Pet health management method and device and storage medium

Abstract

The invention discloses a pet health management method and device and a storage medium, the method is applied to a pet feeder, and the pet health management method comprises the following steps: obtaining current health data of a target pet; performing health analysis processing according to the current health data and a pre-constructed health baseline model to obtain health risk information of the target pet; in response to the health risk information representing that the target pet has a health risk, determining a target nutrition scheme from preset nutrition management schemes according to the health risk information; and putting the nutrition supplement corresponding to the target nutrition scheme. According to the scheme, pet health can be accurately and effectively managed through the health baseline model.

Description

Technical Field

[0001] This application relates to the field of pet product technology, and in particular to a pet health management method, device, and storage medium. Background Technology

[0002] The pet economy is a market economy system derived from pets, encompassing the entire industry chain services such as food, medical care, and boarding. This economic model has developed rapidly as the role of pets has shifted from "animal companions" to "family members."

[0003] Currently, many smart pet care devices have been developed for pet products, such as smart feeding devices. When pet owners are away for extended periods, they can set up feeding schedules so that the smart feeding devices can automatically add food and water to their pets, thereby reducing the stress of feeding their pets and increasing their overall pet ownership happiness.

[0004] Modern automatic feeding methods focus on dispensing food at set times and in measured amounts. While this can solve a pet's hunger problem, it can easily lead pet owners to neglect their pet's eating habits and health, resulting in malnutrition and health problems. Summary of the Invention

[0005] This application provides at least one pet health management method, apparatus, device, and computer-readable storage medium.

[0006] The first aspect of this application provides a pet health management method applied to a pet feeder, comprising: acquiring current health data of a target pet; performing health analysis processing based on the current health data and a pre-constructed health baseline model to obtain health risk information of the target pet; responding to the health risk information indicating that the target pet has a health risk, determining a target nutrition plan from a preset nutrition management plan based on the health risk information; and administering nutritional supplements corresponding to the target nutrition plan.

[0007] In one embodiment, obtaining the current health data of the target pet includes: in response to detecting the eating behavior of the target pet, collecting multimodal data of the target pet during eating; and performing time-series fusion processing on the multimodal data to obtain the current health data of the target pet.

[0008] In one embodiment, health analysis is performed based on the current health data and a pre-built health baseline model to obtain the health risk information of the target pet. This includes: comparing the current health data with the health benchmark data in the health baseline model to obtain the index deviation between the current health data and the health benchmark data; and determining the health risk information based on the index deviation within a preset time period.

[0009] In one embodiment, the step of determining a target nutrition plan from a preset nutrition management plan in response to the health risk information indicating that the target pet has a health risk includes: in response to the health risk information indicating the existence of multiple abnormal data indicators, determining the target risk type of the health risk information based on the multiple abnormal data indicators; the deviation of each abnormal data indicator is greater than the deviation threshold corresponding to each abnormal data indicator; and determining the target nutrition plan from each nutrition management plan based on the mapping weight between the target risk type and each nutrition management plan.

[0010] In one embodiment, after administering the nutritional supplements corresponding to the target nutrition plan, the method further includes: analyzing the subsequent changes in the deviation of the indicator; and adjusting the mapping weight between the target risk type and the target nutrition plan based on the subsequent changes.

[0011] In one embodiment, after administering the nutritional supplements corresponding to the target nutrition plan, the method further includes: acquiring subsequent health data of the target pet; and, in response to the confirmation of the target risk type, incrementally updating the health baseline model based on the subsequent health data and a preset baseline update weight.

[0012] In one embodiment, health analysis processing is performed based on the current health data and a pre-built health baseline model to obtain the health risk information of the target pet, including: performing health analysis processing based on the current health data and the health baseline model to obtain initial health information; acquiring current environmental data corresponding to the time sequence of the current health data; and performing context correction processing on the initial health information based on the current environmental data to obtain the health risk information.

[0013] In one embodiment, before performing health analysis processing based on the current health data and a pre-constructed health baseline model to obtain the health risk information of the target pet, the method further includes: acquiring initial health data of the target pet; performing health screening processing on the target pet based on the initial health data and a preset health threshold to obtain initial screening results of the target pet; and constructing the health baseline model based on the initial health data in response to the initial screening results characterizing the health of the target pet.

[0014] A second aspect of this application provides a pet health management device applied to a pet feeder, comprising: an acquisition module for acquiring current health data of a target pet; a health analysis module for performing health analysis processing based on the current health data and a pre-built health baseline model to obtain health risk information of the target pet; a nutrition management module for determining a target nutrition plan from a preset nutrition management plan in response to the health risk information indicating that the target pet has a health risk; and a supplement dispensing module for dispensing nutritional supplements corresponding to the target nutrition plan.

[0015] A third aspect of this application provides an electronic device, including a memory and a processor, wherein the processor is used to execute program instructions stored in the memory to implement the above-described pet health management method.

[0016] The fourth aspect of this application provides a computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, implement the above-described pet health management method.

[0017] The above-described solution acquires the target pet's current health data and performs health analysis against a pre-built health baseline model. This analysis yields the deviation between the current health data and the baseline model, thereby determining the target pet's health risk information. Based on this risk information, it can be determined whether the target pet faces any health risks. If a health risk is identified, a target nutritional plan can be selected from a pre-defined nutritional management program. Furthermore, it can be chosen to mix the corresponding nutritional supplements with the target pet's food. This allows for continuous long-term acquisition and analysis of the target pet's health data, timely assessment of health risks, generation of health alerts, and the addition of appropriate nutritional supplements during feeding, thus achieving rational and effective pet health management.

[0018] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0020] Figure 1 This is a flowchart illustrating an exemplary embodiment of the pet health management method of this application; Figure 2 This is a block diagram illustrating a pet health management device in an exemplary embodiment of this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application; Figure 4 This is a schematic diagram of the structure of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0021] The solution of this embodiment will now be described in detail with reference to the accompanying drawings.

[0022] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0023] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this document means two or more. Moreover, the term "at least one" in this document means any combination of at least two of any one or more of a plurality of objects. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0024] To facilitate understanding, one of the applicable scenarios of this application will be illustrated by example.

[0025] The pet economy is a market economy system derived from pets, encompassing the entire industry chain services such as food, medical care, and boarding. This economic model has developed rapidly as the role of pets has shifted from "animal companions" to "family members."

[0026] Currently, many smart pet care devices have been developed in the pet supplies industry, such as smart feeding devices, smart water dispensers, smart toilets, smart wearable devices, smart cameras, and so on.

[0027] Taking smart feeding devices as an example, when pet owners are away for extended periods, they can set up a feeding schedule to have the smart feeding device automatically add food and water to their pets, thereby reducing the pressure of feeding them and increasing their happiness in pet ownership.

[0028] Modern automatic feeding methods focus on dispensing food at set times and in measured amounts. While this can solve a pet's hunger problem, it can easily lead pet owners to neglect their pet's eating habits and health, resulting in malnutrition and health problems.

[0029] Please see Figure 1 , Figure 1This is a flowchart illustrating an exemplary embodiment of the pet health management method of this application. The pet health management method of this application can be applied to pet feeders. The pet feeder of this application at least supports the functions of dispensing pet food and nutritional supplements. In addition, it can optionally be expanded to include other functions such as pet water, which are not limited here. Specifically, it may include the following steps: Step S110: Obtain the current health data of the target pet.

