualized baseline or compare it with the individualized baseline. When valid data deviates continuously from the individualized baseline, the system outputs non-diagnostic health alerts to the user terminal based on the location scenario status. Beneficial effects

[0012] By using IMU low-power listening and millimeter-wave radar on-demand wake-up, the continuous working time of the radar can be reduced, extending the battery life of the pet collar.

[0013] By using low-motion candidate states, wearing stability assessment, and acquisition window quality scoring, motion artifacts caused by pet movement, rolling over, scratching, and collar collisions can be reduced.

[0014] By establishing and updating individualized baselines for pets using only acquisition windows that meet quality requirements, fixed threshold errors caused by different body sizes, ages, breeds, and wearing habits can be reduced.

[0015] By combining GNSS positioning scenario status, different sampling and prompting strategies can be adopted in daily, outing, and lost states, taking into account the needs of battery life, positioning, and health prompts.

[0016] The above-mentioned technical effects are the effects that can be achieved or are expected to be improved based on the technical solution of the present invention. The specific improvement range can be further recorded and verified through subsequent prototype testing, simulation or control experiments. Attached Figure Description

[0017] Figure 1 This is the overall flowchart of the method of the present invention.

[0018] Figure 2This is a schematic diagram of the system architecture of the present invention.

[0019] Figure 3 This is a schematic diagram of the pet collar structure and radar / GNSS orientation of the present invention.

[0020] Figure 4 This is a flowchart of the quality scoring process for the data acquisition window in this invention. Detailed Implementation

[0021] In one system embodiment, the system may include a motion state acquisition unit, a state gating unit, a scene determination unit, a radar control unit, a vital sign extraction unit, a quality scoring unit, a baseline processing unit, and a prompt output unit. The motion state acquisition unit acquires motion state data collected by the inertial measurement module; the state gating unit determines low-motion candidate states and wearing stability; the scene determination unit determines the positioning scene state; the radar control unit controls the millimeter-wave radar module to enter the acquisition state; the vital sign extraction unit obtains respiratory rate, heart rate, or their trend data; the quality scoring unit scores the acquisition window; the baseline processing unit updates the individualized baseline or performs baseline comparisons; and the prompt output unit outputs non-diagnostic health prompts.

[0022] In one embodiment, the monitoring host is mounted on a pet collar. The monitoring host includes a main control module, an inertial measurement module, a millimeter-wave radar module, a GNSS positioning module, a power supply module, and a wireless communication module. The millimeter-wave radar module can be a 60GHz millimeter-wave radar module, or a millimeter-wave radar module suitable for detecting chest cavity micro-movements in the 57GHz to 81GHz range. The inertial measurement module can be a six-axis IMU, or a low-power accelerometer, gyroscope, or a combination thereof. The GNSS positioning module can support GPS, BeiDou, GLONASS, Galileo, or a combination thereof. The wireless communication module can be Bluetooth, Wi-Fi, cellular communication, or a combination thereof.

[0023] The monitoring unit can be mounted on the collar, with the millimeter-wave radar module facing the pet's neck or chest area. A millimeter-wave transmission window can be installed on the side of the housing closest to the pet's body. This window can be made of plastic, PC, ABS, silicone, TPU, or other materials suitable for millimeter-wave transmission and pet contact. The GNSS antenna can be positioned on or near the outside of the monitoring unit to minimize obstruction from the pet's body.

[0024] In the daytime routine monitoring embodiment, the inertial measurement module continuously or intermittently acquires acceleration and angular velocity data. When the pet is in a roughly stationary state such as lying down, sitting, or resting briefly, and the changes in acceleration amplitude and angular velocity persist for a first preset time within a relatively wide threshold range, the system can mark it as a first resting candidate state. In this state, shorter or lower frequency radar sampling can be used to obtain daily resting trend data.

[0025] In the nighttime sleep data collection embodiment, when both the acceleration amplitude change and angular velocity change are within a more stringent preset threshold range and remain so for 60 seconds, the system determines that the pet has entered a second resting candidate state or a sleep candidate state. If no collar collision, flipping, or significant posture change is detected within this time, the wearing stability is determined to meet the preset conditions. The main control module wakes up the millimeter-wave radar module to collect 60 seconds of chest cavity micro-motion signals and simultaneously records the motion state data within these 60 seconds.

