Intelligent pet management system and method for health monitoring

Abstract

The invention relates to the field of pet management, and particularly discloses an intelligent pet management system and method for health monitoring, and the method comprises the steps: firstly obtaining pet physiological indexes collected by a built-in integrated sensor of a pet collar and pet life behavior data collected by an intelligent platform, and then carrying out the deep learning technology, and feature extraction and correlation analysis are carried out on the two images, and finally whether a pet health abnormality early warning is sent out is judged through a classifier, so that potential health problems and abnormalities are identified, measures are taken in time, and a pet owner is helped to take care of the health of the pet better.

Description

Technical Field

[0001] The present application relates to the field of pet management, and more specifically, to an intelligent pet management system and method for health monitoring. Background Art

[0002] With the continuous improvement of economic level and changes in lifestyle, pets have become an important member of many families. According to statistics, pet keeping is increasing year by year around the world. Pets not only bring joy to life, but also improve the quality of life of family members to a certain extent. However, along with this comes the increasing demand for pet health management. However, since most family members are busy at work, it is difficult to provide comprehensive pet health care in daily life. This busy lifestyle often leads to the health status of pets being neglected.

[0003] In the prior art, only a single health status of a pet can be recorded, which makes it difficult to fully and accurately reflect the health status of the pet.

[0004] Therefore, an intelligent pet management system and method for health monitoring is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides an intelligent pet management system and method for health monitoring, which first obtains the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform, and then uses deep learning technology to extract features and perform correlation analysis on the two, and finally uses a classifier to determine whether to issue a pet health abnormality warning, thereby identifying potential health problems and abnormalities, so that timely measures can be taken, thereby helping pet owners to better take care of their pets' health.

[0006] According to one aspect of the present application, there is provided an intelligent pet management system for health monitoring, which includes:

[0007] The pet health detection data acquisition module is used to obtain the pet's physiological indicators collected by the integrated sensor built into the pet collar and the pet's life behavior data collected by the smart platform;

[0008] A pet health detection data extraction module, used to extract pet physiological index data text understanding feature vectors and pet life behavior association feature vectors from the pet physiological indexes collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform;

[0009] The pet health abnormality warning module is used to determine whether to issue a pet health abnormality warning based on the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector.

[0010] According to another aspect of the present application, a smart pet management method for health monitoring is provided, which includes:

[0011] Obtain pet physiological indicators collected by the integrated sensors in the pet collar and pet life behavior data collected by the smart platform;

[0012] Extracting pet physiological indicator data text understanding feature vectors and pet life behavior association feature vectors from the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform;

[0013] Based on the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector, it is determined whether to issue a pet health abnormality warning.

[0014] Compared with the prior art, the present application provides an intelligent pet management system and method for health monitoring, which first obtains the pet's physiological indicators collected by the built-in integrated sensor of the pet collar and the pet's life behavior data collected by the intelligent platform, and then uses deep learning technology to perform feature extraction and correlation analysis on the two. Finally, a classifier is used to determine whether to issue an abnormal pet health warning, thereby identifying potential health problems and abnormalities so that timely measures can be taken, thereby helping pet owners to better take care of their pets' health. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of an intelligent pet management system for health monitoring according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of a pet health detection data extraction module in an intelligent pet management system for health monitoring according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a pet physiological indicator feature extraction unit in an intelligent pet management system for health monitoring according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a pet life behavior data feature extraction unit in an intelligent pet management system for health monitoring according to an embodiment of the present application.

[0020] Figure 5 It is a block diagram of a pet health abnormality warning module in an intelligent pet management system for health monitoring according to an embodiment of the present application.

[0021] Figure 6 This is a flow chart of a smart pet management method for health monitoring according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0023] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0024] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.

[0025] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0026] Figure 1 FIG. 1 is a block diagram of an intelligent pet management system for health monitoring according to an embodiment of the present application. Figure 1As shown, according to the embodiment of the present application, the smart pet management system 100 for health monitoring includes: a pet health detection data acquisition module 110, which is used to obtain the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the smart platform; a pet health detection data extraction module 120, which is used to extract the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector from the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the smart platform; a pet health abnormality warning module 130, which is used to determine whether to issue a pet health abnormality warning based on the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector.