[0030] It should be noted that the pet health management method of this application can be applied to one or more (and one or more) pets, and is not limited here.

[0031] The target pet can refer to one or more of all pets in the current application scenario; there is no limitation here. For ease of understanding, the examples below will primarily use a single pet as the target pet.

[0032] Current health data refers to data that characterizes whether the target pet is currently healthy. It can be obtained through analyzing the target pet's daily behavior across one or more dimensions. For example, current health data may include, but is not limited to, data on food intake, excretion, and exercise.

[0033] Optionally, the current health data can also be set by the user (pet owner), which will not be elaborated here.

[0034] Current health data can be acquired through components including, but not limited to, visual sensors, acoustic sensors, mechanical sensors, and environmental sensors, without further limitation. These sensors can be installed on the pet feeder to acquire relevant data, or they can be collected through other smart pet care devices (which have a direct or indirect communication connection with the pet feeder) and sent to the pet feeder, which will not be elaborated on here.

[0035] In this case, other smart pet care devices may process and analyze the relevant data before sending it to the pet feeder, or other smart pet care devices may directly send the unprocessed and unanalyzed relevant data to the pet feeder for processing and analysis; there is no limitation here.

[0036] Therefore, the data modalities of these daily behavioral data can include video, images, audio, text, sensor signals, etc., without limitation here.

[0037] For example, one or more daily behavioral data can be analyzed using techniques such as preset judgment rules, calculation functions, machine learning and / or neural networks to obtain the corresponding current health data, which will not be elaborated here.

[0038] Step S120: Perform health analysis based on current health data and a pre-built health baseline model to obtain health risk information for the target pet.

[0039] Among them, the health baseline model refers to the benchmark used to make judgments during the health analysis process.

[0040] It should be noted that before conducting formal health analysis, a corresponding health baseline model needs to be built based on the pet's historical health data. Historical health data and current health data are essentially the same, but their acquisition timelines differ; historical health data was acquired earlier than current health data.

[0041] A health baseline model can be constructed based on one or more historical health data (such as feeding data, excretion data, exercise data, etc.), and there is no limitation here. Specific methods for constructing a health baseline model can refer to the relevant methods for determining baseline data (or benchmark data) in this technical field. This mainly involves analyzing historical health data to determine baseline data that can characterize the target pet's health status, thereby obtaining the corresponding health baseline model, which will not be elaborated upon here.

[0042] For example, the construction process may include using unsupervised learning algorithms to learn the joint probability distribution of multimodal behavioral features of the target pet in a normal (healthy) state.

[0043] Different pets can correspond to different health baseline models. Here, we will use the health baseline model corresponding to the target pet as an example for explanation.

[0044] For example, by performing health analysis based on the target pet's current health data and its corresponding health baseline model, the degree of deviation (or indicator deviation) between the target pet's current health data and the health baseline model can be determined. Therefore, the target pet's health risk information can be determined based on the indicator deviation.

[0045] Health risk information may include, but is not limited to, one or more of the following: a conclusion on whether the target pet has health risks, the type of health risks the target pet has, and data indicators indicating the existence of health risks.

[0046] It should also be noted that the methods for health analysis and processing may include, but are not limited to: comparing and analyzing the current health data with the health baseline data in the health baseline model, or inputting the current health data and the health baseline model into a pre-trained neural network for analysis; no limitation is made here.

[0047] Step S130: In response to the health risk information indicating that the target pet has a health risk, a target nutrition plan is determined from the preset nutrition management plan based on the health risk information.

[0048] Health risk information can characterize one or more aspects, such as whether the target pet has health risks, the types of health risks the target pet has, and data indicators indicating the presence of health risks.

[0049] A nutrition management plan refers to a program designed to manage the nutritional intake of a pet with specific health risks. Different health risk profiles may require different nutrition management plans. Different pets may also require different nutrition management plans; this is not a limitation here.

[0050] It should be noted that there can be a corresponding nutritional management plan regardless of whether the target pet has any health risks.

[0051] The nutrition management plan can be set by the user or automatically generated by the pet feeder (e.g., generated by built-in nutrition management logic and / or generated by collecting nutrition management knowledge from the Internet), and there are no restrictions here.

[0052] For example, when a target pet has a health risk, a corresponding nutritional management plan can be selected from a pre-defined nutritional management plan library based on the target pet's health risk information. The nutritional management plan library may include at least one nutritional management plan, which includes nutritional supplements (or medications) that the target pet should ingest when it has a health risk.

[0053] Another example is that different breeds of pets, pets of the same breed but different ages, or pets of the same breed and age but different weights may require different nutritional supplements even when facing the same health risks.

[0054] Therefore, when determining a nutrition management plan, in addition to selecting based on the target pet's health risk information, you can also consider the target pet's pet attributes (such as pet breed, pet age, pet weight, etc.).

[0055] For example, based on the target pet's health risk information and pet attribute information, a corresponding nutrition management plan can be selected from a pre-set nutrition management plan library. Alternatively, based on the target pet's health risk information, a corresponding nutrition management plan can be selected from a pre-set nutrition management plan library, and then the nutrition management plan can be adjusted according to the target pet's pet attribute information (such as adjusting the type and dosage of nutritional supplements). There are no limitations on this.

[0056] Step S140: Administer the nutritional supplements corresponding to the target nutrition plan.

[0057] In summary, if a target pet has health risks, and a corresponding nutritional management plan is determined based on the target pet's health risk information, then the nutritional supplements in the nutritional management plan can be administered.

[0058] For example, the nutritional supplement can be administered separately or mixed with the target pet's pet food (including staple food and / or treats) during the administration process; this is not limited to any particular method. Furthermore, the timing of the nutritional supplement administration can be when the target pet's feeding needs are detected, or when the target pet's feeding schedule (the pet food administration schedule) is triggered. Alternatively, the supplement administration schedule can be set up similarly to the methods used in setting up feeding schedules in this technical field; this is not limited to any particular method.

[0059] The mixed feeding process can be either mixing the nutritional supplement with the pet food during the feeding stage before dispensing it into the food bowl (mixing inside the pet feeder), or dispensing the nutritional supplement separately followed by the pet food (mixing in the pet feeder's food bowl), with no specific limitation here.

[0060] Specifically, it could be that the user sets the nutritional supplement according to the recommended way of taking it, or the pet feeder sets it automatically according to the type of nutritional supplement and the acquired nutritional management knowledge, or the pet feeder sets it automatically according to the form of the nutritional supplement (solid, liquid; block, powder, etc.).

[0061] Alternatively, several trial administrations (including individual administrations and / or mixed administrations) can be conducted beforehand. After the trial administrations, the target pet's response to the nutritional supplement is assessed. Response results may include at least one of the following: 1. administration in individual administration, 2. no administration in individual administration, 3. administration in mixed administration, and 4. no administration in mixed administration.

[0062] By analyzing the results of several supplement administration methods in the above examples, we can select the administration method that results in the highest frequency of supplement intake for the target pet. For example, if out of 10 attempts, the pet ate the supplement when administered alone 3 times, did not eat it when administered alone 2 times, and ate it when administered in combination with other methods 5 times, then the combination administration method can be determined as the current administration method for nutritional supplements.

