Animal health monitoring system

Abstract

The present application relates to Internet of Things technology, and discloses an animal health monitoring system, comprising: a wearable device and a cloud server. The wearable device comprises a processor, and an acceleration sensor, a body temperature sensor, and a wireless transceiver connected to the processor. The processor is configured to: determine the posture of an animal on the basis of acceleration data detected by the acceleration sensor, while acquiring the body temperature of the animal from the body temperature sensor; and transmit posture-body temperature data pairs to the cloud server by means of the wireless transceiver. The cloud server is configured to: input multiple sets of posture-body temperature data pairs from the wearable device, arranged in chronological order, into a pre-trained first machine learning model, so as to obtain an animal health status outputted by the first machine learning model. The system not only reduces transmission energy consumption and extends battery life, but also improves the accuracy of animal health status determination.

Description

Animal health monitoring system Technical Field

[0001] This application relates to Internet of Things (IoT) technology, and in particular to the use of IoT to monitor animal health. Background Technology

[0002] This section is intended to provide background or context for understanding the implementation of this application and is for reference only. It should not be construed as an admission by the applicant that this section pertains to prior art that was disclosed before the filing date of this application.

[0003] In livestock production, monitoring animal health is crucial for improving farming efficiency. Currently, the Internet of Things (IoT) technology is widely used in animal health monitoring. A common practice is to equip animals with wearable devices that collect physiological data through various sensors and transmit the data to the cloud for analysis.

[0004] For example, body temperature sensors can collect an animal's body temperature data, and accelerometers can collect its movement data. The raw data collected by these sensors is transmitted wirelessly (such as via Bluetooth or long-range radio communication) to a cloud server for analysis and processing to monitor the animal's health.

[0005] However, this monitoring method has the following problems: First, because it requires frequent transmission of large amounts of raw data to the cloud, wearable devices consume a lot of power and have a short standby time, typically only a few days to one or two weeks. This clearly cannot meet the needs of long-term health monitoring of animals in livestock production. Second, simply transmitting single data points such as body temperature or movement data to the cloud for analysis cannot comprehensively reflect the animal's health status and is prone to misjudgment. Summary of the Invention

[0006] The purpose of this application is to provide an animal health monitoring system that reduces transmission energy consumption, extends battery life, and improves the accuracy of animal health status assessment.

[0007] This application discloses an animal health monitoring system, including: a wearable device and a cloud server;

[0008] The wearable device includes a processor, and an accelerometer, a body temperature sensor, and a wireless transceiver connected to the processor; the processor is configured to: determine the animal's posture based on acceleration data detected by the accelerometer, and simultaneously acquire the animal's body temperature from the body temperature sensor; and transmit the posture-body temperature data pair to the cloud server via the wireless transceiver.

[0009] The cloud server is configured to input multiple sets of posture-body temperature data pairs from the wearable device, sorted chronologically, into a pre-trained first machine learning model to obtain the animal health status output by the first machine learning model.

[0010] In a preferred embodiment, there are multiple wearable devices;

[0011] The first machine learning model was trained based on posture-temperature data from multiple wearable devices.

[0012] In a preferred embodiment, the wearable device further includes a memory; the processor is configured to store the posture-body temperature data pairs in the memory and transmit multiple sets of the posture-body temperature data pairs stored in the memory to the cloud server via the wireless transceiver at a predetermined time.

[0013] In a preferred embodiment, the processor is further configured to: determine the hibernation duration based on the animal's current posture, control the wearable device to enter a hibernation state, wake up the wearable device after the hibernation duration, and perform the next detection, wherein at least two different postures correspond to different hibernation durations.

[0014] In a preferred embodiment, the wearable device further includes a gyroscope and a magnetometer;

[0015] The processor is also configured to determine the animal's posture based on data detected by the accelerometer, the gyroscope, and the magnetometer.

[0016] In a preferred embodiment, the processor uses a second machine learning model in the wearable device to determine the animal's posture based on acceleration data detected by the accelerometer.

[0017] In a preferred embodiment, the processor is further configured to dynamically adjust the sampling frequencies of the accelerometer, the gyroscope, the magnetometer, and the body temperature sensor based on the determined attitude.

[0018] In a preferred embodiment, the system further includes environmental parameter sensors arranged in the animal rearing environment for measuring environmental parameters of the environment in which the animal is located;

[0019] The environmental parameter sensor periodically uploads environmental parameters to the cloud server;

[0020] The wearable device also includes a clock, and the processor is further configured to obtain time information from the clock to determine the posture and obtain the body temperature, form a timestamp, and send the posture-body temperature data pair, the corresponding timestamp, and the identifier of the wearable device to the cloud server through the wireless transceiver;

[0021] The cloud server determines the environmental parameter sensors of the environment in which the wearable device is located based on the identifier of the wearable device, and determines the environmental parameters corresponding to the posture-body temperature data pair based on the timestamp; the posture-body temperature data pair and its corresponding environmental parameters are input into the first machine learning model to obtain the animal health status output by the first machine learning model.

[0022] In a preferred embodiment, the environmental parameters include one or any combination of the following: ambient temperature, ambient humidity, and ambient air pressure;

[0023] The environmental parameter sensor includes one or any combination of the following:

[0024] Temperature sensors for measuring ambient temperature, humidity sensors for measuring ambient humidity, and air pressure sensors for measuring ambient air pressure.