[0027] In the above-mentioned smart pet management system 100 for health monitoring, the pet health detection data acquisition module 110 is used to obtain pet physiological indicators collected by the built-in integrated sensor of the pet collar and pet life behavior data collected by the intelligent platform. It should be understood that with the development of the economy and the change of lifestyle, pets have become the core members of many families. Statistics show that the number of pets kept worldwide continues to rise. Pets not only add fun to life, but also improve the quality of life of family members to a certain extent. However, this also brings higher demands for pet health management. Because most family members are busy at work and it is difficult to provide comprehensive pet care in daily life, this busy lifestyle often makes the health status of pets ignored. Current technology can usually only record a single health indicator of pets, and it is difficult to fully and accurately reflect the overall health status of pets. Therefore, in the technical solution of the present application, by taking the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform, and combining deep learning technology, it is judged whether it is necessary to issue a pet health abnormality warning to identify potential health problems and abnormal conditions, so as to take appropriate measures in time. This will take care of the health of pets more effectively and ensure their well-being.

[0028] Specifically, the sensors built into the pet collar can monitor and record the pet's physiological indicators in real time. These indicators include heart rate, body temperature, and activity level. The sensors can monitor these indicators around the clock and transmit the data to the smart platform wirelessly, providing detailed and continuous physiological information. On the other hand, the smart platform collects pet's life behavior data through various data collection methods, including pet behavior monitoring (such as eating, drinking, sleeping, etc.) and environmental parameters (such as temperature, humidity, etc.). These data help supplement the monitoring of physiological indicators to provide a more comprehensive health assessment. These behavioral data combined with physiological indicators can provide deeper insights into pet health, thereby achieving comprehensive monitoring of pet health status, timely identification of abnormalities, providing health warnings, and taking action in time before problems become serious. This comprehensive data acquisition and analysis method not only improves the efficiency of pet health management, but also enables pet owners to better take care of their pets in their busy lives to ensure their health and happiness.

[0029] In the above-mentioned smart pet management system 100 for health monitoring, the pet health detection data extraction module 120 is used to extract pet physiological indicator data text understanding feature vectors and pet life behavior association feature vectors from the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the smart platform. In this way, the smart pet health management system can comprehensively evaluate the pet's physiological state and life behavior, and provide more accurate health warnings and intervention suggestions. This multi-dimensional data analysis method enables the system to monitor the health status of the pet in real time, detect potential problems in a timely manner, and provide effective care suggestions for pet owners.

[0030] Figure 2 FIG. 1 is a block diagram of a pet health detection data extraction module in an intelligent pet management system for health monitoring according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the pet health detection data extraction module 120 includes: a pet physiological indicator feature extraction unit 121, used to perform feature extraction on the pet physiological indicators collected by the built-in integrated sensor of the pet collar to obtain the pet physiological indicator data text understanding feature vector; a pet life behavior data feature extraction unit 122, used to perform feature extraction on the pet life behavior data collected by the smart platform to obtain the pet life behavior associated feature vector.

[0031] It should be understood that the sensors in pet collars can monitor a variety of physiological indicators in real time, such as heart rate, body temperature, activity level, etc. These data are usually raw, continuous time series data. Although the values ​​directly obtained from the sensors contain a lot of information, they are difficult to be used directly for health assessment without processing and analysis. The main task of feature extraction is to extract key features from these raw data for better understanding and application. Specifically, the process of feature extraction involves converting raw data into feature vectors, which are high-dimensional representations of data that can capture potential patterns and important information in physiological data.

[0032] Furthermore, the smart platform collects pets' daily activity data through various means, such as food intake, water intake, sleep duration, activity level, etc. These data provide detailed information about pets' daily lives, but the raw data are usually large, redundant, and may contain noise. Therefore, feature extraction becomes a key step, which can convert these data into feature vectors with high information content, which is convenient for further analysis and application. Specifically, the feature extraction process maps the raw behavioral data to a high-dimensional feature space, and these feature vectors can capture important patterns and associations in the behavioral data.