[0063] Alternatively, the weight of each feeding method can be calculated by statistically analyzing the supplement feeding results using the examples above. For instance, the frequency of occurrence of various supplement feeding results can be quantified to obtain the weight of the corresponding feeding method, which will not be elaborated here. Then, the feeding method is selected based on the weight of each feeding method (e.g., selecting the feeding method with the highest weight).

[0064] Furthermore, in addition to determining the initial weight of each delivery method through the trial delivery process described in the aforementioned example, and then selecting the appropriate delivery method based on the initial weight, the pet's supplement consumption results can be recorded during one or more nutritional supplement delivery processes after the trial delivery process. Similarly, the weight calculation method described in the aforementioned example or other weight calculation methods in this technical field can be used to update the weight of each delivery method, thereby adapting to the dynamic changes in the pet's eating habits.

[0065] For example, each time a single-feeding treatment is detected, the weight of that single-feeding treatment is increased; each time a single-feeding treatment is detected as not eating, the weight of that single-feeding treatment is decreased; each time a mixed-feeding treatment is detected as eating, the weight of the single-feeding treatment is increased; and each time a mixed-feeding treatment is detected as not eating, the weight of the single-feeding treatment is decreased. Alternatively, after a period of nutritional supplement administration, the weight of each feeding method is recalculated based on the recorded supplement consumption results of the target pets over that period; this will not be elaborated upon here.

[0066] As can be seen, this application obtains the current health data of the target pet and performs health analysis processing on it against a pre-constructed health baseline model of the target pet. This allows for the determination of the deviation between the current health data and the health baseline model, thereby identifying the target pet's health risk information. Based on this health risk information, it can be determined whether the target pet has a health risk. When a health risk exists, a target nutrition plan can be determined from a pre-set nutrition management program based on the health risk information. Furthermore, it allows for the selection of mixing the corresponding nutritional supplements with the target pet's food. This enables long-term, continuous acquisition and analysis of the target pet's health data, timely assessment of potential health risks, generation of health warnings, and the addition of appropriate nutritional supplements during feeding according to a suitable nutrition management program, achieving rational and effective management of pet health.

[0067] For ease of understanding and explanation, this embodiment provides an example of a pet feeder to which the pet health management method of this application can be applied.

[0068] An example pet feeder may include at least: The multimodal perception module may include a camera, microphone array, high-precision weight sensor, environmental temperature and humidity sensor, and identification unit, and is mainly used to collect relevant data. For example, it can continuously collect multimodal time-series data of the target pet during each feeding process, as well as its associated environmental and pet identification information, through one or more of the visual sensor, acoustic sensor, weight sensor, environmental sensor, and identification unit.

[0069] The data processing and modeling module, which may include a processor and memory, is primarily used to run relevant algorithms. For example, it can process collected multimodal time-series data, build and continuously update a corresponding health baseline model for each target pet, compare current health data with the health baseline model to assess health risks, and determine appropriate nutritional management plans. It can also iteratively optimize the health baseline model.

[0070] Nutrition management module: This includes multi-compartment nutrient storage units, precision metering and dispensing mechanisms, and mixing devices, primarily used for dispensing nutritional supplements. For example, it controls the execution mechanism to accurately dispense the corresponding nutritional supplements according to the nutrition management plan.

[0071] Its multi-compartment nutrient solution compartment can contain several independent compartments (such as at least 4 independent compartments), which can store probiotics, urinary health supplements, joint nutrients, vitamins and other nutrient supplements respectively.

[0072] Its precision metering and dispensing mechanism can use components such as stepper motors to drive micro screws to achieve precise quantitative dispensing at the milligram level.

[0073] Its mixing device can use a micro-stirring paddle or air jet in the food bowl to ensure that the nutritional supplements are evenly mixed with the pet food.

[0074] System control and communication module: mainly used to coordinate the operation of various modules in the pet feeder and to communicate data with the user terminal and / or other smart pet devices.

[0075] Based on the above embodiments, this embodiment illustrates the method for constructing a healthy baseline model before step S110. Specifically, the method of this embodiment may include at least the following steps S001 to S003: Step S001 involves registering pet profiles for each pet in the current application scenario. Pet profiles may include one or more of the following: pet identity information, pet attribute information, such as pet ID, breed, age, weight, and health history; these are not limited here.

[0076] There are various specific methods for registration, including but not limited to the identification and detection function based on the pet feeder, the pet feeder obtaining information from other smart pet-raising devices with communication connections, and / or user-defined settings, etc., which are not limited here.

[0077] Step S002: During the baseline learning period of a preset time period (e.g., 10-14 days), relevant data is collected from the target pet for establishing the pet profile to learn and construct the health baseline model corresponding to the target pet.

[0078] Alternatively, during the baseline learning period, the system may simply collect data without implementing nutritional management.

[0079] In the process of building a health baseline model, relevant data collected (such as the multimodal historical health data mentioned in the previous embodiments) can be used to build a dynamic health baseline model for the target pet.

[0080] The baseline health model can be calculated based on the average value, or it can be represented by a "normal behavior feature space (the behavior feature space of the target pet in a healthy state)" learned through techniques such as Gaussian Mixture Model (GMM) or One-Class Support Vector Machine (SVM).

[0081] For example, the health baseline model of pet A can be represented as a joint probability distribution consisting of multiple data indicators such as eating rate, chewing frequency, and estimated water consumption. For GMM (Global Model), the degree of anomaly in data (such as current health data) can be assessed by calculating its log-likelihood under this distribution; a likelihood below a corresponding threshold is considered anomaly. For One-Class SVM (One-Class Streaming VM), the degree of anomaly is measured by calculating the distance from the data to the decision boundary. For details, please refer to the relevant explanations of the above-mentioned example methods in this technical field; they will not be elaborated upon here.

[0082] It should be noted that, in order to improve the reliability of the baseline health model and the accuracy of health analysis and processing, a health screening period can be selectively set before step S002 (baseline learning period) in this embodiment.

[0083] For example, before starting the formal baseline learning period, the system can first perform a short-term (e.g., 2-3 days) health screening period. During the health screening period, the system collects the target pet's initial health data as usual, and then compares the initial health data with the pre-stored general health threshold range to determine whether the target pet may have health risks from the beginning, so as to avoid learning abnormal data during the construction of the health baseline model.

[0084] The general health threshold range can be set by the user and / or obtained from other smart pet-raising devices and / or from the internet; no specific limitations are imposed here. The general health threshold range is typically derived from data collection and analysis of a large number of healthy pets; further details will not be elaborated upon here.

[0085] Simultaneously, the system can prompt the user to confirm via the app on the user's terminal that the target pet currently has no obvious health risks (such as vomiting, diarrhea, lethargy, or other disease symptoms). The system will only normally enter the baseline learning period and start the subsequent baseline learning process when the initial health data is within the general health threshold range, or when the initial health data is within the general health threshold range and the user confirms that the target pet is in a healthy state.

[0086] Therefore, the health screening mechanism effectively prevents the nutrition management system from mislearning the pet's initial abnormal health status as a "normal baseline," thus ensuring the accuracy and reliability of the health baseline model from the source.

[0087] Step S003: After constructing the health baseline model, the health baseline model can be dynamically updated using a preset online learning algorithm.

