Pet state analysis method and system based on remote monitoring

Abstract

The invention provides a pet state analysis method and system based on remote monitoring, and relates to the technical field of pet state evaluation. The method comprises the following steps: collecting multi-source historical health monitoring data about a pet, and determining a plurality of multi-source combination states of the pet; performing physiological behavior joint analysis based on morphological change on the plurality of multi-source combination states of the pet, generating a plurality of physiological behavior joint modes of the pet, and determining a recessive compensation mode in each physiological behavior joint mode; constructing a state association map of the pet, and performing anomaly analysis on the multi-source real-time health monitoring data of the pet to determine a state anomaly event; and determining a target joint mode to which the pet belongs according to the multi-source real-time health monitoring data, performing compensation stability analysis on the pet according to a recessive compensation mode of the target joint mode, and generating a compensation imbalance evaluation result of the pet. According to the invention, accurate evaluation of the health state of the pet in the transportation process is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of pet status assessment, and in particular to a pet status analysis method and system based on remote monitoring. Background Art

[0002] With the widespread application of IoT devices, pet health management has gradually shifted from traditional manual monitoring to intelligent management based on remote monitoring and data analysis. During the transportation of pets, real-time collection of pet health data such as body temperature, heart rate, activity level, ambient temperature and humidity, etc., relying on wearable devices, environmental sensors and other technologies, has become an important means of ensuring pet health. These real-time data can be used to dynamically evaluate the health status of pets and provide a basis for risk warning and health intervention decisions during transportation.

[0003] However, some pet remote health monitoring technologies use anomaly detection methods based on fixed thresholds or single indicators, ignoring the complex environmental adaptability and behavioral compensation patterns of pets during transportation. During transportation, pets may experience short-term physiological changes, such as increased body temperature or increased heart rate, due to external environmental pressures such as high temperature, vibration, and noise. These changes are usually accompanied by behavioral changes in pets, such as reduced activity and prolonged static time as self-regulation. However, these behavioral changes will mask physiological abnormalities to a certain extent, causing the health status of pets to be misjudged as normal, and even delay the discovery of potential health risks. In the early stages of health problems, pets use behavioral adjustments to mask physiological stress, which can easily delay or ignore abnormal signals. Developing an intelligent system that can accurately analyze the health status of pets during transportation, identify abnormal conditions, and assist in decision-making is of great practical significance for improving the health protection of pets during transportation. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a pet status analysis method and system based on remote monitoring. Based on the pet's own historical normal behavior pattern data and combined with real-time data during transportation, it dynamically evaluates the changes in the pet's health status and identifies implicit compensatory behaviors. It can accurately evaluate the level of health changes of the pet under short-term stress.

[0005] To achieve the above object, the present invention provides a pet status analysis method based on remote monitoring in a first aspect, comprising: Collect multi-source historical health monitoring data about the pet, including the pet's historical behavior data, historical physiological data and historical environmental data, extract state features from the multi-source historical health monitoring data, and determine multiple multi-source combined states of the pet; Performing a physiological behavior joint analysis based on morphological changes on multiple multi-source combination states of the pet, generating multiple physiological behavior joint patterns of the pet, extracting compensation features from multiple multi-source combination states in each physiological behavior joint pattern, and determining the implicit compensation pattern in each physiological behavior joint pattern; A state association map of the pet is constructed according to the timestamp information corresponding to each multi-source combination state, and an abnormal analysis is performed on the multi-source real-time health monitoring data of the pet during transportation based on the state association map to determine the state abnormal events in the multi-source real-time health monitoring data; The state feature sequence of abnormal state events is extracted from multi-source real-time health monitoring data, and the physiological behavior is matched with the state feature sequence through multiple physiological behavior joint patterns to determine the target joint pattern to which the pet belongs. The compensatory stability analysis of the pet is performed according to the implicit compensation pattern of the target joint pattern to generate the pet's compensatory imbalance assessment result.