[0033] Figure 3 FIG. 1 is a block diagram of a pet physiological index feature extraction unit in an intelligent pet management system for health monitoring according to an embodiment of the present application. Figure 3 As shown, in a specific embodiment of the present application, the pet physiological indicator feature extraction unit 121 includes: a pet physiological indicator semantic feature encoding subunit 1211, which is used to perform semantic feature encoding on the pet physiological indicator collected by the built-in integrated sensor of the pet collar to obtain a pet physiological indicator data vector sequence; a pet physiological indicator semantic understanding subunit 1212, which is used to pass the pet physiological indicator data vector sequence through the pet physiological indicator data vector sequence semantic understanding Bi-LSTM model to obtain the pet physiological indicator data text understanding feature vector.

[0034] It should be understood that although the physiological indicators collected by the sensors of pet collars are detailed, directly analyzing the raw data may be very complicated and time-consuming. In order to efficiently extract useful information from these data, semantic feature coding came into being. The main purpose of semantic feature coding is to convert raw physiological data into a sequence of feature vectors that can effectively represent the key features and patterns of the data. By applying deep learning models (such as recurrent neural networks (RNNs), long short-term memory networks (LSTMs), etc.), the system can capture the temporal dependencies and potential patterns in the data. For example, fluctuations in heart rate and periodic changes in body temperature can all be extracted through feature coding to extract representative features.

[0035] Furthermore, the pet physiological index data vector sequence is processed through the Bi-LSTM (bidirectional long short-term memory network) model in order to achieve in-depth understanding and precise analysis of pet physiological data. This process helps to reveal the potential patterns and relationships in the data, thereby improving the accuracy and practicality of health monitoring. Specifically, the Bi-LSTM model is a deep learning model that can process time series data. Its bidirectional characteristics enable the model to consider the information of the previous and next time when understanding the data, which is particularly important for the analysis of physiological data. Among them, the Bi-LSTM model consists of two LSTM networks, one for processing the forward information of the sequence and the other for processing the reverse information. In this way, the model can capture the complex time dependency in the data and integrate the previous and next context information to generate a more comprehensive feature representation. When processing pet physiological data, Bi-LSTM can identify patterns such as periodic changes and sudden abnormalities in the data. For example, by analyzing the changing patterns of heart rate and body temperature, Bi-LSTM can effectively capture abnormal fluctuations, long-term trends, etc., which are important bases for evaluating pet health.

[0036] In a specific embodiment of the present application, the pet physiological indicator semantic feature encoding subunit 1211 includes: passing the pet physiological indicators collected by the built-in integrated sensor of the pet collar through a pet physiological indicator semantic context encoder including an embedding layer to obtain a plurality of pet physiological indicator semantic feature vectors; arranging the plurality of pet physiological indicator semantic feature vectors into the pet physiological indicator data vector sequence.

[0037] It should be understood that the embedding layer is a key component in deep learning models, which can map discrete or high-dimensional data into a continuous, low-dimensional embedding space. In the case of physiological indicators, the embedding layer can map the original physiological data values ​​into a more compact and semantically characteristic space. Through this mapping, the different dimensions of the original data are transformed into vectors with higher semantic density, which is convenient for capturing potential patterns and relationships in the data. The pet physiological indicator semantic context encoder combines the embedding layer with the context encoding technology, which not only embeds the physiological data into the feature space, but also considers the context information of the data. In this way, the model can better understand the temporal relationship and dependency between the data. For example, the changing pattern of heart rate and the fluctuation of body temperature can be identified by the context encoder to generate semantic feature vectors with practical significance. These feature vectors not only retain the temporal information of the data, but also enhance the ability to identify potential health problems in the data. Specifically, the pet physiological indicators collected by the built-in integrated sensor of the pet collar are converted into embedding vectors using an embedding layer of a pet physiological indicator semantic context encoder including an embedding layer to obtain a sequence of pet physiological indicator embedding vectors; and the sequence of pet physiological indicator embedding vectors is encoded based on global context semantics using a converter-based Bert model of a pet physiological indicator semantic context encoder including an embedding layer to obtain multiple pet physiological indicator semantic feature vectors.

[0038] Furthermore, in the pet physiological indicator data, each physiological indicator can be converted into a semantic feature vector through the aforementioned semantic feature encoder. These vectors can capture the key features and contextual information of the physiological data, making the processing of the original data more efficient and meaningful. These feature vectors are embedded and encoded to form a physiological indicator data vector sequence in order to retain the semantic information in the data. The arrangement of the vector sequence facilitates further processing using advanced analysis and prediction models. For example, deep learning models such as LSTM (Long Short-Term Memory Network) and Transformer can be trained using the contextual information of time series data to identify potential health problems or predict future health status. By inputting feature vector sequences into these models, the system is able to more accurately capture long-term dependencies and complex patterns in the data, thereby providing more insightful analysis results.