[0088] It should be noted that the health data collected in this application is essentially a high-dimensional time-series data stream. The core of dynamically updating the health baseline model lies in the fact that it learns not only the static distribution of health data, but also the evolutionary patterns and co-change laws of health data over time.

[0089] For example, the system can employ time-decay-based online learning algorithms (such as online Gaussian mixture models or incremental support vector machines) to continuously incorporate data verified as "healthy" into the parameters of the health baseline model with small weights.

[0090] This approach allows the health baseline model to adaptively and slowly track normal data drift caused by physiological changes in target pets due to age, season, etc. This avoids false alarms while remaining highly sensitive to pathological changes that deviate from long-term health trends.

[0091] For example, as a puppy grows older, its food intake gradually increases, causing the "food intake" data in its corresponding health baseline model to slowly rise. However, if its food intake suddenly drops sharply in a short period of time, the system can immediately identify this as an anomaly (a sign of health risk).

[0092] Based on the above embodiments, this embodiment illustrates the method for obtaining the current health data of the target pet in step S110.

[0093] Among them, obtaining the target pet's current health data may include at least obtaining the target pet's multimodal eating data when eating, and merging the eating data over a period of time as current health data to reflect whether the target pet exhibits health risks in its eating behavior.

[0094] Specifically, the method for obtaining the current health data of the target pet in step S110 may include at least the following steps S111 to S112: Step S111: In response to detecting the feeding behavior of the target pet, collect multimodal data of the target pet during feeding.

[0095] The method for detecting the feeding behavior of the target pet can refer to one or more feeding detection methods supported in this technical field, which will not be elaborated here.

[0096] For example, the method for detecting eating behavior in this embodiment may include, but is not limited to: 1. Use visual detection methods to determine whether the target pet is eating (e.g., use visual sensors to capture images including the food bowl to determine whether the target pet is eating: such as using images to determine whether the target pet is eating, or using images to determine whether the pet food in the food bowl has decreased, etc.).

[0097] 2. Determine whether the target pet is eating by using a weight detection method (e.g., determine whether the target pet is eating by using the weight change of the food bowl collected by a weight sensor: a decrease in the weight in the food bowl usually indicates that eating is occurring).

[0098] 3. Determine whether the target pet is eating by using acoustic detection methods (e.g., determine whether the target pet is eating by using ambient sounds collected by acoustic sensors: the presence of chewing, licking and other eating sounds usually indicates the presence of eating behavior).

[0099] 4. Combine the various methods in the above examples to jointly determine whether the target pet is eating (e.g., satisfying one or more of the following conditions: visual detection of eating behavior, decrease in weight in the food bowl, or collection of eating sounds, to indicate the presence of eating behavior).

[0100] Furthermore, after confirming that the target pet is eating, multimodal data of the target pet during eating can be collected (multimodal eating data). Multimodal data may include, but is not limited to: eating posture (including body posture and / or chewing posture, etc.), eating rate, chewing force, total amount of food consumed, etc.

[0101] Among them, the eating posture can be data collected by a visual sensor, the eating rate and total amount of food can be data collected by a weight sensor, and the chewing force can be data collected by an acoustic sensor (quantifying the collected chewing sounds). The specific data collection method can also be adjusted as needed according to the actual application scenario, and is not limited here.

[0102] Step S112: Perform time-series fusion processing on the modal data in the multimodal data to obtain the current health data of the target pet.

[0103] Based on the steps outlined above, after obtaining the multimodal data of the target pet during feeding, the various modal data can be time-series fused. Specific methods for time-series fusion processing can be found in relevant multimodal data fusion techniques within this technical field, and will not be elaborated upon here.

[0104] For example, data features from each modality can be extracted and transformed into the same feature space. The data features of each modality are then time-aligned according to their acquisition time sequence. Finally, the time-aligned data features are subjected to feature fusion processing to obtain the current health data of the target pet. The feature fusion processing can be feature splicing and / or feature merging, which is not limited here.

[0105] Based on the above embodiments, this embodiment illustrates a specific implementation scenario for collecting and analyzing various modal data as described above. Specifically, this embodiment may include at least the following methods: 1. Pet identification is achieved by acquiring videos and / or images of target pets through visual sensors. This allows for the assignment of a unique identifier to each pet in multi-pet households through visual analysis, ensuring the accuracy of subsequent data collection, modeling, health analysis, and nutrition management.

[0106] Furthermore, pet facial recognition algorithms can be integrated into visual inspection. Convolutional neural networks (CNNs) can be used to extract and compare the facial features of the target pet (such as facial shape, pattern and texture, nose print features, etc.) to achieve accurate identification of individual pets in multi-pet households. This will not be elaborated on here.

[0107] 2. Record video of the target pet eating using a visual sensor. For example, a camera with a resolution of 1080P or higher can be selected, recording video at a frame rate of no less than 15fps. The specific camera and recording specifications can be set as needed and are not limited here.

[0108] Subsequently, computer vision algorithms (such as animal pose estimation models based on OpenPose) can be used to extract the feeding posture of the target pet (such as head lifting height, facial muscle movement amplitude during chewing, and tongue licking frequency).

[0109] Optionally, during the target pet's non-feeding phase, the overall behavioral posture (such as whether it curls up, its activity range, movement frequency, sleep duration, etc.) of the target pet can be analyzed using captured video streams (which can be video streams captured by pet feeders and / or other smart feeding devices). A pre-trained deep learning model (such as ResNet or Vision Transformer) is then used to analyze the overall behavioral posture and calculate the target pet's lethargy index (or anxiety index), which reflects the target pet's mental and emotional state (a higher index indicates greater lethargy). The lethargy index can then be combined with current health data as a factor in assessing health risk information; details will not be elaborated here.

[0110] 3. Acquire the sounds of the target pet eating using acoustic sensors. For example, a MEMS microphone array with a signal-to-noise ratio ≥60dB can be selected to collect the audio of the target pet during the eating process.

[0111] For example, sound source localization technology and noise reduction algorithms can be used to separate eating sounds from ambient sounds in the acquired audio, and features such as the Mel-frequency cepstral coefficients (MFCC) of the eating sounds can be extracted. Therefore, data such as the chewing force, chewing rhythm, and chewing symmetry of the target pet can be quantified based on the characteristics of the eating sounds. For instance, if the intensity of chewing sounds on the left side of the target pet is significantly lower than that on the right side, it may indicate that the target pet has oral health problems.

[0112] 4. Obtain data such as the pet's eating rate through weight sensors. For example, a high-precision weighing sensor (with a weighing accuracy down to 0.1g) can be used to record changes in the weight of the food bowl at a frequency of 10Hz, and generate a data correlation curve of weight versus time. From this, the instantaneous eating rate (g / min), the frequency of feeding interruptions, etc., can be calculated.

[0113] 5. Acquire environmental data about the target pet's environment using environmental sensors. The environmental sensors may collect one or more types of data, and this is not limited here. For example, the environmental sensors may be temperature and humidity sensors and / or light intensity sensors.

[0114] For example, since different pets may have different eating habits under different environmental conditions, which may affect their normal eating behavior and thus interfere with the accuracy of health analysis, this embodiment can also perform contextual correction processing based on environmental data during eating behavior analysis.

[0115] For example, in hot and humid environments (or in dark environments), the target pet's appetite may decrease due to normal physiological processes, so such situations should not be misjudged as health abnormalities (or health risks).