[0025] In a preferred embodiment, a wireless gateway device is also included, which establishes a wireless connection with the plurality of wearable devices and the plurality of environmental parameter sensors, and uploads data from the wearable devices and the environmental parameter sensors to the cloud server.

[0026] In this embodiment, by integrating an accelerometer and a body temperature sensor into a wearable device, and having a processor determine the animal's posture based on acceleration data while simultaneously acquiring its body temperature, the posture-body temperature data pair is uploaded to a cloud server to be input into a pre-trained first machine learning model for health status prediction. This enables precise monitoring and intelligent analysis of the animal's health status. This technical solution significantly reduces the massive transmission of raw data and the burden of cloud processing. By generating posture-body temperature data pairs at the terminal, it not only reduces device power consumption and extends standby time but also continuously improves the accuracy and real-time performance of health assessments through the continuous learning of the cloud model.

[0027] Furthermore, by using posture-temperature data from multiple wearable devices for joint training, the scale of training data can be effectively expanded, the generalization ability and prediction accuracy of machine learning models can be improved, and thus more accurate analysis of animal health status at the group level can be performed.

[0028] Furthermore, by incorporating a memory into wearable devices and periodically uploading posture-temperature data pairs in batches, energy consumption caused by frequent data transmission can be reduced, the battery life of wearable devices can be improved, and reliable operation can be ensured for a long time under limited power conditions.

[0029] Furthermore, by dynamically determining the sleep duration of wearable devices based on the animal's current posture, allowing the devices to enter sleep mode in a low-activity state, thereby reducing unnecessary measurements and transmissions, energy consumption can be significantly reduced and the device's working time extended.

[0030] Furthermore, by integrating accelerometers, gyroscopes, and magnetometers into wearable devices and having a processor fuse these data to determine animal posture, the accuracy and robustness of posture recognition can be improved, reducing judgment bias caused by the limitations of a single sensor.

[0031] Furthermore, by deploying a second machine learning model on the wearable device to perform attitude recognition on the acceleration data, data can be processed quickly at the front end, reducing reliance on cloud computing resources and lowering the frequency of raw data uploads, thereby achieving more efficient edge computing and energy saving.

[0032] Furthermore, by dynamically adjusting the sampling frequencies of the accelerometer, gyroscope, magnetometer, and body temperature sensor according to the determined attitude, different attitudes can correspond to different sensor sampling frequencies. This can improve data sampling accuracy when critical behaviors occur, reduce the sampling rate to save power when there are no significant attitude changes, and optimize the overall monitoring strategy.

[0033] Furthermore, by incorporating environmental parameter sensors into the system and combining environmental temperature, humidity, or air pressure with posture-body temperature data pairs to form timestamp-correlated data input into the cloud model, a more comprehensive health status analysis can be obtained, improving the accuracy and adaptability of predictions. Through multi-dimensional correlation analysis of environmental parameters such as temperature, humidity, and air pressure with animal posture-body temperature data, external causes of animal health abnormalities can be identified more accurately, thereby enabling more refined health risk prediction and intervention strategies.

[0034] The various technical features disclosed in the above-described invention, the various technical features disclosed in the following embodiments and examples, and the various technical features disclosed in the accompanying drawings can be freely combined to form various new technical solutions (all of which should be considered as having been recorded in this specification), unless such a combination of technical features is technically infeasible. For example, in one example, feature A+B+C is disclosed, and in another example, feature A+B+D+E is disclosed. Features C and D are equivalent technical means that serve the same function, and technically only one needs to be used; it is impossible to use both simultaneously. Feature E can be technically combined with feature C. Therefore, the solution A+B+C+D should not be considered as having been recorded because it is technically infeasible, while the solution A+B+C+E should be considered as having been recorded. Attached Figure Description

[0035] Figure 1 is a schematic diagram of an animal health monitoring system according to an embodiment of this application. Detailed Implementation

[0036] In the following description, many technical details are presented to help the reader better understand this application. However, those skilled in the art will understand that the technical solutions claimed in this application can be implemented even without these technical details and various variations and modifications based on the following embodiments.

[0037] Explanation of some concepts:

[0038] Wearable devices are electronic devices that can be worn on animals. They integrate various sensors and processors to collect and process the animal's physiological and behavioral data and transmit the data to external systems via wireless communication. They are typically characterized by their small size, light weight, and low power consumption.

[0039] A machine learning model is a computer-implemented model that can autonomously learn patterns and relationships from historical data to make predictions or decisions on new data without explicitly programming specific rules.

[0040] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0041] This application relates to an animal health monitoring system, comprising: wearable devices and a cloud server. Typically, there are multiple wearable devices, each worn on an animal. For example, the wearable device can be an ear tag, attached to the ear of a pig, cow, or sheep. Other types can also be used, such as collar-type wearable devices, harness-type wearable devices, ankle bracelet-type wearable devices, etc. The cloud server can be a computer on the Internet, a computer cluster, or a virtual host, etc. Figure 1 is a schematic diagram of a typical example of an animal health monitoring system.