[0039] Figure 4 FIG. 1 is a block diagram of a pet life behavior data feature extraction unit in an intelligent pet management system for health monitoring according to an embodiment of the present application. Figure 4As shown, in a specific embodiment of the present application, the pet life behavior data feature extraction unit 122 includes: a pet life behavior data preprocessing subunit 1221, used to perform data preprocessing on the pet life behavior data collected by the intelligent platform to obtain a pet life behavior associated input matrix; a pet life behavior feature encoding subunit 1222, used to pass the pet life behavior associated input matrix through a pet life behavior feature encoder to obtain the pet life behavior associated feature vector.

[0040] It should be understood that smart platforms usually collect pet life behavior data including activity, eating habits, sleep patterns, etc. These data are usually recorded in raw form and may include a variety of different sensor data, timestamps, and measurements. The complexity and diversity of raw data make direct analysis difficult, so data preprocessing is required to extract useful information and create a unified data structure. In a specific embodiment, the first step of data preprocessing usually involves data cleaning and normalization. This includes removing noise, filling missing values, and standardizing data formats. Next, data preprocessing may include feature extraction and transformation, such as extracting daily activity patterns from activity data and analyzing eating frequency and amount from dietary records. After processing the data, the final step of preprocessing is to construct a pet life behavior association input matrix. This matrix organizes the processed behavioral data into a certain structure, usually arranged in time series or behavioral categories. Each element of the matrix can represent behavioral characteristics at a specific time point or time period, such as activity, diet, sleep time, etc. In this way, the data becomes more structured and consistent, which is convenient for further analysis and modeling.

[0041] Furthermore, the pet life behavior association input matrix is ​​processed through the pet life behavior feature encoder to convert the structured behavior data into a more condensed and high-information density feature representation, thereby improving the efficiency and accuracy of data analysis. This process is crucial for a deep understanding of pet behavior patterns and their health status. Among them, the pet life behavior association input matrix usually contains pre-processed behavior data, which is organized by time series or behavior categories, covering information such as activity volume, eating habits, and sleep patterns. However, the data volume of these matrices is large and complex, and it may be difficult to directly extract useful information from them. In order to make more effective use of these data, it is necessary to convert them into more refined feature vectors with practical application value. The role of the pet life behavior feature encoder is to map the raw behavior data in these matrices into a new feature space. The feature encoder processes the matrix through a deep learning model, such as a convolutional neural network (CNN) or a transformer, and extracts key features and patterns from it. These feature encoders can capture the complex relationships and dependencies in the data and convert them into high-dimensional information representations, called feature vectors. The generation of feature vectors makes pet behavior data more compact and semantically informative. Through this mapping, the feature vector can condense the key information in the matrix, retain the core features of the data, and eliminate redundancy and noise. This process makes subsequent data analysis and modeling more efficient. Specifically, each layer of the pet life behavior feature encoder is used to perform convolution processing, mean pooling processing based on the local feature matrix, and nonlinear activation processing on the input data in the forward pass of the layer to output the pet life behavior associated feature vector by the last layer of the pet life behavior feature encoder, wherein the input of the pet life behavior feature encoder is the pet life behavior associated input matrix.

[0042] In a specific embodiment of the present application, the pet life behavior data preprocessing subunit 1221 includes: embedding the pet life behavior data collected by the intelligent platform into a pet life behavior data encoder to obtain a plurality of pet life behavior word vector sequences; and arranging the plurality of pet life behavior word vector sequences in two dimensions to obtain the pet life behavior associated input matrix.