[0116] 6. Obtain the target pet's excretion data through the excretion sensor in the smart toilet. Similarly, the excretion sensor can actually include one or more types; for ease of explanation, the various sensors installed in the smart toilet are collectively referred to as excretion sensors here. For specific details, please refer to the sensors commonly installed in smart toilets in this technical field and their supported functions, etc., which are not limited here.

[0117] Therefore, the excretion data of the target pet can be obtained through communication connection with the smart toilet (including but not limited to excretion type (urine and feces), excretion frequency, excretion volume, excrement form (which can be analyzed through visual images) and physicochemical indicators such as excrement pH value and crystallinity, which are not limited here).

[0118] Furthermore, these excretion data can be correlated with feeding data over a certain period of time, together forming key indicators for assessing the health of the target pet's digestive and urinary systems. For example, a normal food intake but a significant decrease in the frequency of bowel movements may indicate a risk of constipation. Abnormal changes in both water intake and urine output may indicate kidney or urinary tract risks.

[0119] Referring to the foregoing embodiments, the pet feeder of this application can optionally employ a system architecture that combines an embedded microprocessor (such as the ARM Cortex-A series) with a cloud platform. The complex model training process is chosen to be completed on the cloud platform, while the pre-trained lightweight model is deployed on the pet feeder terminal for real-time inference. The processor module can also include a built-in causal discovery submodule, employing techniques such as PC algorithms or NOTEARS models to analyze long-term collected multimodal data metrics to learn the underlying causal network structure between the metrics.

[0120] Based on the above embodiments, this embodiment illustrates the method in step S120 for performing health analysis processing based on current health data and a pre-built health baseline model to obtain health risk information of the target pet. Specifically, the method of this embodiment may include at least the following steps S121 to S122: Step S121: Compare and analyze the current health data with the health baseline data in the health baseline model to obtain the deviation of the indicators between the current health data and the health baseline data.

[0121] For example, the collected current health data can be compared and analyzed with the corresponding health benchmark data in the health baseline model to obtain the deviation of the indicators between the current health data and the health benchmark data.

[0122] It should be noted that since there can be one or more types of current health data and health benchmark data, there can be one or more corresponding indicator deviations. For ease of understanding, this embodiment will use one indicator deviation as an example.

[0123] The indicator deviation represents how much the current health data deviates from the health baseline data. When indicator deviation exists, the deviation typically includes the current health data being higher than the health baseline data, or the current health data being equal to the health baseline data. Each health baseline data can have a corresponding indicator deviation threshold, or a corresponding health threshold range.

[0124] Step S122: Determine health risk information based on the deviation of indicators within a preset time period.

[0125] In the process of determining health risk information based on the deviation of indicators within a preset time period, health risk information can be determined based on the deviation of indicators at a certain moment within the preset time period, or based on the change of indicator deviation over time within the preset time period (such as the deviation increasing or decreasing). No limitation is made here.

[0126] Health risk information may include whether there is a health risk or not.

[0127] For example, if the current health data deviates significantly from the health benchmark data based on an indicator that deviates from a threshold (or a health threshold range), then the current health data can be determined to be an abnormal indicator (i.e., the target pet has a health risk). Otherwise, the current health data can be determined to be a normal indicator (i.e., the target pet does not have a health risk).

[0128] Based on the above embodiments, this embodiment describes the method for determining health risk information according to the deviation of indicators within a preset time period in step S122. The existing health risks can be classified. For example, health risks can be classified by quantifying them using a risk index (e.g., a higher risk index indicates a higher and more severe health risk), or they can be classified by a stage-based system (e.g., classifying health risks from severe to mild as severe, moderate, and mild). This embodiment mainly uses the example of health risk information including mild and severe health risks for illustration.

[0129] It should be noted that different levels of health risk can correspond to different levels of attention, meaning different nutritional management plans can be set according to different levels, which will not be elaborated here. For example, mild health risk can correspond to an observation-level warning (low attention, generating a minor alert and not administering nutritional supplements); moderate health risk can correspond to a attention-level warning (medium attention, generating a significant alert and administering nutritional supplements); and high health risk can correspond to an action-level warning (high attention, generating a severe alert and administering nutritional supplements).

[0130] Specifically, the method for determining health risk information based on the deviation of indicators within a preset time period in this embodiment may include at least the following steps S1221 to S1222: Step S1221: In response to the presence of a single abnormal data indicator in the health risk level representation, it is determined that the target pet has a mild health risk.

[0131] Referring to the foregoing embodiments, if among the multiple data indicators of the current health data, if there is only a single abnormal data indicator in the health risk level indicator, it can be determined that the target pet has a mild health risk.

[0132] For example, if a single data indicator (such as the rate of eating) is lower than X% (such as 15%) of the corresponding baseline average for N consecutive days (such as 5 days), the rate of eating is considered an abnormal data indicator, and a mild health risk can be identified.

[0133] Optionally, if there is only a single abnormal data indicator, and the deviation between the abnormal data indicator and the corresponding health benchmark data is too large (greater than the corresponding preset maximum deviation limit, which is usually larger than the indicator deviation threshold, such as X% being 30% in the above example), then it is still possible to choose to trigger the corresponding operation for moderate or severe health risk, which is not limited here.

[0134] Step S1222: In response to the presence of multiple abnormal data indicators in the health risk level representation, it is determined that the target pet has a severe health risk.

[0135] If multiple abnormal data indicators are found in the health risk level indicators among the current health data, it can be determined that the target pet has a severe health risk.

[0136] For example, if multiple data indicators in the current health data (such as eating rate, chewing force, and total food intake) show a significant decrease at the same time (indicating an abnormality), it can be determined that the target pet has a serious health risk.

[0137] Optionally, to improve the sensitivity of health risk assessment, it is also possible to determine that the target pet has a severe health risk when multiple data indicators are simultaneously abnormal (simultaneously decreasing or simultaneously increasing), even if some or all of the multiple data indicators do not exceed the corresponding threshold.

[0138] Building upon this foundation, the causal discovery submodule is activated. This module can analyze historical health data to construct causal paths (e.g., "increased water intake and decreased food intake" corresponding to "kidney disease or metabolic disease"); or determine the type of health risk based on preset causal paths. If current health data matches this causal path, the system can associate the health risk type of severe health risk with "kidney / metabolic problems" instead of "loss of appetite," thereby greatly improving the accuracy of health warnings.

[0139] Specifically, the methods supported by the aforementioned causal discovery submodule may also include root cause analysis. For example, when the system detects and identifies multiple data indicators with synchronization anomalies through the coupling relationship of multiple data indicators, root cause analysis can be performed through the causal discovery submodule.

[0140] This submodule can construct a directed acyclic graph (DAG) between key health data indicators (such as food intake, water intake, exercise, and weight) based on historical health data over a certain period of time, using causal discovery models such as the PC algorithm and NOTEARS, in order to characterize the potential causal relationships between each health data indicator.

[0141] In real-time monitoring, if changes in multiple health data indicators are found to conform to an abnormal pattern along a causal path (for example, historical health data showing an abnormal increase in water intake is often the cause of an abnormal decrease in food intake), the system can associate the current health risk information with a specific organ system (such as the kidneys or endocrine system) with higher confidence. This elevates the system's decision-making logic from "predicting possible symptoms" to "the cause of current symptoms," providing a deeper, more rule-based basis for generating more targeted nutritional management plans.