[0042] The wearable device includes a processor, and an accelerometer, a body temperature sensor, and a wireless transceiver connected to the processor. The processor is configured to determine the animal's posture based on acceleration data detected by the accelerometer, and simultaneously acquire the animal's body temperature from the body temperature sensor. The posture-body temperature data pair is transmitted to a cloud server via the wireless transceiver. The wireless transceiver can be Bluetooth, Wi-Fi, Zigbee, or other low-power wireless communication devices.

[0043] The cloud server is configured to input multiple sets of posture-temperature data pairs from wearable devices, ordered chronologically, into a pre-trained first machine learning model to obtain the animal's health status output by the model. The animal's health status can be, for example, healthy, slightly unwell, or sick. Users can access the cloud server using their terminal devices to obtain the animal's health status. The cloud server can also proactively send notification messages to the user's terminal when it detects a possibility that the animal is sick.

[0044] The applicant found that an animal's posture is strongly correlated with its mental and physical health.

[0045] Taking pigs as an example, the following factors relate to their posture and mental and health status:

[0046] (1) Pigs sitting in a dog-sitting posture indicate difficulty breathing, which is often associated with pneumonia, heart failure, pleurisy, or anemia.

[0047] (2) When standing, the head and neck are stretched forward, which also indicates difficulty breathing. Pigs with pleurisy usually stand with their backs arched.

[0048] (3) Lame pigs are usually unwilling to stand.

[0049] (4) Head and neck tilting or circular motion is commonly seen in otitis media, otitis media, meningitis, and brain abscess. The direction of head and neck bending or circular motion is centered on the affected side.

[0050] (6) Healthy pigs always have their ears erect or extended forward. If the ears droop or stick back, it indicates that the pig is not in good spirits. Running around randomly and not responding to external sounds indicates that the pig may be deaf or blind.

[0051] Even a pig's sleeping posture can reflect its health condition; pigs typically sleep on their side or stomach.

[0052] 1. The side-lying position is the most common sleeping position of pigs. Pigs usually lie on one side with their heads resting on their front or hind hooves.

[0053] 2. The prone position is a more relaxed posture for pigs when they sleep. They will stretch out their heads and front hooves and lie on the ground.

[0054] 3. Of course, in addition to these two positions, pigs also have some other sleeping positions that are not common among pigs, such as lying on their backs and lying on their stomachs. Different pigs may have different sleeping habits, but in general, lying on their sides and lying on their stomachs are their favorite sleeping positions.

[0055] Sick pigs often lie flat on the ground, appearing uncomfortable. Some pigs are even unable to sleep due to discomfort and will sit like dogs – this is the familiar "dog sit" to pig farmers. Of course, you can't determine if a pig is sick just by looking at its sleeping posture; you also need to observe its overall condition.

[0056] When healthy pigs sleep, their muscles are relaxed and their breathing rhythm is even; however, this is not the case for sick pigs. Sick pigs usually feel uncomfortable when they sleep. They will lie down exhausted and often exhibit abnormal behaviors such as difficulty breathing, panting, coughing, and sitting like a dog.

[0057] This application utilizes pre-training and pre-learning to make preliminary judgments about animal postures within wearable devices. This not only reduces the need for large-scale data storage and raw data uploads but also avoids data alterations caused by changes in sensors or the environment. Therefore, using these postures to aid in the learning of future large-scale models and improve the accuracy of animal health monitoring and judgment will be of great significance.

[0058] The working principle of an accelerometer is rooted in Newton's second law, which states that the acceleration of an object is directly proportional to the force acting upon it. In actual operation, an accelerometer can sensitively detect subtle changes in the acceleration of an object during its motion and accurately convert this abstract physical quantity into a recognizable and processable electrical signal for output, thereby achieving effective monitoring and in-depth analysis of the object's motion state. Furthermore, accelerometers possess strong collaborative capabilities; they can work in conjunction with other types of sensors, such as gyroscopes and magnetometers, using advanced angle fusion algorithms to accurately calculate attitude angles, including yaw, pitch, and roll angles. These attitude angles act like precise coordinates of the object's motion, meticulously depicting its specific location and motion state in space.

[0059] Based on this principle, one embodiment of the smart ear tag in this application is equipped with an accelerometer and further features a gyroscope. This combination enables the smart ear tag to make preliminary and accurate judgments about the details of an animal's movements and posture changes. This initial judgment data generated at the front end is properly stored as state data by the smart ear tag system and then uploaded to a big data model. This provides rich and valuable data material for subsequent, more in-depth and complex artificial intelligence learning, thereby driving the entire animal monitoring and analysis system towards greater intelligence and efficiency.

[0060] These motion sensors have the potential to be further integrated with temperature sensors through algorithmic fusion, enabling a more in-depth and comprehensive analysis of an animal's health. This is because an animal's body temperature can largely reflect its potential disease state. Take pigs, for example; within a specific temperature range, changes in their body temperature are actually a manifestation of their instinctive self-care. For instance, when a pig's body temperature rises slightly, its immunity strengthens accordingly, helping it resist external diseases. Therefore, when a pig's body temperature is below 41 degrees Celsius, using medication to reduce fever requires extreme caution.