[0043] It should be understood that the pet life behavior data collected by the smart platform usually includes activity, eating habits, sleep patterns, etc. These data exist in raw form and may have large dimensions and complex time series characteristics. Directly processing these high-dimensional, raw data will be very complex and prone to noise. In order to make these data more structured and analyzable, it is necessary to convert the data into a more refined and meaningful representation through an embedding encoder. The core function of the embedding encoder is to map the raw data into a low-dimensional embedding space, in which each data point (such as a specific behavior pattern) is represented as a word vector. Word vectors are dense vectors used to represent discrete data in deep learning, which can capture semantic information and potential relationships in the data. In pet life behavior data, word vectors can convert behavioral features such as activity level, diet, sleep time, etc. into feature representations that can be used for further analysis. Multiple pet life behavior word vector sequences processed by the embedding encoder can effectively capture and express key patterns and features in the behavior data. For example, an increase or decrease in activity, changes in eating habits, etc. can be reflected by changes in these word vectors. This conversion not only reduces the dimension of the data, but also retains the core information of the data, making subsequent analysis and modeling more efficient and accurate. Specifically, the pet life behavior data collected by the intelligent platform is segmented to obtain a pet life behavior word sequence; the embedding layer of the pet life behavior data embedding encoder is used to map each pet life behavior word in the pet life behavior word sequence to a word embedding vector to obtain multiple pet life behavior word vector sequences.

[0044] Furthermore, the word vector sequences of pet life behaviors are transformed from the original behavior data by embedding encoders, and each word vector represents the behavior characteristics at a specific time point or time period. These word vector sequences contain information about pet activities, diet, sleep, etc., but when these sequences exist alone, it may be difficult to fully display their complex behavior patterns and their changing trends. Arranging these word vector sequences in two dimensions, that is, organizing them into a matrix structure, helps to achieve systematic management of data. In this two-dimensional matrix, one dimension usually represents the time series, while the other dimension can represent different behavior characteristics or different time windows. This arrangement allows the behavior characteristics of each time point to be accurately located and displayed in the matrix, thereby retaining the time series information of the behavior data and its corresponding characteristics. This two-dimensionally arranged matrix form provides a unified input format for data analysis and modeling, allowing machine learning models and statistical analysis tools to process data more effectively. The rows of the matrix may represent different time steps or time periods, while the columns represent various behavior characteristics or indicators at these time points. This structured data representation can reveal the changing trends of behavior patterns, the relationships between behavior characteristics, and long-term health trends.

[0045] In the above-mentioned intelligent pet management system 100 for health monitoring, the pet health abnormality warning module 130 is used to determine whether to issue a pet health abnormality warning based on the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector. It should be understood that pet physiological indicator data usually includes body temperature, heart rate, respiratory rate, etc., and these physiological data can directly reflect the physical health of the pet. However, analyzing physiological indicator data alone may not fully capture the complexity of health abnormalities, because physiological data may be affected by various factors, such as environmental changes, eating habits, etc. Therefore, converting these physiological indicator data into text understanding feature vectors can better capture the key features and potential health information in the data. At the same time, the pet life behavior association feature vector provides detailed information about the pet's daily behavior patterns, such as activity level, eating behavior, and sleep patterns. These behavioral characteristics can reveal the health trends and behavioral changes of pets and help identify abnormal patterns related to health conditions. Combining the physiological indicator data text understanding feature vector and the life behavior association feature vector for analysis can achieve a multi-dimensional assessment of the health status of the pet. This comprehensive analysis not only improves the accuracy of health warnings, but also enhances the real-time monitoring capability of the pet's health status.

[0046] Figure 5 FIG. 1 is a block diagram of a pet health abnormality warning module in an intelligent pet management system for health monitoring according to an embodiment of the present application. Figure 5 As shown, in a specific embodiment of the present application, the pet health abnormality warning module 130 includes: a pet health data feature fusion unit 131, which is used to adjust the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector based on the offset information corresponding to the feature fine-grainedness to obtain a pet health abnormality warning judgment feature vector; a pet health abnormality warning classification judgment unit 132, which is used to pass the pet health abnormality warning judgment feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a pet health abnormality warning.

[0047] It should be understood that the fusion of pet physiological indicator data text understanding feature vectors and pet life behavior association feature vectors to obtain pet health abnormal warning judgment feature vectors is to integrate information from different data sources to provide a more comprehensive and accurate health monitoring and abnormal warning system. This fusion can integrate the advantages of physiological data and behavioral data, and improve the detection ability and warning accuracy of pet health status. In one embodiment of the present application, the fusion process combines the two feature vectors by utilizing machine learning or deep learning models to extract representative patterns and abnormal signals. Through the training model, the system can learn the relationship between normal health status and abnormal status, and can issue a warning when potential health problems are detected. This method makes health monitoring not only based on a single data source, but also comprehensively considers various information, thereby improving the accuracy and reliability of the warning.