[0142] Based on the above embodiments, this embodiment describes the method in step S130 of determining a target nutrition plan from a preset nutrition management plan in response to health risk information indicating that the target pet has a health risk.

[0143] The system in this application can pre-build a "mapping knowledge base between health risk information and nutrition management plans". This knowledge base defines the correspondence between abnormal feature vectors of different health risk information and specific nutritional supplement formulas and dosages.

[0144] When specific health risk information is identified (a health data indicator deviates significantly from the corresponding baseline data, or multiple health data points deviate abnormally), the system can query the mapping knowledge base and automatically generate and execute a personalized nutrition management plan based on the preset nutrition management plans in the base.

[0145] Specifically, when multiple abnormal data indicators are present in the changes of health risk information, the method for determining the target nutrition plan from the preset nutrition management plan based on the health risk information in this embodiment may include at least the following steps S1301 to S1302: Step S1301: In response to the presence of multiple abnormal data indicators in the health risk information representation, the target risk type of the health risk information is determined based on the multiple abnormal data indicators; the deviation of each abnormal data indicator is greater than the deviation threshold corresponding to each abnormal data indicator.

[0146] For example, the judgment logic for determining the target risk type based on multiple abnormal data indicators in this application may include, but is not limited to, setting the following conditions: 1. Abnormal data indicators (decreased eating rate + decreased chewing strength + decreased food intake + normal lethargy index) determine the target risk type (loss of appetite, indigestion).

[0147] 2. Abnormal data indicators (increased water intake + decreased food intake) determine the target risk type (early kidney disease, diabetes risk).

[0148] 3. Abnormal data indicators (abnormal eating posture (such as head tilting while eating) + decreased chewing symmetry) determine the target risk type (oral pain, periodontal disease).

[0149] 4. Abnormal data indicators (decreased physical activity + normal eating rate + normal food intake + elevated lethargy index) determine the target risk type (joint discomfort, decreased vitality).

[0150] Furthermore, relevant knowledge in this technical field can be incorporated to set or adjust the judgment logic for determining the target risk type as needed. Specifically, the corresponding judgment logic can also be generated and set through Retrieval-Augmented Generation (RAG) technology, which is not limited here.

[0151] Step S1302: Determine the target nutrition plan from each nutrition management plan based on the mapping weight between the target risk type and each nutrition management plan.

[0152] Based on the steps outlined above, each type of health risk has a corresponding mapping weight with a preset nutrition management plan (hereinafter referred to as a preset nutrition plan) in the knowledge base. Therefore, when selecting a corresponding nutrition management plan, the appropriate plan can be determined and implemented based on the mapping weight for the current health risk type.

[0153] For example, the preset nutrition plans corresponding to each target risk type in the knowledge base may include, but are not limited to, settings based on the following conditions: 1. The target risk type (loss of appetite, indigestion) can be matched with a corresponding nutritional management plan (administering probiotics (e.g., 100mg)).

[0154] 2. The target risk type (early kidney disease, diabetes risk) can correspond to the nutritional management plan (administering kidney health nutrients (e.g., 150mg)).

[0155] 3. Target risk types (oral pain, periodontal disease) can be matched with corresponding nutritional management plans (administering oral health soothing agents (e.g., 80mg)).

[0156] 4. The target risk type (joint discomfort, decreased vitality) can be matched with a nutritional management plan (injection of joint nutrients (e.g., 120mg) + B complex vitamins (e.g., 50mg)).

[0157] Based on the above embodiments, this embodiment describes the method after administering the nutritional supplements corresponding to the target nutrition plan in step S140.

[0158] It should be noted that this embodiment may also provide a method for nutrition management feedback and model iteration.

[0159] Specifically, after administering the nutritional supplements corresponding to the target nutrition plan, this embodiment may further include at least the following steps: Analyze the subsequent changes in the deviation of the indicators; based on the subsequent changes, adjust the mapping weights between the target risk type and the target nutrition plan.

[0160] For example, after executing a nutrition management plan, the system may initiate a feedback optimization cycle. This feedback optimization cycle may include at least: 1. Nutritional Management Effectiveness Tracking: The system continuously tracks the target pet's key health indicators (such as food intake, lethargy index, etc.) and / or previously abnormal data indicators for a certain period of time (such as 1-3 days) after the implementation of the nutritional management plan, and obtains the subsequent changes in the deviation of these data indicators.

[0161] Subsequent changes may include, but are not limited to, calculating the recovery rate of these data indicators relative to the indicators before the implementation of the nutrition management plan.

[0162] For example, whether a data indicator that significantly decreased before implementing a nutrition management plan increased after implementing the plan, and by how much (or the ratio between the increase and the deviation of the data indicator), will not be elaborated here. Conversely, the same logic can be applied to a data indicator that significantly increased before implementing a nutrition management plan, which will also not be elaborated here.

[0163] 2. Nutritional Management Effectiveness Evaluation: If the indicator recovery rate exceeds the indicator recovery threshold (e.g., 70%), the strategy of "nutritional management plan adopted for the target risk type" can be deemed effective. If the indicator recovery rate is lower than the indicator recovery threshold, or if these data indicators continue to deteriorate, the strategy of "nutritional management plan adopted for the target risk type" is deemed ineffective.

[0164] 3. Knowledge base optimization: Strategy reinforcement: For strategy pairs marked as "effective", the system increases the mapping weight or confidence of the mapping relationship in its "mapping knowledge base between health risk information and nutrition management plan", so that when similar health risk information is detected in the future, the system will give higher priority to making decisions from the preset nutrition plan.

[0165] Strategy Exploration: For strategy pairs marked as "invalid", the system can trigger preset backup management plans (such as reducing the types and / or amounts of nutritional supplements; increasing pet energy intake by providing pet treats, etc.), and / or send a serious prompt to the user terminal via the APP: "Current nutritional management is not effective, it is recommended to seek medical attention".

[0166] 4. Model Iteration: All or part of the current health data that has been verified as "healthy" can be used to incrementally update the target pet's health baseline model.

[0167] Thus, through the above closed-loop solution, the system has achieved intelligent evolution from passively triggering the execution of nutrition management plans based on health risk information to continuously learning from interactive experiences and constantly optimizing future decisions, significantly improving the accuracy and adaptability of long-term health management for target pets.

[0168] Based on the above embodiments, this embodiment should be noted that in actual application scenarios, if the target pet is diagnosed with a health problem, the initially determined health baseline model may be difficult to apply to the health management of target pets in sick, recovering, or in a period of time.

[0169] Therefore, this application may also require dynamic updates to the health baseline model to make it more closely reflect the actual changes in the health status of the target pet.

[0170] Specifically, after administering the nutritional supplements corresponding to the target nutrition plan, this embodiment may further include at least the following steps: Acquire subsequent health data of the target pet; in response to the confirmation of the target risk type, incrementally update the health baseline model based on the subsequent health data and the preset baseline update weights.

[0171] After administering nutritional supplements, the method for obtaining current health data in the aforementioned embodiments can be used to continuously obtain subsequent health data for the target pet.

[0172] You can choose to use a preset baseline update weight (a smaller weight; the specific weight value can be set as needed and is not limited here) to update the health baseline model with subsequent health data increments (updating the corresponding health benchmark data), so that the health baseline model is applicable to the target pet's sick state, sick state, and recovery state, providing a more accurate benchmark reference.