[0006] Preferably, a physiological behavior joint analysis based on morphological changes is performed on multiple multi-source combined states of the pet to generate multiple physiological behavior joint patterns of the pet, including: A combined state sequence of multi-source historical health monitoring data is constructed according to the timestamp information of the multi-source combined state, and a sliding window analysis is performed on the combined state sequence. Multiple local state fluctuation parameters of the combined state sequence are extracted according to the transfer characteristics of the multiple multi-source combined states in each window. The transfer paths of the multiple multi-source combined states in each window are smoothed and analyzed to obtain multiple local path curvatures of the combined state sequence. A physiological and behavioral collaborative analysis is performed on the multiple multi-source combined states in each window to obtain multiple local modal synchronization parameters of the combined state sequence. Multiple local state feature vectors corresponding to the combined state sequence are constructed, and the multiple local state feature vectors are clustered to generate multiple physiological and behavioral joint patterns of the pet.

[0007] Preferably, performing abnormal analysis on the multi-source real-time health monitoring data of the pet during transportation based on the state association graph to determine abnormal state events in the multi-source real-time health monitoring data includes: Determine multiple real-time combined states in multi-source real-time health monitoring data, construct multiple real-time state paths of multi-source real-time health monitoring data, perform abnormal analysis on multiple real-time state paths according to the state association map, determine abnormal transfer paths, and locate abnormal state events based on the abnormal transfer paths; The state association graph uses multiple multi-source combination states as graph nodes, calculates the transition probability between multi-source combination states according to the timestamp information corresponding to each multi-source combination state, constructs the directed edges in the state association graph according to the transition probability, and determines multiple reference transfer paths in the state association graph through the path mining algorithm. If the real-time state path does not match any reference transfer path, it is marked as an abnormal transfer path.

[0008] Preferably, determining the implicit compensation pattern in each physiological-behavioral joint pattern includes: Multiple local state fragments are cut out from multiple local state feature vectors included in the joint pattern of physiological behavior, and the pattern frequency and pattern coverage of each local state fragment are calculated. The fragment co-occurrence parameters of each local state fragment are determined according to the pattern frequency and pattern coverage, and multiple candidate state fragments are determined. Based on the multiple candidate state fragments, multiple local state feature vectors included in the implicit compensatory pattern in the joint pattern of physiological behavior are extracted.

[0009] Preferably, the pattern coverage of a local state segment is the ratio of the total number of physiological behavior joint patterns to the number of physiological behavior joint patterns that contain the local state segment.

[0010] Preferably, the state association graph is subjected to path mining by using a Viterbi algorithm to determine a plurality of reference transfer paths in the state association graph.

[0011] A second aspect of the present invention provides a pet status analysis system based on remote monitoring, which is used to implement the above-mentioned pet status analysis method based on remote monitoring, comprising: A state feature extraction module is used to collect multi-source historical health monitoring data about the pet, including the pet's historical behavior data, historical physiological data and historical environmental data, extract state features from the multi-source historical health monitoring data, and determine multiple multi-source combined states of the pet; An implicit compensation analysis module is used to perform a physiological behavior joint analysis based on morphological changes on multiple multi-source combination states of the pet, generate multiple physiological behavior joint patterns of the pet, extract compensation features from multiple multi-source combination states in each physiological behavior joint pattern, and determine an implicit compensation pattern in each physiological behavior joint pattern; A real-time anomaly analysis module is used to construct a state association map of the pet according to the timestamp information corresponding to each multi-source combination state, and to perform an anomaly analysis on the multi-source real-time health monitoring data of the pet during transportation based on the state association map to determine the state anomaly events in the multi-source real-time health monitoring data; The real-time status assessment module is used to extract the status feature sequence of abnormal status events from multi-source real-time health monitoring data, match the status feature sequence with physiological behavior through multiple physiological behavior joint patterns, determine the target joint pattern to which the pet belongs, and perform compensatory stability analysis on the pet according to the implicit compensation pattern of the target joint pattern to generate the pet's compensatory imbalance assessment result.