[0048] In particular, in the technical solution of the present application, pet physiological indicators are collected by sensors, processed by the embedding layer and the Bi-LSTM model, and mainly capture the time series information and contextual relationship of physiological data. These features are usually highly abstract and have strong temporal and semantic characteristics. On the other hand, pet life behavior data is processed by the embedding encoder and the feature encoder, focusing on capturing behavior patterns and associated features, which usually focus on the spatial and relational characteristics of the behavior. Therefore, there are large differences between physiological indicator features and behavioral features in terms of semantic level and dimension. This difference leads to the offset of the fine-grained dimension of the features, that is, the expression form and detail level of the two in the feature space are inconsistent, thus affecting the fusion effect. Simply fusion of the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector may not effectively align the physiological indicator features and behavioral features, and may cause information loss or structural distortion. Specifically, the time series information of physiological features and the spatial relationship of behavioral features cannot be balanced in the feature vector after simple fusion. For example, subtle changes in physiological data may be diluted and cannot be effectively reflected in the final feature vector, and local behavioral patterns in behavioral data may not be fully reflected. This structural mismatch causes the final feature vector in the simple fusion process to be distorted in some important details, reducing the accuracy of the early warning system. Therefore, in the technical solution of the present application, the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector are adjusted based on the offset information corresponding to the fine-grained features to obtain the pet health abnormality early warning judgment feature vector.

[0049] Among them, the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector are adjusted based on the offset information corresponding to the feature fine-grainedness to obtain the pet health abnormality warning judgment feature vector, including: calculating the position difference between the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector to obtain the offset information adjusted differential feature vector; calculating the position point multiplication between the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector to obtain the offset information adjusted position point multiplication feature vector; calculating the position addition between the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector to obtain the offset information adjusted position addition feature vector; adding the offset information adjusted differential feature vector and the offset information adjusted position point multiplication feature vector. The multiplication feature vector and the offset information adjusted position-added feature vector are cascaded and input into a one-dimensional convolution layer and maximum pooling is performed to obtain an offset information adjusted convolutional coding pooling feature vector; the square root of the two norm of the offset information adjusted convolutional coding pooling feature vector is weighted and added with the one norm of the offset information adjusted convolutional coding pooling feature vector to obtain an offset information adjustment factor; the offset information adjusted position point multiplication feature vector and the square root of the length of the pet physiological indicator data text understanding feature vector are added by position to obtain an offset information adjusted intermediate vector; the offset information adjustment factor is used as the first weighting coefficient and the second weighted hyperparameter is used as the second weighting coefficient to calculate the position-weighted sum of the offset information adjusted position-added feature vector and the offset information adjusted intermediate vector to obtain a pet health abnormality early warning judgment feature vector.

[0050] Among them, the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector are adjusted based on the offset information corresponding to the feature fine-grainedness to obtain the pet health abnormality warning judgment feature vector, including:

[0051]

[0052] Among them, V 1 Represents the pet physiological indicator data text understanding feature vector, V 2 Represents the pet life behavior associated feature vector, represents positional subtraction, ⊙ represents positional multiplication, represents positional addition, conv1D represents a one-dimensional convolutional layer, MaxPool represents maximum pooling, V μ Indicates the offset information to adjust the convolutional coding pooling feature vector, ‖·‖ 1 represents a norm, ‖·‖ 2represents the bi-norm, [·] represents the cascade, α represents the first weighted hyperparameter, τ represents the offset information adjustment factor, L represents the length of the pet physiological index data text understanding feature vector, β represents the second weighted hyperparameter, V f Represents the pet health abnormality warning judgment feature vector.

[0053] In the technical solution of the present application, in the process of fusing the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector, the feature fine-grained dimension offset between the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector is taken into account, which may cause the fused pet health abnormality warning judgment feature vector to have local structural collapse. Based on this, in the technical solution of the present application, the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector are adjusted based on the offset information corresponding to the feature fine-grainedness.

[0054] Specifically, based on the restricted expression of the foreground-background structure distinction of the high-dimensional manifold of the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector by the semantic difference vector amplitude between the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector, the key feature fine-grained correspondence between the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector is modeled, and then the active fine-grained correspondence adjustment propagation is performed through the back-propagation displacement compensation information between the corresponding feature values ​​of the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector, so as to avoid the imbalance at the feature vector level of the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector in a focused manner, thereby synchronizing the micro dimensions of the two feature vectors, ensuring that in the fusion process, the information between the feature vectors can be effectively aligned and combined, thereby avoiding abnormal accumulation of local structures.