[0061] Pigs' normal body temperature ranges from 38 to 39.5 degrees Celsius, significantly higher than that of humans. During pig farming, medication and vaccinations can cause fluctuations in a pig's body temperature. Furthermore, pigs' body temperature can rise after stress or strenuous activity; even a sow's postpartum body temperature reaching 40.3 degrees Celsius is considered normal.

[0062] Based on the temperature rise in sick pigs, it can be roughly divided into the following typical stages: First, the low-grade fever stage, with body temperature between 40-41 degrees Celsius. This stage is commonly seen in local inflammations such as mastitis and gastroenteritis that occur after farrowing, as well as some chronic infectious diseases such as chronic swine fever and paratyphoid fever. Second, the moderate fever stage, with body temperature between 41-42 degrees Celsius. This stage often occurs in acute viral infectious diseases, such as acute swine fever and swine influenza, or locally infectious diseases with significant impact, such as pneumonia. Third, the high fever stage, with body temperature between 42-43 degrees Celsius. This stage is usually closely related to acute infectious diseases and inflammation, such as swine fever, pneumonia, lobar pneumonia, and pleuropneumonia. Finally, the extremely high fever stage, with body temperature reaching above 43 degrees Celsius, often occurs in severe acute infectious diseases, such as swine erysipelas, anthrax, and acute septicemic streptococcal diseases. By accurately monitoring and assessing changes in pig body temperature, combined with data obtained from motion sensors, we can provide crucial and precise information for pig health management and disease prevention. This will help the pig farming industry develop in a more scientific and efficient direction, ensuring the health of pig herds and maximizing farming efficiency.

[0063] For example, you can select some typical pig postures, such as sitting with a hunched back, standing, walking, walking briskly, lying on your side, lying face down, looking up, looking down, etc., and summarize the raw data into posture data before uploading and processing.

[0064] To effectively reduce energy consumption and save power, a strategy of storing data first and then uploading it for processing can be adopted. Specifically, during the data acquisition phase, when the data volume accumulates to a certain scale, it is then uploaded and processed in a concentrated manner. This avoids the power consumption caused by frequent data transmission. Furthermore, the sampling frequency can be dynamically adjusted based on the animal's posture to achieve energy-saving optimization. For example, when a pig is resting, its posture is relatively stable with minimal movement. In this case, the entire sensor module can be put into sleep mode, pausing data sampling to significantly reduce power consumption. When the pig begins to walk briskly, its movement intensity and speed increase significantly, and its body posture changes frequently and dramatically. In this situation, the sampling frequency can be increased accordingly to accurately capture its movement data and posture change details. When the pig is lying on its side or engaging in non-vigorous exercise, its posture changes are relatively gentle, so the sampling frequency can be appropriately reduced. This ensures effective monitoring of its basic state while minimizing data generation, thereby reducing power consumption, extending sensor battery life, and improving the overall system's energy efficiency.

[0065] Optionally, in one embodiment, the first machine learning model is trained based on posture-temperature data pairs from multiple wearable devices. Posture-temperature data pairs from multiple animals can be continuously monitored, and the animals' health status can be manually determined (i.e., the posture-temperature data pairs are labeled) to form training data for supervised training of the first machine learning model.

[0066] The first machine learning model only needs to identify the health status of an animal based on the posture-body temperature data of multiple animals sorted by time. Various models can be used to implement the first machine learning model. Below are some non-restrictive examples.

[0067] LSTM (Long Short-Term Memory) models are an option. They are effective at processing time-series data and can learn the long-term dependencies between animal behavior and body temperature changes. LSTMs can receive posture-body temperature sequence data within continuous time windows and predict the animal's health status.

[0068] GRU (Gated Recurrent Unit) networks are another option. Compared to LSTM, GRU has a simpler structure, fewer parameters, and faster training speed, while still effectively capturing temporal features. For resource-constrained scenarios, GRU is a good choice.

[0069] Temporal Convolutional Networks (TCNs) can also be considered. TCNs process temporal data through causal convolutions, offering better parallelism and faster training compared to recurrent neural networks. They can effectively extract local pattern features from animal behavior and physiological indicators.

[0070] The encoder part of the Transformer model can also be used. Through the self-attention mechanism, it can capture long-distance dependencies in the sequence, making it particularly suitable for processing animal behavior patterns over long time spans.

[0071] Optionally, in one embodiment, the wearable device further includes a memory. The processor is configured to store attitude-temperature data pairs in the memory and transmit multiple sets of attitude-temperature data pairs stored in the memory to a cloud server via a wireless transceiver at predetermined intervals. For example, it can report once every hour or once a day.

[0072] Optionally, in one embodiment, the processor is also configured to dynamically determine an appropriate sleep duration based on the animal's current posture and to perform low-power management of the wearable device accordingly. When the system detects that the animal is in a specific posture (e.g., lying down, standing still, moving slowly, or running / walking with high activity) via an accelerometer (and optionally a gyroscope and magnetometer), the processor allocates different sleep durations according to a pre-set strategy. For example, when the processor detects that the animal is in a relatively stable, low-activity posture (such as lying down to rest) for a long time, it can put the wearable device into a longer sleep state to save energy. When the animal is in a highly active state (e.g., eating or moving quickly), the processor can shorten the sleep duration to ensure that the device is woken up more frequently and collects data, thereby capturing key dynamic changes that may affect the animal's health in a timely manner.