[0012] The present invention has the following beneficial effects: The present invention constructs a joint feature vector based on fluctuation intensity, path curvature and modal synchronization through time series discrete coding and morphological analysis of multi-source health monitoring data of pets, and quantifies the dynamic change relationship of pets' physiological behavior patterns; captures the characteristic change law of pets' self-regulatory behavior under environmental pressure through cross-window fragment co-occurrence parameter extraction and path transfer probability modeling, mines abnormal state change phenomena of pets through state association maps, and conducts abnormal state assessment of pets in combination with abnormal detection of real-time transfer paths, thereby realizing accurate identification of complex compensatory failures during pet transportation, effectively distinguishing healthy adaptive behaviors from potential health risks, improving the reliability and warning timeliness of pet health status assessment under stress, providing decision-making basis for remote pet status monitoring, and providing good health protection for pet transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The figure is a flowchart of an exemplary pet status analysis method based on remote monitoring in an embodiment of the present invention.

[0014] Figure 2 This is a structural diagram of an exemplary pet status analysis system based on remote monitoring in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to make those skilled in the art better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0016] See also Figure 1 , a pet status analysis method based on remote monitoring provided by an embodiment of the present invention comprises the following steps: Step S1, collecting multi-source historical health monitoring data about the pet, including the pet's historical behavior data, historical physiological data and historical environmental data, extracting state features from the multi-source historical health monitoring data, and determining multiple multi-source combined states of the pet.

[0017] It is worth noting that during the transportation of pets, various smart sensors such as temperature and humidity sensors, microphones, light sensors, etc. can be used to remotely collect various health data of pets in transportation devices such as pet boxes, so as to realize remote monitoring of the pet's status. Compared with the environment in which pets live in daily life, during transportation, pets may experience short-term physiological changes due to external environmental pressures such as high temperature, vibration, and noise, and pets will also conduct related self-conditioning, resulting in some abnormal phenomena exhibited by pets being concealed to a certain extent. It is necessary to deeply analyze the multi-source data of pets during transportation and timely discover potential abnormal signals to improve the health protection of pets during transportation.

[0018] Among them, the multi-source historical health monitoring data about pets is specifically the historical normal data corresponding to the individual's daily performance, which includes the pet's normal self-regulation of changes in the external environment under environmental disturbances, and can be used to model the daily health status of pets. Multi-source historical health monitoring data includes but is not limited to the pet's historical behavior data, historical physiological data, and historical environmental data, etc. The historical behavior data is used to characterize the activity of the pet, such as the acceleration data, activity pattern and other behavior-related information of the pet collected by the acceleration sensor, the historical physiological data involves, for example, the pet's body temperature, heart rate, respiratory rate and other physical sign-related information, and the historical environmental data includes, for example, the temperature, humidity, noise and other external stress factors of the pet's environment. By dynamically discretizing and encoding the multi-source historical health monitoring data, for example, dividing the heart rate, body temperature, environmental noise, etc. into multiple discrete coding symbols, the discrete coding symbols corresponding to the multiple data related to the health of the pet in a certain period of time are combined, for example, the activity of the pet in a certain window is distributed in the M3 interval, the heart rate is in the A1 interval, the body temperature is in the T2 interval, the environmental temperature is in the H4 interval, etc., and the multi-source combination state corresponding to the pet in different time periods is obtained.

[0019] Step S2, performing a physiological behavior joint analysis based on morphological changes on the pet's multiple multi-source combination states, generating multiple physiological behavior joint patterns of the pet, extracting compensatory features from the multiple multi-source combination states in each physiological behavior joint pattern, and determining an implicit compensation pattern in each physiological behavior joint pattern.

[0020] Among them, in the process of mining the joint patterns of physiology and behavior, we focus on the evolutionary characteristics of the pet's comprehensive morphology, integrate the overall physiological and behavioral performance of the pet, capture the dynamic process characteristics of the pet's state changes, and explore the uniqueness of individual physiological patterns.

[0021] In this process, the combined state sequence of multi-source historical health monitoring data is first constructed according to the timestamp information of the multi-source combined state, and the combined state sequence is subjected to sliding window analysis to extract multiple deep features in different windows.

[0022] Specifically, multiple local state fluctuation parameters of the combined state sequence are extracted according to the transfer characteristics of multiple multi-source combined states in each window. In this process, , Indicates the number of state types, that is, the number of types of multi-source combination states involved in the window. It represents the number of state transitions, that is, the number of state transitions involving multiple sources in the window. The local state fluctuation parameter is used to quantify the severity of state changes per unit time to avoid frequent switching between a few states being mistaken for high fluctuations.