[0055] Furthermore, the pet health abnormality warning judgment feature vector integrates the pet's physiological indicator data and life behavior data. By comprehensively analyzing these features, it can reflect the complex information of the pet's health status. Although these feature vectors can provide rich health-related data, it may be very difficult to make accurate judgments directly from these high-dimensional data. Therefore, using classifiers to process these feature vectors is a key step in converting data into specific decisions. In the field of pet health warning, classifiers are trained to identify and classify different health states, such as normal or abnormal. This classification result provides a clear basis for judgment, and the system can decide whether to issue a health abnormality warning based on this result. If the classifier detects an abnormal signal, it will trigger the warning mechanism to notify the pet owner to take further actions, such as medical examination or adjusting the pet's living environment. Through the processing of the classifier, not only the automated analysis of a large amount of complex data is achieved, but also the efficiency and accuracy of health monitoring are ensured. The classifier can identify potential abnormal patterns during the processing process, reducing the need for manual intervention, while improving the response speed and accuracy of the warning system to various health problems.

[0056] In summary, the embodiment of the present application first obtains the pet's physiological indicators collected by the built-in integrated sensor of the pet collar and the pet's life behavior data collected by the smart platform, and then uses deep learning technology to perform feature extraction and correlation analysis on the two. Finally, a classifier is used to determine whether to issue an abnormal pet health warning, thereby identifying potential health problems and abnormalities so that timely measures can be taken, thereby helping pet owners to better take care of their pets' health.

[0057] As described above, the smart pet management system 100 for health monitoring according to the embodiment of the present application can be implemented in various terminal devices. In one example, the smart pet management system 100 for health monitoring can be integrated into the terminal device as a software module and / or a hardware module. For example, the smart pet management system 100 for health monitoring can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the smart pet management system 100 for health monitoring can also be one of the many hardware modules of the terminal device.

[0058] Alternatively, in another example, the smart pet management system 100 for health monitoring and the terminal device may also be separate devices, and the smart pet management system 100 for health monitoring may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0059] Figure 6 FIG. 1 is a flow chart of an intelligent pet management method for health monitoring according to an embodiment of the present application. Figure 6 As shown, according to the embodiment of the present application, the intelligent pet management method for health monitoring includes: S110, acquiring pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform; S120, extracting the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector from the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform; S130, judging whether to issue a pet health abnormality warning based on the pet physiological indicator data text understanding feature vector and the pet life behavior association feature vector.

[0060] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned smart pet management method for health monitoring have been referred to above. Figures 1 to 5 The invention has been introduced in detail in the description of the smart pet management system for health monitoring, and therefore, its repeated description will be omitted.

[0061] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0062] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0063] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0064] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0065] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0066] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.

Claims

1. An intelligent pet management system for health monitoring, characterized in that: include: The pet health detection data acquisition module is used to obtain the pet's physiological indicators collected by the integrated sensor built into the pet collar and the pet's life behavior data collected by the smart platform; A pet health detection data extraction module, used to extract pet physiological index data text understanding feature vectors and pet life behavior association feature vectors from the pet physiological indexes collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform; The pet health abnormality warning module is used to determine whether to issue a pet health abnormality warning based on the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector.

2. The smart pet management system for health monitoring according to claim 1, characterized in that: The pet health detection data extraction module includes: A pet physiological index feature extraction unit, used for extracting features of the pet physiological index collected by the built-in integrated sensor of the pet collar to obtain a text comprehension feature vector of the pet physiological index data; The pet life behavior data feature extraction unit is used to extract features from the pet life behavior data collected by the intelligent platform to obtain the pet life behavior related feature vector.

3. The smart pet management system for health monitoring according to claim 2 is characterized in that: The pet physiological index feature extraction unit comprises: A pet physiological index semantic feature encoding subunit, used for performing semantic feature encoding on the pet physiological index collected by the built-in integrated sensor of the pet collar to obtain a pet physiological index data vector sequence; The pet physiological indicator semantic understanding subunit is used to pass the pet physiological indicator data vector sequence through the pet physiological indicator data vector sequence semantic understanding Bi-LSTM model to obtain the pet physiological indicator data text understanding feature vector.