[0073] Furthermore, this posture-based dynamic hibernation duration control can be optimized by combining historical data and environmental information. For example, if the system detects through long-term monitoring that an animal is generally in a low-activity, resting state during a specific time period (such as the early morning), the device can automatically increase the hibernation duration during this period to maximize battery life. Conversely, if the animal is more active or in the stage of potential disease outbreak during a specific time period, the system can shorten the hibernation cycle to increase the data sampling frequency, making it easier to detect abnormal conditions in a timely manner.

[0074] This strategy dynamically optimizes the power supply and data acquisition frequency of the wearable device under different states, minimizing power consumption while ensuring monitoring accuracy, thereby extending battery life and reducing maintenance and replacement costs. This intelligent low-power management scheme lays the foundation for the long-term, stable operation of the system in real-world large-scale livestock scenarios. Optionally, in one embodiment, the wearable device also includes other sensors, such as gyroscopes, magnetometers, heart rate sensors, sound sensors, etc. The processor is also configured to determine the animal's posture based on data detected by the accelerometer, gyroscope, and magnetometer, resulting in a more accurate animal posture determination than using the accelerometer alone.

[0075] Optionally, in one embodiment, the processor can use a second machine learning model in the wearable device to determine the animal's posture based on acceleration data detected by an accelerometer. The second machine learning model can be implemented in various ways. Below are some non-limiting examples:

[0076] Traditional machine learning models:

[0077] Random Forest: After extracting features from acceleration data (such as mean, variance, frequency domain features, peak features, etc.), a random forest is used to classify the feature vectors. This method is easy to implement and has good performance and robustness for small-scale datasets.

[0078] Support Vector Machine (SVM): Similarly, feature engineering is used to transform acceleration sequences into feature vectors, and SVM is then used for multi-class pose recognition. SVM performs well in classifying high-dimensional data and has high training efficiency when the data size is moderate.

[0079] Deep learning models:

[0080] Convolutional Neural Networks (CNNs): Acceleration data can be treated as a one-dimensional time series. This data is then input into a one-dimensional convolutional neural network, which automatically learns features and outputs a pose classification result. CNNs can reduce the workload of manually designing features in feature extraction.

[0081] Long Short-Term Memory Networks (LSTM) or Gated Recurrent Units (GRUs): By using LSTM or GRUs to model acceleration signals to address the temporal correlation of time-series data, the persistence and dynamic changes of animal posture over time can be better captured.

[0082] Temporal Convolutional Networks (TCNs): These networks efficiently model time series using one-dimensional convolutions and dilated convolutions, balancing performance and computational efficiency in capturing both long-term dependencies and short-term features.

[0083] Combining CNN and LSTM (i.e., first extracting local features through CNN, and then modeling global temporal information through LSTM) is also a common strategy, which can extract effective spatial and temporal features in the model at the same time.

[0084] Lightweight Models and Embedded Optimization:

[0085] Considering the computing power and energy consumption limitations of wearable devices, lightweight models can be selected, or the aforementioned models can be pruned, quantized, and distilled. For example:

[0086] MicroNets, such as MicroCNN or quantized LSTM, are deep learning architectures that have undergone model compression and optimization, yet still maintain relatively good recognition accuracy when running on embedded devices.

[0087] TinyML tools: Compress and quantize neural networks using TinyML toolchains (such as TensorFlow Lite Micro) to make them suitable for resource-constrained hardware platforms, thus enabling them to run efficiently in wearable devices.

[0088] In summary, a second machine learning model can be implemented using traditional machine learning models (such as random forests and SVMs) to deep learning models (CNNs, LSTMs, GRUs, etc.) and their lightweight versions, to accurately and efficiently identify animal postures based on accelerometer data. When selecting a specific model, factors such as model accuracy, computational and storage resources, as well as the actual operating conditions and application scenarios of the wearable device should be comprehensively considered.

[0089] Optionally, for training the second machine learning model, the wearable device can initially upload the raw data from the accelerometer to the cloud. Then, the second machine learning model can be trained on the cloud using big data to determine the animal's posture based on the acceleration data. Although this process consumes a significant amount of power, it is a pre-training process. The trained second machine learning model is then implanted into the wearable device, after which the wearable device no longer needs to upload the raw accelerometer data.

[0090] Optionally, in one embodiment, the processor may also use non-machine learning algorithms to determine the animal's posture based on the outputs of the accelerometer, gyroscope, and magnetometer.

[0091] Optionally, in one embodiment, the processor is also configured to dynamically adjust the sampling frequencies of the accelerometer, gyroscope, magnetometer, and body temperature sensor based on the determined attitude.

[0092] Optionally, in one embodiment, the system further includes environmental parameter sensors deployed in the animal's rearing environment to measure environmental parameters of the environment in which the animal is located. The environmental parameter sensors periodically upload these environmental parameters to a cloud server. The wearable device also includes a clock, and the processor is further configured to obtain time information for determining posture and acquiring body temperature from the clock, forming a timestamp, and transmitting the posture-body temperature data pair, the corresponding timestamp, and the wearable device's identifier to the cloud server via a wireless transceiver. The cloud server determines the environmental parameter sensors in the environment where the wearable device is located based on the wearable device's identifier, and determines the environmental parameters corresponding to the posture-body temperature data pair based on the timestamp. The posture-body temperature data pair and its corresponding environmental parameters are input into a first machine learning model to obtain the animal's health status output by the first machine learning model.