[0023] A smoothing analysis is performed on the transfer paths of multiple multi-source combination states in each window, and multiple local path curvatures of the combination state sequence are calculated, including the angle between the directed edges corresponding to any two adjacent multi-source combination states in the window. The average of the multiple angles in the window is taken as the local path curvature of the multiple multi-source combination states in the window, which characterizes the smoothness or mutation of the physiological state changes, that is, considering that the pet may mask physiological abnormalities by behavioral stillness, but the path turning will expose the internal disorder.

[0024] A physiological and behavioral synergy analysis is performed on multiple multi-source combined states in each window, and multiple local modal synchronization parameters of the combined state sequence are calculated, including determining the maximum number of behavioral and physiological changes of the pet within the window, and then analyzing the number of synergistic changes between the pet's behavior and physiology within a specific period of time. The ratio of the number of synergistic changes between behavior and physiology to the maximum number of changes between the pet's behavior and physiology is used as the local modal synchronization parameter in the window to describe the synergistic efficiency between behavioral regulation and physiological response. A higher ratio indicates that the pet's compensatory behavior is a healthy compensation, such as a decrease in activity accompanied by a decrease in heart rate. A lower ratio indicates a failure of compensation, such as cessation of activity but a continuous increase in heart rate.

[0025] Finally, the local state feature vectors corresponding to the combined state sequence in different windows are obtained through the above-mentioned feature construction, which are used to describe the overall morphological characteristics of the pet's physiological-behavioral patterns in different windows, and then multiple local state feature vectors are clustered, for example, through the DBSCAN algorithm to generate multiple physiological-behavioral joint patterns of the pet. Different physiological-behavioral joint patterns represent the behavioral-physiological pattern categories of the pet under different macroscopic perspectives in a normal state, such as the overall performance of the pet in a quiet state, sleeping state, exercise state, etc.

[0026] On the basis of multiple physiological and behavioral joint patterns, we further consider physiological compensatory behaviors caused by some minor external pressures in real life, such as a temporary increase in heart rate and a decrease in activity, that is, the actual performance of pets under the normal adaptation range of healthy individuals. The above physiological and behavioral joint patterns represent the overall performance of pets in a specific mode, but they also include some data related to the normal compensatory behaviors of pets. For example, in quiet mode, some data may be affected by certain external environments, such as excessive noise, causing the pet to temporarily change to a quiet state, and at the same time, it is affected by minor influences, causing the heart rate to increase to a certain extent compared to the normal state. For these implicit compensatory behaviors that are stress changes but are still in the normal range because they have not reached the pathological level, further mining of implicit compensation patterns is carried out.

[0027] The extraction of implicit compensatory patterns contained in different physiological behavior joint patterns includes cutting out multiple local state fragments from multiple local state feature vectors contained in the physiological behavior joint pattern, that is, determining multiple micro-combinations, such as a combination of fluctuation characteristics and modal synchronization characteristics, a combination of path transfer and modal synchronization, etc., and calculating the pattern frequency and pattern coverage of each local state fragment, wherein the pattern frequency is specifically the frequency of occurrence of the local state fragment in the physiological behavior joint pattern to which it belongs, and the pattern coverage is the ratio of the total number of physiological behavior joint patterns to the corresponding number of multiple physiological behavior joint patterns containing the local state fragment, that is, the state fragments that appear frequently in a specific pattern but are relatively rare in other patterns are representative compensatory behavior patterns, and finally the pattern frequency of the local state fragments is compared. The pattern coverage of local state fragments is weightedly corrected to obtain the fragment co-occurrence parameters of each local state fragment, and representative local state fragments are screened out. For example, the local state fragments whose fragment co-occurrence parameters are greater than the preset co-occurrence threshold are marked as candidate state fragments, and according to the total number of candidate state fragments involved in the local state feature vector, representative local state feature vectors are selected to characterize the implicit compensation pattern in the physiological and behavioral joint pattern, thereby realizing the mining of the implicit compensation pattern in the physiological and behavioral joint pattern. On the basis of discarding the conventional features with broad consensus in the physiological and behavioral joint pattern, the features with consensus in the pattern are used as the implicit metabolic pattern exhibited by the pet in a normal state, which is used to characterize the normal stress state of the pet in response to external environmental stimuli under normal performance.