4. The smart pet management system for health monitoring according to claim 3 is characterized in that: The pet physiological index semantic feature encoding subunit includes: Passing the pet physiological indicators collected by the built-in integrated sensor of the pet collar through a pet physiological indicator semantic context encoder including an embedding layer to obtain a plurality of pet physiological indicator semantic feature vectors; The plurality of pet physiological indicator semantic feature vectors are arranged into the pet physiological indicator data vector sequence.

5. The smart pet management system for health monitoring according to claim 4 is characterized in that: The pet life behavior data feature extraction unit comprises: A pet life behavior data preprocessing subunit, used for preprocessing the pet life behavior data collected by the intelligent platform to obtain a pet life behavior correlation input matrix; The pet life behavior feature encoding subunit is used to pass the pet life behavior associated input matrix through a pet life behavior feature encoder to obtain the pet life behavior associated feature vector.

6. The smart pet management system for health monitoring according to claim 5, characterized in that: The pet life behavior data preprocessing subunit includes: Embedding the pet life behavior data collected by the intelligent platform into an encoder to obtain a plurality of pet life behavior word vector sequences; The multiple pet life behavior word vector sequences are arranged in two dimensions to obtain the pet life behavior association input matrix.

7. The smart pet management system for health monitoring according to claim 6, characterized in that: The pet health abnormality warning module includes: A pet health data feature fusion unit is used to adjust the pet physiological index data text understanding feature vector and the pet life behavior related feature vector based on the offset information corresponding to the fine-grained features to obtain a pet health abnormality warning judgment feature vector; The pet health abnormality warning classification judgment unit is used to pass the pet health abnormality warning judgment feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a pet health abnormality warning.

8. The smart pet management system for health monitoring according to claim 7, characterized in that: The pet health data feature fusion unit includes: Calculating the position difference between the pet physiological index data text understanding feature vector and the pet life behavior related feature vector to obtain an offset information adjusted differential feature vector; Calculate the position point multiplication between the pet physiological index data text understanding feature vector and the pet life behavior related feature vector to obtain the offset information adjusted position point multiplication feature vector; Calculate the positional addition between the pet physiological index data text understanding feature vector and the pet life behavior related feature vector to obtain the offset information adjusted positional addition feature vector; The offset information adjustment differential feature vector, the offset information adjustment position point product feature vector and the offset information adjustment position addition feature vector are cascaded and input into a one-dimensional convolution layer and maximum value pooling is performed to obtain an offset information adjustment convolutional coding pooling feature vector; The square root value of the two-norm of the offset information adjusted convolutional coding pooling feature vector is weighted and added to the one-norm of the offset information adjusted convolutional coding pooling feature vector to obtain an offset information adjustment factor; The offset information adjustment position dot product feature vector and the square root of the length of the pet physiological index data text understanding feature vector are added according to the position to obtain an offset information adjustment intermediate vector; Using the offset information adjustment factor as the first weighting coefficient and the second weighted hyperparameter as the second weighting coefficient, the position-weighted sum of the offset information adjustment position-added feature vector and the offset information adjustment intermediate vector is calculated to obtain the pet health abnormality warning judgment feature vector.

9. An intelligent pet management method for health monitoring, characterized in that: include: Obtain pet physiological indicators collected by the integrated sensors in the pet collar and pet life behavior data collected by the smart platform; Extracting pet physiological indicator data text understanding feature vectors and pet life behavior association feature vectors from the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform; Based on the pet physiological indicator data text understanding feature vector and the pet life behavior related feature vector, it is determined whether to issue a pet health abnormality warning.

10. The smart pet management method for health monitoring according to claim 9, characterized in that: Extracting pet physiological indicator data text understanding feature vectors and pet life behavior related feature vectors from the pet physiological indicators collected by the built-in integrated sensor of the pet collar and the pet life behavior data collected by the intelligent platform includes: Extracting features of the pet's physiological indicators collected by the built-in integrated sensor of the pet collar to obtain a text comprehension feature vector of the pet's physiological indicator data; Feature extraction is performed on the pet life behavior data collected by the intelligent platform to obtain the pet life behavior related feature vector.

Images