[0026] In the acquisition window quality scoring embodiment, the system constructs a quality score based on the signal quality indicators output by the millimeter-wave radar module, radar echo intensity, target range gate stability, waveform missing rate, respiratory band energy, heartbeat band energy, spectral stability, periodicity index, signal-to-noise ratio, and the degree of IMU interference during acquisition. If the pet rolls over, scratches, runs, jumps, or bumps into its collar during acquisition, the acquisition window is marked as invalid or low confidence and is not used to update the individualized baseline.

[0027] In the vital sign extraction embodiment, the system can perform bandpass filtering, frequency domain analysis, time-frequency analysis, mode decomposition, adaptive filtering, peak detection, or a combination thereof on the chest cavity micro-motion signal to obtain respiratory rate, heart rate, or their trend data. The specific algorithm can be configured according to the data type of the hardware module output and the pet's body size.

[0028] In the individualized baseline implementation, the system establishes an initial individualized baseline for the pet within 3 days of initial use based on multiple acquisition windows that meet quality requirements, and continuously refines it as new valid acquisition windows are added during subsequent use. The baseline may include a resting respiratory rate baseline, a resting heart rate baseline, a heart rate variability trend, a sleep duration baseline, an activity intensity baseline, or a location activity range baseline.

[0029] In the location scenario implementation, the system determines the daily status, out-of-town status, or missing status based on GNSS positioning data, electronic fence status, wireless communication connection status, or user-defined status. In the daily status, the system can reduce the GNSS sampling frequency and wireless communication reporting frequency; in the out-of-town or missing status, the system can increase the GNSS sampling frequency and reporting frequency, and adjust the display method of health prompts.

[0030] In the health alert implementation, when respiratory rate, heart rate, or trend data consistently deviate from the pet's individualized baseline across multiple consecutive data collection windows that meet quality requirements, and this deviation cannot be explained by activity level or wear instability, the system outputs a non-diagnostic health alert to the user terminal. This alert may include the status level, confidence level, deviation item, main basis, observation suggestions, care suggestions, or veterinary advice. The above alerts are for daily health observation and care assistance and are not intended as disease diagnosis or treatment conclusions.

[0031] In the large language model interpretation embodiment, the system can input structured monitoring results, pet individualized baselines, recent effective data collection window statistics, and location scene status into the large language model, which will then generate user-oriented explanatory text or care suggestions. The large language model is only used for interpretation and suggestion generation and does not directly replace data collection window quality scores and individualized baseline comparisons.

[0032] In other alternative implementations, the monitoring host may also include a microphone module for collecting pet sounds or ambient sounds and using sound events as supplementary information for anomaly alerts. The microphone module is not a necessary component for the status gating and millimeter-wave vital sign acquisition of this invention.

Claims

1. A non-contact pet vital sign collection and health alert method based on state gating, applied to a monitoring host worn on a pet collar, the monitoring host comprising an inertial measurement module, a millimeter-wave radar module, a GNSS positioning module, a main control module, and a wireless communication module, characterized in that, The method includes: S1, acquiring motion state data collected by the inertial measurement module, and determining whether the pet has entered a low motion candidate state based on the motion state data; S2, determining whether the wearing stability of the monitoring host meets preset conditions based on the motion state data; S3, acquiring or determining the positioning scene state related to the pet, wherein the positioning scene state includes at least one of daily state, outing state, lost state, or unknown state; S4, when the low motion candidate state continues for a first preset time and the wearing stability meets the preset conditions, controlling the millimeter-wave radar module to enter the acquisition state from a low-power state; S5, acquiring the pet's neck within a second preset time. S6. Collect chest cavity micromotion signals in the lower or frontal region and simultaneously record motion state data within the second preset time period; S7. Obtain respiratory rate, heart rate, or their trend data based on the chest cavity micromotion signals; S8. At least based on the motion state data within the second preset time period and the signal quality parameters of the chest cavity micromotion signals, perform a quality score on the current acquisition window, and mark acquisition windows with quality scores below a threshold as invalid or low confidence; S9. Only use the respiratory rate, heart rate, or their trend data corresponding to acquisition windows that meet the quality requirements to update the pet's individualized baseline or compare with the pet's individualized baseline, and output non-diagnostic health prompts to the user terminal in conjunction with the positioning scene status.