[0028] Step S3: construct a state association map of the pet according to the timestamp information corresponding to each multi-source combination state, perform an abnormal analysis on the multi-source real-time health monitoring data of the pet during transportation based on the state association map, and determine the state abnormal events in the multi-source real-time health monitoring data.

[0029] Among them, the state association graph is used to describe the state transition characteristics of the pet. For the construction of the state association graph, multiple multi-source combination states are used as graph nodes of the state association graph. Each multi-source combination state is unique in the graph, and the transition probability between the multi-source combination states is calculated in combination with the timestamp information corresponding to the multi-source combination state. For example, the total number of times the current multi-source combination state is transformed into the remaining multi-source combination states in history is counted, and the transition probability between any two multi-source combination states is calculated according to the number of transitions corresponding to each multi-source combination state, including the corresponding transition probabilities in bidirectional transfers. Finally, the directed edges in the state association graph are constructed in combination with the directionality of the transition probability, and multiple reference transfer paths in the state association graph are determined by a path mining algorithm. For example, the reference transfer path is determined by path mining through the Viterbi algorithm to reflect the most likely evolution law of physiological behavior.

[0030] For the multi-source real-time health monitoring data collected in real time during the transportation of pets, first determine the multiple real-time combination states in the multi-source real-time health monitoring data, associate any two adjacent real-time combination states, construct multiple real-time state paths of the multi-source real-time health monitoring data, and perform abnormal analysis on the multiple real-time state paths according to the state association map. If a real-time state path does not match any reference transfer path in the state association map, it is marked as an abnormal transfer path. In this way, some abnormal performances of pets in the transportation process are excavated, and in this way, the abnormal transfer path of the pet is determined and the abnormal state event is located based on the abnormal transfer path, which is used to characterize the abnormal phenomenon of the pet compared with the normal performance during the transportation process.

[0031] Step S4, extracting the state feature sequence of abnormal state events from multi-source real-time health monitoring data, performing physiological behavior matching on the state feature sequence through multiple physiological behavior joint patterns, determining the target joint pattern to which the pet belongs, performing compensatory stability analysis on the pet according to the implicit compensation pattern of the target joint pattern, and generating a compensatory imbalance assessment result for the pet.

[0032] Among them, for the detected abnormal state event, the relevant state feature sequence is extracted from the real-time health monitoring data. Specifically, the target data of this event can be extracted from the real-time health monitoring data according to the abnormal transfer path involved in the abnormal state event and a certain time range before and after the time information of the abnormal transfer path. The selection of the target data can be reasonably set according to the actual performance of the pet, and it is not specifically limited in this embodiment. Thus, multiple state feature sequences of the pet regarding the abnormal state event are generated according to the target data, that is, the state feature sequences corresponding to the target data, and refer to the construction of the aforementioned combined state sequence, which is used to characterize the deep state characteristics of the pet in the abnormal state event, and then the state feature sequences are respectively matched with the physiological behavior joint mode to determine the target joint mode to which the pet belongs, that is, considering that the pet exhibits certain compensatory behaviors in the current state, it may be difficult to accurately identify the current state mode of the pet simply based on, for example, behavioral thresholds. After determining the target joint mode to which the pet belongs, a compensatory stability analysis is performed on the pet according to the implicit compensation mode of the target joint mode, including determining the offset characteristics of the pet's current implicit compensation mode from the target joint mode, which may be specifically the distance from the cluster center of the implicit compensation mode. The cluster center of the implicit compensation mode may be determined based on the average level corresponding to multiple local state feature vectors contained therein. At the same time, based on the offset direction between the target joint mode and the implicit compensation mode of the target joint mode, a state evaluation is performed on the pet's current offset characteristics.