[0173] Alternatively, the health baseline model can be incrementally updated based on subsequent health data after the user confirms the target risk type (the user can issue a confirmation command). This allows for continuous reminders to the user about the health risks associated with that pet, even before the user confirms the target risk type, ensuring timely reminders are not lost after the health baseline model is updated.

[0174] Alternatively, you can proactively screen the health data collected after administering nutritional supplements. All or part of the health data verified as "healthy" (normal) can then be used to incrementally update the target pet's health baseline model.

[0175] Based on the above embodiments, this embodiment describes the method in step S120 for performing health analysis processing based on current health data and a pre-built health baseline model to obtain health risk information of the target pet.

[0176] Specifically, the method in this embodiment for obtaining health risk information of a target pet by performing health analysis based on current health data and a pre-built health baseline model may include at least the following steps: Health analysis is performed based on current health data and a health baseline model to obtain initial health information; current environmental data corresponding to the time series of current health data is obtained; and contextual correction is performed on the initial health information based on the current environmental data to obtain health risk information.

[0177] In conjunction with the foregoing embodiments, during the health analysis process based on current health data and a health baseline model, the current health data of the target pet may be affected by the current environmental data, resulting in certain errors in the analyzed health information.

[0178] Therefore, in this embodiment, after performing health analysis based on the current health data and health baseline model to obtain initial health information, the initial health information can be context-corrected using the current environmental data to obtain more accurate health risk information.

[0179] For specific methods, please refer to the relevant descriptions of the foregoing embodiments, which will not be repeated here.

[0180] Based on the above embodiments, this embodiment describes the method prior to step S110.

[0181] Specifically, the method in this embodiment before performing health analysis processing based on current health data and a pre-built health baseline model in step S110 to obtain the health risk information of the target pet may further include at least the following steps: Obtain the initial health data of the target pet; perform health screening on the target pet based on the initial health data and preset health thresholds to obtain the initial screening results of the target pet; in response to the initial screening results characterizing the health of the target pet, construct a health baseline model based on the initial health data.

[0182] In conjunction with the foregoing embodiments, the method of this embodiment can pre-determine whether the target pet is healthy, thereby using the initial health data of "health" to construct a health baseline model of the target pet, making the subsequent health analysis processing using the health baseline model more accurate.

[0183] For specific methods, please refer to the relevant descriptions of the foregoing embodiments, which will not be repeated here.

[0184] Based on the above embodiments, to facilitate a detailed understanding of the application scenarios of the method of this application, this embodiment provides examples of several specific application scenarios in which this application can be implemented.

[0185] Application Scenario 1: Early detection, accurate diagnosis, and closed-loop nutritional management of chronic kidney disease in an elderly cat named "Pet A".

[0186] Background information: Pet A, a 12-year-old Persian cat, has a baseline learning period of 14 days. A health baseline model (including eating rate, water intake, chewing force, etc.) is established for it in a healthy state.

[0187] Health analysis process: 1) Day 22: The system detected that pet A's average daily food intake decreased from the health baseline of 48g to 42g (-12.5%), triggering the corresponding action for mild health risk (e.g., recording relevant information but not giving a prompt).

[0188] 2) Day 25: The health risk trend continues, with food intake decreasing to 39g (-18.8%) and the average feeding rate decreasing from 4.2g / min to 3.5g / min (-16.7%). The abnormal trend detection conditions for a single data indicator are met, triggering the corresponding actions for moderate health risk (such as notifying the user via the APP: "Pet A's appetite continues to decline, please pay attention").

[0189] 3) Day 28: An abnormal coupling relationship was detected among multiple data indicators, with a synchronous upward trend in "water consumption" (estimated by observing the frequency of water bowl usage via camera or by estimating water bowl weight). This indicates that "decreased appetite + increased water consumption" synergistically constitute a severe health risk. Simultaneously, after analyzing historical health data, the causal discovery submodule determines that the probability (confidence level) of "increased water consumption" as the cause and "decreased food intake" as the effect is as high as 85%. Therefore, the system can immediately determine that pet A has a severe health risk.

[0190] System decision-making and execution: 1) Diagnosis: The algorithm maps and associates the health risk type with "early stage of chronic kidney disease" based on a knowledge base and causal analysis.

[0191] 2) Nutritional management: Automatically and precisely add 150mg of kidney-prescription nutritional powder to pet A's food.

[0192] 3) Warning: The APP sends a red warning to the user: "Analysis shows that pet A has a high risk of early kidney dysfunction! Kidney nutritional support has been initiated. Please arrange medical examination as soon as possible."

[0193] Results and Validation (Closed Loop): 1) If a user takes their pet A to the vet and the vet confirms through blood tests that the creatinine level is slightly elevated, early-stage chronic kidney disease can be diagnosed.

[0194] 2) The system continuously monitored and confirmed that after nutritional management, Pet A's food intake gradually increased to 43g within a week, while its water intake also decreased slightly. The effect evaluation module marked this "kidney nutrient" intervention as "effective" and recorded this effective case.

[0195] 3) Model Iteration: The system updates the dynamic health baseline model of pet A with a small weight increment after the pet is diagnosed and tends to be healthy and stable under nutritional management. This makes the health baseline model adapt to the pet's current "stable with disease" state and provides more accurate benchmark data for future testing.

[0196] Application Scenario 2: Social appetite reduction identification of "Pet B" in multi-pet households.

[0197] Background information: The current household has two cats, "Pet B" (timid and compliant) and "Pet C" (strong and dominant). The system has established corresponding health baseline models for both cats and accurately recorded their respective health data through an identification unit.

[0198] Health analysis process: 1) The system detected that "Pet B's" eating rate and food intake decreased significantly within a week.

[0199] 2) Simultaneously, data analysis showed that pet B's abnormal eating behavior was related to pet C's simultaneous approach to the food bowl. Pet B's "lethargy index" was normal during daily playtime but increased during feeding time. Multi-pet behavior analysis indicated that this was a typical behavioral abnormality caused by "pet competition stress."

[0200] System decision-making and execution: The app can be used to send suggestions to users: "It has been detected that pet B is eating less due to competition from other pets. It is recommended to feed the pet in different areas or at different times to ensure that it eats enough."

[0201] Meanwhile, the pet feeder can generate two feeding plans for pet B and pet C that are far apart, and can also choose to add appetite-enhancing nutritional supplements when feeding pet B to improve pet B's appetite.

[0202] Result: After the user adopted the suggestion, pet B's eating data returned to normal. This example demonstrates the system's ability to distinguish between health risks caused by health problems and behavioral problems in complex scenarios, avoiding false alarms about health issues.

[0203] It should be further noted that the entity executing the pet health management method can be a pet health management device. For example, the pet health management method can be executed by a terminal device, server, or other processing device. The terminal device can be a user equipment (UE), computer, mobile device, user terminal, terminal, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, in-vehicle device, wearable device, etc. In some possible implementations, the pet health management method can be implemented by a processor calling computer-readable instructions stored in memory.

[0204] Figure 2 This is a block diagram illustrating a pet health management device as shown in an exemplary embodiment of this application. Figure 2 As shown, this exemplary pet health management device 200 can be applied to a pet feeder and includes: an acquisition module 210, a health analysis module 220, a nutrition management module 230, and a supplement dispensing module 240. Specifically: The acquisition module 210 is used to acquire the current health data of the target pet.