2. The method according to claim 1, characterized in that, The low-motion candidate states include a first resting candidate state and a second resting candidate state; the first resting candidate state is used for roughly stationary monitoring in daytime or daily scenarios, and the second resting candidate state is used for higher stability monitoring in nighttime sleep or continuously stable scenarios.

3. The method according to claim 1, characterized in that, The wearing stability meets the preset conditions, including: the acceleration amplitude and / or angular velocity change of the inertial measurement module are within the preset threshold range, or the monitoring host does not experience flipping, collision or large attitude change within a preset time.

4. The method according to claim 1, characterized in that, The signal quality parameters include at least one of the following: signal quality index output by the millimeter-wave radar module, radar echo intensity, target range gate stability, waveform missing rate, respiratory band energy, heartbeat band energy, spectral stability, periodicity index, or signal-to-noise ratio.

5. The method according to claim 1, characterized in that, When the motion status data within the second preset time period indicates that the pet is rolling over, running, jumping, scratching, colliding with its collar, or undergoing a significant change in posture, the corresponding collection window will be marked as invalid or low confidence.

6. The method according to claim 1, characterized in that, Step S6 includes performing bandpass filtering, frequency domain analysis, time-frequency analysis, mode decomposition, adaptive filtering, peak detection, or a combination thereof on the thoracic cavity micro-motion signal to obtain respiratory rate, heart rate, or their trend data.

7. The method according to claim 1, characterized in that, The pet individualization baseline is established within 3 days after the first use based on multiple collection windows that meet the quality requirements, and is continuously updated according to the rolling window during subsequent use; The individualized baseline for pets includes at least one of the following: resting respiratory rate baseline, resting heart rate baseline, heart rate variability trend, sleep duration baseline, activity intensity baseline, or location activity range baseline.

8. The method according to claim 1, characterized in that, The positioning scenario state is determined based on the location data collected by the GNSS positioning module, the electronic fence status, the wireless communication connection status, the user-defined status, or a combination thereof; the GNSS sampling frequency is reduced in normal conditions, and the GNSS sampling frequency and / or wireless communication reporting frequency is increased in the out-of-town or lost conditions.

9. The method according to claim 1, characterized in that, The non-diagnostic health alerts include at least one of the following: status level, confidence level, deviation items, main basis, observation suggestions, care suggestions, or veterinary suggestions, and the non-diagnostic health alerts are not considered as disease diagnoses or treatment conclusions; it also includes inputting the non-diagnostic health alerts, the pet's individualized baseline, and the statistical results of the recent effective data collection window into a large language model, which generates user-oriented explanatory text or care suggestions, and the large language model does not directly replace the quality score in step S7 and the baseline comparison in step S8.

10. A non-contact pet vital sign collection and health alert system based on state gating, characterized in that, include: The motion state acquisition unit is used to acquire motion state data collected by the inertial measurement module. The state gating unit is used to determine whether the pet has entered a low-motor candidate state and whether the wearing stability meets the preset conditions. Scene determination unit, used to acquire or determine the state of the location scene; The radar control unit is used to control the millimeter-wave radar module to enter the acquisition state when the low motion candidate state continues for a first preset time and the wearing stability meets the preset conditions. The vital signs extraction unit is used to obtain respiratory rate, heart rate, or their trend data based on chest cavity micromotion signals; The quality scoring unit is used to score the quality of the acquisition window based on the signal quality parameters of the motion state data and the chest cavity micromotion signal during the acquisition period. The baseline processing unit is used to update the individualized pet baseline or compare it with the individualized pet baseline based on the acquisition window that meets the quality requirements; The prompt output unit is used to output non-diagnostic health prompts.

Images