[0033] For example, a pet may show a stress response of reduced activity after being stimulated by external stimuli, and its heart rate may be somewhat higher than normal levels. By analyzing the changes in different health monitoring items between the target joint mode and the implicit compensation mode of the target joint mode, for example, determining the reference levels of body temperature, heart rate, and activity level in the target joint mode and the corresponding implicit compensation mode, the pet's current offset for different health monitoring items is determined, and the pet's current compensatory imbalance score is calculated according to the scoring weights corresponding to the different health monitoring items determined in advance based on expert experience, thereby generating a compensatory imbalance assessment result for the pet. Some pets are timid in nature, and their heart rate will increase significantly after being stimulated by external sounds. In the process of calculating the compensatory imbalance score, the score weight corresponding to the heart rate offset can be increased, and the focus can be on the pet's current heart rate changes. In this way, the health status of the pet during transportation can be accurately analyzed and evaluated, and abnormal performance of the pet compared to the normal state can be identified. By analyzing the changes in the pet's compensatory imbalance score in real time, it can be assisted in evaluating whether the pet can restore its normal health state through self-adjustment, and the changing trend of the pet's health level under external pressure can be analyzed. Health intervention decision support is provided for the changes in the pet's state during transportation, and health protection is provided for the pet during transportation.

[0034] Based on the same concept as the above-mentioned pet status analysis method based on remote monitoring, an embodiment of the present invention further provides a pet status analysis system based on remote monitoring.

[0035] See also Figure 2 , a pet status analysis system based on remote monitoring provided by an embodiment of the present invention includes: The state feature extraction module 01 is used to collect multi-source historical health monitoring data about the pet, including the pet's historical behavior data, historical physiological data and historical environmental data, extract state features from the multi-source historical health monitoring data, and determine multiple multi-source combined states of the pet; Implicit compensation analysis module 02, used for performing physiological behavior joint analysis based on morphological changes on multiple multi-source combination states of the pet, generating multiple physiological behavior joint patterns of the pet, extracting compensation features from multiple multi-source combination states in each physiological behavior joint pattern, and determining implicit compensation patterns in each physiological behavior joint pattern; The real-time abnormality analysis module 03 is used to construct a state association map of the pet according to the timestamp information corresponding to each multi-source combination state, and to perform abnormality analysis on the multi-source real-time health monitoring data of the pet during transportation based on the state association map to determine abnormal state events in the multi-source real-time health monitoring data; The real-time status assessment module 04 is used to extract the status feature sequence of abnormal status events from multi-source real-time health monitoring data, perform physiological behavior matching on the status feature sequence through multiple physiological behavior joint patterns, determine the target joint pattern to which the pet belongs, perform compensation stability analysis on the pet according to the implicit compensation pattern of the target joint pattern, and generate a compensation imbalance assessment result for the pet.

[0036] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A pet status analysis method based on remote monitoring, characterized in that: include: Collect multi-source historical health monitoring data about the pet, including the pet's historical behavior data, historical physiological data and historical environmental data, extract state features from the multi-source historical health monitoring data, and determine multiple multi-source combined states of the pet; Performing a physiological behavior joint analysis based on morphological changes on multiple multi-source combination states of the pet, generating multiple physiological behavior joint patterns of the pet, extracting compensation features from multiple multi-source combination states in each physiological behavior joint pattern, and determining the implicit compensation pattern in each physiological behavior joint pattern; A state association map of the pet is constructed according to the timestamp information corresponding to each multi-source combination state, and an abnormal analysis is performed on the multi-source real-time health monitoring data of the pet during transportation based on the state association map to determine the state abnormal events in the multi-source real-time health monitoring data; The state feature sequence of abnormal state events is extracted from multi-source real-time health monitoring data, and the physiological behavior is matched with the state feature sequence through multiple physiological behavior joint patterns to determine the target joint pattern to which the pet belongs. The compensatory stability analysis of the pet is performed according to the implicit compensation pattern of the target joint pattern to generate the pet's compensatory imbalance assessment result.