[0093] Environmental parameters can include ambient temperature, ambient humidity, and ambient air pressure. Environmental parameter sensors can include temperature sensors that measure ambient temperature, humidity sensors that measure ambient humidity, and air pressure sensors that measure ambient air pressure.

[0094] A timestamp is data that represents time, but it is not necessarily a single time value. For example, if environmental parameter sensors are used periodically, the timestamp can be an integer representing how many cycles have passed since the agreed start time.

[0095] Optionally, in one embodiment, to improve the accuracy and applicability of the system in assessing animal health status, the system can further expand the range of environmental parameters collected and the methods of utilization. In this embodiment, in addition to basic environmental parameter sensors such as temperature, humidity, and air pressure, more types of sensors can be deployed to obtain more comprehensive environmental information, for example:

[0096] Harmful gas sensors: used to detect the concentration of harmful or irritating gases (such as ammonia and hydrogen sulfide) in the breeding environment to assess the potential negative impact of the breeding environment on the animal's respiratory system and overall health.

[0097] Dust / particulate matter sensors: used to detect the concentration of particulate matter (such as PM2.5, PM10) in the environment, thereby assessing the air quality level of the breeding environment and its potential impact on animal health.

[0098] Light sensors: used to measure the light intensity and light cycle in the enclosure, so as to perform correlation analysis between environmental light conditions and animal behavior (feeding, exercise, resting cycles).

[0099] Acoustic sensors: used to monitor noise levels in enclosures or pastures and assess the impact of the sound environment on animal stress levels and health.

[0100] In this embodiment, the environmental parameter sensor uploads multi-dimensional environmental data, including ambient temperature, humidity, air pressure, gas concentration, dust particle count, light intensity, and noise level, to the cloud server at set time intervals (e.g., every few minutes to every few hours, the specific interval can be flexibly adjusted according to actual needs).

[0101] Wearable devices still use a built-in clock to timestamp the posture and body temperature data collected, and then transmit the posture-body temperature data pairs with timestamps and device identifiers to the cloud server. Based on the wearable device's identification information, the cloud server matches and associates the posture-body temperature data pairs corresponding to the device with environmental data measured at the same (or close to) timestamp by the environmental parameter sensors corresponding to its location.

[0102] After obtaining complete data correlations, the cloud server inputs multimodal data, including posture, body temperature, and various environmental parameters, into a pre-trained first machine learning model. This model utilizes these environmental parameters as part of its input during training to improve the accuracy of health status predictions. For example, if the model identifies an animal exhibiting abnormally elevated body temperature and atypical postures (such as lethargy or rapid breathing) in environments with high temperature, high humidity, or high concentrations of harmful gases, the probability of the model issuing a warning about its health status will increase accordingly.

[0103] This embodiment has the following technical effects:

[0104] 1. Improve prediction accuracy and scenario adaptability: With more comprehensive environmental parameter data, the model can better distinguish between short-term abnormal behavior caused by environmental changes and real health problems caused by diseases, thereby reducing false alarms or missed alarms.

[0105] 2. Support for precise decision-making and intervention: Based on the comprehensive analysis of environmental data and animal status data, keepers can promptly identify unhealthy feeding conditions (such as overcrowding, poor ventilation, high concentration of harmful gases, etc.) and make targeted adjustments (increasing ventilation, controlling lighting, improving the cleanliness of the enclosure, etc.).

[0106] 3. Facilitating Large-Scale Intelligent Livestock Management: In large-scale farming scenarios, integrating environmental parameter sensors, wearable devices, and cloud-based intelligent analysis helps provide dynamic and precise health monitoring and recommendations for each animal. Through continuous data accumulation and model training, the system can continuously improve its understanding of the complex relationship between environmental changes and animal health, thereby achieving a higher level of automated and intelligent livestock management.

[0107] In summary, by collecting richer environmental parameters and fusing deep data, a more comprehensive, accurate, and scalable foundation is provided for animal health monitoring and early warning, enabling the first machine learning model to make more reliable predictions of animal health status in complex and ever-changing breeding environments.

[0108] Optionally, in one embodiment, the system further includes a wireless gateway device that establishes wireless connections with multiple wearable devices and multiple environmental parameter sensors, and uploads data from the wearable devices and environmental parameter sensors to a cloud server. The wireless gateway device is not mandatory; for example, if the wearable devices can directly access the internet (e.g., via an ultra-low power cellular network), the wireless gateway device can be omitted.

[0109] Optionally, in one embodiment, the system employs a multi-layered wireless network architecture, including multiple wireless gateway devices. Each wireless gateway device is responsible for covering wearable devices and environmental parameter sensors within a specific area, establishing stable wireless connections with these devices. The wireless gateway devices feature an industrial-grade design, are waterproof and dustproof, and can be installed in appropriate locations within the animal husbandry environment.