[0205] The health analysis module 220 is used to perform health analysis based on current health data and a pre-built health baseline model to obtain health risk information for the target pet.

[0206] The nutrition management module 230 is used to respond to health risk information indicating that the target pet has health risks, and to determine the target nutrition plan from the preset nutrition management plan based on the health risk information.

[0207] The supplement delivery module 240 is used to deliver the nutritional supplements corresponding to the target nutrition plan.

[0208] In this exemplary pet health management device, by acquiring the current health data of the target pet and performing health analysis processing against a pre-built health baseline model of the target pet, the deviation between the current health data and the health baseline model can be obtained, thereby determining the target pet's health risk information. Based on the health risk information, it can be determined whether the target pet has a health risk. When the target pet has a health risk, a target nutrition plan can be determined from the preset nutrition management plan based on the health risk information, and the corresponding nutritional supplements can be mixed with the target pet's food for administration. This allows for long-term, continuous acquisition and analysis of the target pet's health data, timely determination of whether the target pet has a health risk and generation of health warnings, and the addition of corresponding nutritional supplements when feeding the target pet according to an appropriate nutrition management plan, thus achieving reasonable and effective management of pet health.

[0209] It should be noted that the apparatus and method provided in the above embodiments belong to the same concept, and the specific ways in which each module and unit performs operations have been described in detail in the method embodiments, and will not be repeated here. In practical applications, the apparatus provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the apparatus can be divided into different functional modules to complete all or part of the functions described above, and this is not a limitation.

[0210] The functions of each module can be found in the pet health management method implementation example, and will not be repeated here.

[0211] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application. The electronic device 100 includes a memory 101 and a processor 102. The processor 102 is used to execute program instructions stored in the memory 101 to implement the steps in any of the above-described embodiments of the pet health management method. In a specific implementation scenario, the electronic device 100 may include, but is not limited to, a microcomputer or a server. In addition, the electronic device 100 may also include mobile devices such as laptops and tablets, which are not limited here.

[0212] Specifically, processor 102 controls itself and memory 101 to implement the steps in any of the above-described pet health management method embodiments. Processor 102 may also be referred to as a CPU (Central Processing Unit). Processor 102 may be an integrated circuit chip with signal processing capabilities. Processor 102 may also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor. Furthermore, processor 102 may be implemented using integrated circuit chips.

[0213] In this exemplary electronic device, by acquiring the current health data of the target pet and performing health analysis processing against a pre-built health baseline model of the target pet, the deviation of the current health data from the health baseline model can be obtained, thereby determining the target pet's health risk information. Based on the health risk information, it can be determined whether the target pet has a health risk. When the target pet has a health risk, a target nutrition plan can be determined from a preset nutrition management plan based on the health risk information, and the corresponding nutritional supplements can be mixed with the target pet's food for administration. This allows for long-term, continuous acquisition and analysis of the target pet's health data, timely determination of whether the target pet has a health risk and generation of health warnings, and the addition of corresponding nutritional supplements when feeding the target pet according to an appropriate nutrition management plan, thus achieving reasonable and effective management of pet health.

[0214] Please see Figure 4 , Figure 4 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present application. The computer-readable storage medium 110 stores program instructions 111 that can be executed by a processor. The program instructions 111 are used to implement the steps in any of the above-described embodiments of the pet health management method.

[0215] In this exemplary storage medium, by running the program instructions stored in the medium, the current health data of the target pet is acquired, and this data is compared with a pre-built health baseline model of the target pet for health analysis. The deviation between the current health data and the health baseline model is obtained, thereby determining the target pet's health risk information. Based on the health risk information, it can be determined whether the target pet has a health risk. If the target pet has a health risk, a target nutrition plan can be determined from a preset nutrition management plan based on the health risk information. It is also possible to choose to mix the corresponding nutritional supplements with the target pet's food. This allows for long-term, continuous acquisition and analysis of the target pet's health data, timely determination of whether the target pet has a health risk and generation of health warnings. By adding appropriate nutritional supplements when feeding the target pet according to a suitable nutrition management plan, the pet's health can be managed rationally and effectively.

[0216] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0217] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0218] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0219] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for pet health management, characterized in that, The method is applied to a pet feeder, and the method includes: Obtain the target pet's current health data; Based on the current health data and the pre-built health baseline model, health analysis and processing are performed to obtain the health risk information of the target pet. In response to the health risk information indicating that the target pet has a health risk, a target nutrition plan is determined from the preset nutrition management plan based on the health risk information; The nutritional supplements corresponding to the target nutrition plan are administered.

2. The method according to claim 1, characterized in that, The acquisition of the target pet's current health data includes: In response to detecting the feeding behavior of the target pet, multimodal data of the target pet during feeding is collected; The current health data of the target pet is obtained by performing time-series fusion processing on the modal data in the multimodal data.

3. The method according to claim 1, characterized in that, Based on the current health data and the pre-built health baseline model, health analysis is performed to obtain the health risk information of the target pet, including: The current health data is compared and analyzed with the health baseline data in the health baseline model to obtain the index deviation between the current health data and the health baseline data; The health risk information is determined based on the deviation of the indicator within a preset time period.

4. The method according to claim 3, characterized in that, The step of responding to the health risk information indicating that the target pet has a health risk, and determining a target nutrition plan from a preset nutrition management plan based on the health risk information, includes: In response to the presence of multiple abnormal data indicators in the health risk information representation, the target risk type of the health risk information is determined based on the multiple abnormal data indicators; the deviation of each abnormal data indicator is greater than the corresponding deviation threshold. Based on the mapping weights between the target risk type and each nutrition management program, the target nutrition program is determined from each nutrition management program.

5. The method according to claim 4, characterized in that, After administering the nutritional supplements corresponding to the target nutrition plan, the method further includes: Analyze the subsequent changes in the deviation of the aforementioned indicator; Based on the subsequent changes, adjust the mapping weights between the target risk type and the target nutrition plan.

6. The method according to claim 1, characterized in that, After administering the nutritional supplements corresponding to the target nutrition plan, the method further includes: Obtain subsequent health data of the target pet; In response to the confirmation of the target risk type, the health baseline model is incrementally updated based on the subsequent health data and the preset baseline update weights.

7. The method according to claim 1, characterized in that, The step of performing health analysis based on the current health data and a pre-built health baseline model to obtain the health risk information of the target pet includes: Based on the current health data and the health baseline model, health analysis and processing are performed to obtain initial health information; Obtain the current environmental data corresponding to the current health data time series; The initial health information is subjected to context correction processing based on the current environmental data to obtain the health risk information.

8. The method according to claim 1, characterized in that, Before performing health analysis based on the current health data and a pre-built health baseline model to obtain the health risk information of the target pet, the method further includes: Obtain the initial health data of the target pet; Based on the initial health data and preset health thresholds, a health screening process is performed on the target pet to obtain the initial screening results of the target pet; In response to the initial screening results characterizing the health of the target pet, a health baseline model is constructed based on the initial health data.

9. An electronic device, characterized in that, The method includes a memory and a processor, the processor being configured to execute program instructions stored in the memory to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by the processor, they implement the method described in any one of claims 1 to 8.

Images