2. A pet status analysis method based on remote monitoring according to claim 1, characterized in that: Perform a physiological and behavioral joint analysis based on morphological changes on multiple multi-source combined states of pets to generate multiple physiological and behavioral joint patterns of pets, including: A combined state sequence of multi-source historical health monitoring data is constructed according to the timestamp information of the multi-source combined state, and a sliding window analysis is performed on the combined state sequence. Multiple local state fluctuation parameters of the combined state sequence are extracted according to the transfer characteristics of the multiple multi-source combined states in each window. The transfer paths of the multiple multi-source combined states in each window are smoothed and analyzed to obtain multiple local path curvatures of the combined state sequence. A physiological and behavioral collaborative analysis is performed on the multiple multi-source combined states in each window to obtain multiple local modal synchronization parameters of the combined state sequence. Multiple local state feature vectors corresponding to the combined state sequence are constructed, and the multiple local state feature vectors are clustered to generate multiple physiological and behavioral joint patterns of the pet.

3. A pet status analysis method based on remote monitoring according to claim 1, characterized in that: Based on the state association graph, the multi-source real-time health monitoring data of the pet during transportation is analyzed for abnormalities, and the abnormal state events in the multi-source real-time health monitoring data are determined, including: Determine multiple real-time combined states in multi-source real-time health monitoring data, construct multiple real-time state paths of multi-source real-time health monitoring data, perform abnormal analysis on multiple real-time state paths according to the state association map, determine abnormal transfer paths, and locate abnormal state events based on the abnormal transfer paths; The state association graph uses multiple multi-source combination states as graph nodes, calculates the transition probability between multi-source combination states according to the timestamp information corresponding to each multi-source combination state, constructs the directed edges in the state association graph according to the transition probability, and determines multiple reference transfer paths in the state association graph through the path mining algorithm. If the real-time state path does not match any reference transfer path, it is marked as an abnormal transfer path.

4. A pet status analysis method based on remote monitoring according to claim 2, characterized in that: Identify implicit compensatory patterns within each physiological-behavioral joint pattern, including: Multiple local state fragments are cut out from multiple local state feature vectors included in the joint pattern of physiological behavior, and the pattern frequency and pattern coverage of each local state fragment are calculated. The fragment co-occurrence parameters of each local state fragment are determined according to the pattern frequency and pattern coverage, and multiple candidate state fragments are determined. Based on the multiple candidate state fragments, multiple local state feature vectors included in the implicit compensatory pattern in the joint pattern of physiological behavior are extracted.

5. A pet status analysis method based on remote monitoring according to claim 4, characterized in that: The pattern coverage of a local state segment is the ratio of the total number of physiological behavior joint patterns to the number of physiological behavior joint patterns that contain the local state segment.

6. A pet status analysis method based on remote monitoring according to claim 3, characterized in that: The Viterbi algorithm is used to perform path mining on the state association graph to determine multiple reference transfer paths in the state association graph.

7. A pet status analysis system based on remote monitoring, characterized in that: The system is used to implement a pet status analysis method based on remote monitoring as described in any one of claims 1 to 6, comprising: A state feature extraction module is used to collect multi-source historical health monitoring data about the pet, including the pet's historical behavior data, historical physiological data and historical environmental data, extract state features from the multi-source historical health monitoring data, and determine multiple multi-source combined states of the pet; An implicit compensation analysis module is used to perform a physiological behavior joint analysis based on morphological changes on multiple multi-source combination states of the pet, generate multiple physiological behavior joint patterns of the pet, extract compensation features from multiple multi-source combination states in each physiological behavior joint pattern, and determine an implicit compensation pattern in each physiological behavior joint pattern; A real-time anomaly analysis module is used to construct a state association map of the pet according to the timestamp information corresponding to each multi-source combination state, and to perform an anomaly analysis on the multi-source real-time health monitoring data of the pet during transportation based on the state association map to determine the state anomaly events in the multi-source real-time health monitoring data; The real-time status assessment module is used to extract the status feature sequence of abnormal status events from multi-source real-time health monitoring data, match the status feature sequence with physiological behavior through multiple physiological behavior joint patterns, determine the target joint pattern to which the pet belongs, and perform compensatory stability analysis on the pet according to the implicit compensation pattern of the target joint pattern to generate the pet's compensatory imbalance assessment result.

Images