[0110] The wireless gateway device supports multiple wireless communication protocols, including Bluetooth Low Energy (BLE), Wi-Fi, ZigBee, and LoRa. For short-range communication, the gateway uses BLE, Wi-Fi, or ZigBee protocols to exchange data with wearable devices and environmental parameter sensors. These protocols are characterized by low power consumption, making them suitable for long-term operation of battery-powered devices. For scenarios requiring long-range coverage, the gateway can use the LoRa protocol to achieve wireless communication over a range of several kilometers.

[0111] To improve communication reliability, each wireless gateway device (referred to as "gateway") can be equipped with mesh networking capabilities. Gateways can communicate with each other to form a self-organizing network. When a gateway fails or has a weak signal, data can be transmitted through other gateways via a detour, ensuring stable system operation. Simultaneously, each wearable device and environmental parameter sensor can also establish connections with multiple gateways, achieving communication redundancy.

[0112] The wireless gateway device possesses local storage and computing capabilities. When the network connection is unstable, the gateway can temporarily store data uploaded by the device and upload it to the cloud server in batches once the network is restored. The gateway can also preprocess and compress data to reduce network transmission load. In addition, the gateway has a simple anomaly detection function, which can quickly detect and report device malfunctions or abnormal situations.

[0113] Each wireless gateway device can be equipped with Ethernet and 4G / 5G communication modules, enabling it to access the internet via wired or wireless means and transmit data to a cloud server. The gateway automatically selects the optimal network connection method, switching to the wireless network when the wired network is unavailable to ensure continuous data transmission.

[0114] The system employs a hierarchical data transmission strategy. Wearable devices and environmental parameter sensors select different transmission methods based on the importance and urgency of the data. For routine monitoring data, batch transmission can be used to reduce communication frequency. For detected abnormal data or alarm information, the data is immediately reported to the cloud server through the gateway to ensure timely detection and handling of problems.

[0115] Optionally, in one embodiment, the system may also utilize image recognition technology to achieve automatic annotation of pose data, thereby training the first machine learning model more efficiently.

[0116] Specifically, in this embodiment, the system also includes a video monitoring device, such as a camera installed in the animal husbandry environment. The implementation of this embodiment is as follows:

[0117] 1. Training the third machine learning model:

[0118] First, prepare a large dataset of animal images (e.g., images from different individuals, postures, and rearing environments of pigs). Then, manually label the animal postures in these images (e.g., standing, lying down, lame, eating, sitting, etc.) to create a high-quality training dataset.

[0119] The third machine learning model was trained under supervised supervision using this labeled image dataset, enabling it to automatically identify animal poses based on single-frame images. Because this training is based on a large amount of manually labeled image data, the third machine learning model exhibits strong pose recognition capabilities and good robustness to complex environments. The third machine learning model can employ a convolutional neural network structure, such as ResNet or MobileNet image recognition networks.

[0120] 2. Data collection and automatic annotation process:

[0121] In practical use, cameras are deployed in animal husbandry environments to continuously record video of multiple animals wearing wearable devices.

[0122] Meanwhile, wearable devices worn on the animals (including accelerometers and optional sensors such as gyroscopes and magnetometers) record the animals' status at a certain sampling frequency and upload the data to the cloud server wirelessly.

[0123] In a cloud server or edge computing unit, the video stream captured by the camera is decomposed into consecutive image frames. These image frames are then input into a pre-trained third-party machine learning model to automatically identify the animal posture at each time point. Since the individual animal can be identified in the scene captured by the camera (e.g., through specific markers, fence positioning, image recognition algorithms, or other positioning methods), the system can associate the posture identified in the image with a specific timestamp and sensor data from a specific wearable device (corresponding to a specific animal).

[0124] In other words, for each timestamp, the animal pose label obtained from the camera image can be matched one-to-one with the acceleration data uploaded by the wearable device at the same moment (e.g., a sequence of acceleration vectors over a predetermined period of time prior to that moment). In this way, the accelerometer data (and optional gyroscope and magnetometer data) can be automatically labeled using the pose recognition results from a third machine learning model without human intervention.

[0125] 3. Train the second machine learning model:

[0126] Through the aforementioned automatic annotation method, a large number of "attitude-acceleration data" samples are generated rapidly and efficiently. This automatically annotated data is then input into a second machine learning model for training, enabling the model to predict animal posture using accelerometer (and optionally gyroscope and magnetometer) data. Due to the large amount of training data and accurate annotation, the second machine learning model has stronger generalization ability and higher posture recognition accuracy for different environments and different animals.

[0127] 4. Push the trained second machine learning model to the animal's wearable device.

[0128] Data can be continuously collected using cameras and accelerometers to optimize the second machine learning model. The optimized second machine learning model can be periodically sent to various wearable devices, thereby continuously improving the accuracy of the second machine learning model in detecting animal postures.

[0129] This embodiment has the following effects:

[0130] 1. Reduced Cost and Difficulty of Manual Annotation: Traditional methods require manual annotation of the attitude corresponding to accelerometer data, which is both time-consuming and prone to errors. However, by first training a third machine learning model to automatically annotate the images, and then achieving automated annotation through time synchronization between video and accelerometer data, the workload and cost of manual annotation are greatly reduced.

[0131] 2. Improve data annotation consistency and quality: The third machine learning model, trained on a large amount of labeled image data, achieves higher pose recognition accuracy and consistency than human visual observation. Using this model to automatically label accelerometer data can improve the quality and consistency of training data, thereby enhancing the predictive performance of the final first machine learning model.

[0132] 3. Accelerated Model Iteration and Optimization: With the automatic annotation process implemented, when the system adds or updates accelerometer data, it can quickly obtain new, high-quality training data without additional manual annotation. This enables the first machine learning model to iterate and optimize more quickly, adapting more rapidly to different scenarios or groups of animals, thereby improving the system's flexibility and adaptability.

[0133] This embodiment enables the construction of an efficient and automated closed-loop process for training data generation and model optimization, thereby improving the overall accuracy and reliability of animal health monitoring systems.

[0134] In various embodiments of this application, the processor may be a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Digital Signal Processor (DSP), a Microcontroller Unit (MCU), a Neural Processing Unit (NPU), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or other programmable logic devices. The aforementioned memory may be random access memory (RAM), flash memory, a hard disk, or a solid-state drive, etc.

[0135] It should be noted that in this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this application, if it refers to performing an action according to an element, it means performing the action at least according to that element, including two cases: performing the action only according to that element, and performing the action according to that element and other elements. Expressions such as "multiple," "repeatedly," and "various" include two, two times, two kinds, and more than two, more than two times, and more than two kinds.

[0136] This specification includes combinations of various embodiments described herein. Individual references to embodiments are made (e.g., "one embodiment," "some embodiments," or "preferred embodiments"); however, these embodiments are not mutually exclusive unless indicated to be mutually exclusive or are readily apparent to those skilled in the art. It should be noted that the word "or" is used in a non-exclusive sense throughout this specification unless the context explicitly indicates or requires it.

[0137] All references to this specification are considered to be incorporated integrally into the disclosure of this application so that they can serve as the basis for modifications if necessary. Furthermore, it should be understood that the above descriptions are merely preferred embodiments of this specification and are not intended to limit the scope of protection of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the scope of protection of one or more embodiments of this specification.

[0138] In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An animal health monitoring system, characterized in that, include: Wearable devices and cloud servers; The wearable device includes a processor, and an accelerometer, a body temperature sensor, and a wireless transceiver connected to the processor; the processor is configured to: determine the animal's posture based on acceleration data detected by the accelerometer, and simultaneously acquire the animal's body temperature from the body temperature sensor; The posture-body temperature data pair is transmitted to the cloud server via the wireless transceiver; The cloud server is configured to input multiple sets of posture-body temperature data pairs from the wearable device, sorted chronologically, into a pre-trained first machine learning model to obtain the animal health status output by the first machine learning model.

2. The animal health monitoring system of claim 1, wherein, There are multiple wearable devices; The first machine learning model was trained based on posture-temperature data from multiple wearable devices.

3. The animal health monitoring system of claim 1, wherein, The wearable device also includes a memory; the processor is configured to store the posture-body temperature data pairs in the memory and transmit multiple sets of the posture-body temperature data pairs stored in the memory to the cloud server via the wireless transceiver at a predetermined time.

4. The animal health monitoring system of claim 1, wherein, The processor is also configured to: determine the hibernation duration based on the animal's current posture, control the wearable device to enter a hibernation state, wake up the wearable device after the hibernation duration, and perform the next detection, wherein at least two different postures correspond to different hibernation durations.

5. The animal health monitoring system of claim 1, wherein, The wearable device also includes a gyroscope and a magnetometer; The processor is also configured to determine the animal's posture based on data detected by the accelerometer, the gyroscope, and the magnetometer.

6. The animal health monitoring system of claim 1, wherein, The processor uses a second machine learning model in the wearable device to determine the animal's posture based on acceleration data detected by the accelerometer.

7. The animal health monitoring system of claim 5, wherein, The processor is also configured to dynamically adjust the sampling frequencies of the accelerometer, the gyroscope, the magnetometer, and the body temperature sensor based on the determined attitude.

8. The animal health monitoring system of claim 1, wherein, The system also includes environmental parameter sensors deployed in the animal husbandry environment for measuring environmental parameters of the environment in which the animals are located; The environmental parameter sensor periodically uploads environmental parameters to the cloud server; The wearable device also includes a clock, and the processor is further configured to obtain time information from the clock to determine the posture and obtain the body temperature, form a timestamp, and send the posture-body temperature data pair, the corresponding timestamp, and the identifier of the wearable device to the cloud server through the wireless transceiver; The cloud server determines the environmental parameter sensors of the environment in which the wearable device is located based on the identifier of the wearable device, and determines the environmental parameters corresponding to the posture-body temperature data pair based on the timestamp; the posture-body temperature data pair and its corresponding environmental parameters are input into the first machine learning model to obtain the animal health status output by the first machine learning model.

9. The animal health monitoring system of claim 8, wherein, The environmental parameters include one or any combination of the following: ambient temperature, ambient humidity, and ambient air pressure. The environmental parameter sensor includes one or any combination of the following: A temperature sensor for measuring ambient temperature, a humidity sensor for measuring ambient humidity, and a barometric pressure sensor for measuring ambient barometric pressure.

10. The animal health monitoring system of claim 8, wherein, A wireless gateway device is also included, which establishes wireless connections with the plurality of wearable devices and the plurality of environmental parameter sensors, and uploads data from the wearable devices and the environmental parameter sensors to the cloud server.

Images