Search

CN-122020486-A - Pet health monitoring system based on microbiome gene detection

CN122020486ACN 122020486 ACN122020486 ACN 122020486ACN-122020486-A

Abstract

The invention relates to the technical field of pet health monitoring, and discloses a pet health monitoring system based on microbiome gene detection. The system comprises a data acquisition and fluidization module, a cross-layer association and primary screening module, a dynamic baseline modeling module, an anomaly comparison and background analysis module and a deep threat traceability module. The system acquires a pet gene sequence and metadata through formulating an acquisition scheme to form a real-time data stream, further performs cross-level correlation analysis, identifies an abnormal interaction mode and generates an abnormal marked data set with weight, builds and dynamically calibrates a personalized baseline model by utilizing the data set and the data stream, compares the real-time data with the baseline to find deviation, analyzes an abnormal background by combining the metadata, and finally analyzes the metadata based on an abnormal identification result to trace the deep health threat mode and the root cause. The invention realizes continuous dynamic monitoring and multi-level correlation analysis of the pet microbiome and improves the accuracy of early health risk early warning.

Inventors

  • JIA GUANGMIN
  • LI ZHENPING
  • YANG DAWEI

Assignees

  • 深圳罗米智能科技有限公司

Dates

Publication Date
20260512
Application Date
20260324

Claims (10)

  1. 1. A pet health monitoring system based on microbiome genetic testing, the system comprising: The data acquisition and fluidization module is used for determining a microbiome gene data acquisition scheme for pet health monitoring, acquiring gene sequence data and sample metadata from a pet biological sample according to the scheme, and concatenating the acquired data into a real-time gene data stream according to time sequence; The cross-level correlation and primary screening module is used for carrying out cross-level correlation analysis by taking the level logic of the real-time gene data flow as the context, combing different levels of correlation by combining the scene context generated by the data, finding out the correlation and abnormal interaction mode of the gene data among the levels, marking the found result as a preliminary abnormal signal, and collecting the preliminary abnormal signal into an abnormal marked data set with weight; The dynamic baseline modeling module is used for constructing a dynamic baseline model by depending on the real-time gene data flow and the abnormal marker data set, calibrating the baseline model by using the result of cross-level correlation analysis, and updating the dynamic parameter adjustment along with the data; the anomaly comparison and background analysis module is used for comparing the data acquired in real time with the dynamic baseline model, finding out deviated abnormal activities, marking the deviated abnormal behaviors, and transmitting the deviation recognition results to the sample metadata for analysis to obtain abnormal background information; and the deep threat tracing module is used for analyzing sample metadata based on an abnormal recognition result of the dynamic baseline model, finding out a potential deep health threat mode and an abnormal signal, tracing the threat mode and the abnormal signal, and determining a root cause and an abnormal path.
  2. 2. The pet health monitoring system based on microbiome gene detection of claim 1, wherein the determining the microbiome gene data collection scheme of the pet health monitoring comprises obtaining the gene sequence data and the sample metadata from the pet biological sample according to the scheme, and concatenating the obtained data into a real-time gene data stream according to the time sequence, and the method specifically comprises: Firstly, combining the age stage, daily diet type and recent health state of the pet, determining whether a biological sample to be acquired is a fecal sample, an oral sample or a skin sample, and determining that the gene data acquisition requirement corresponding to each type of sample is full-gene sequencing depth, functional gene targeting acquisition range and sample metadata must contain acquisition time, acquisition parts and recent diet records of the pet; performing gene data capture on biological samples collected in real time according to specific requirements, extracting intestinal microbial genes from fecal samples, extracting oral microbial genes from oral samples and extracting skin microbial genes from skin samples; Classifying the captured gene data according to sample sources, carding species classification information according to a microorganism species layer, carding gene function information according to a functional gene layer and carding sequence base arrangement information according to a gene sequence layer, and analyzing the gene sequence data and sample metadata comprising acquisition time, acquisition positions, recent diet records of pets and varieties of pets; And serially connecting the analyzed gene sequence data according to the acquisition time, and attaching sample metadata at corresponding time points to the gene sequence data serially connected at the same time to form a real-time gene data stream.
  3. 3. The pet health monitoring system based on microbiome gene detection of claim 2, wherein the hierarchical logic of the real-time gene data stream is taken as a context to carry out cross-hierarchical correlation analysis, different hierarchical associations are combed by combining scene contexts generated by data, the association and abnormal interaction mode of the gene data among the hierarchical associations are found, the found result is marked as a preliminary abnormal signal, and the preliminary abnormal signal is collected into an abnormal marked data set with weight, and the pet health monitoring system specifically comprises: Firstly dividing gene sequence data into continuous time slices according to acquisition time according to the generation logic of a real-time gene data stream, arranging data in each time slice according to a species layer, a functional gene layer and a gene sequence layer, and extracting scene context information such as the environmental state of an acquisition part, the types of components in pet diet and the types of recently contacted articles from each time slice data; Aiming at a species layer and a functional gene layer in each time slice, analyzing the association between the abundance of a certain type of microorganism species and the expression abundance of a corresponding functional gene, judging whether the association accords with the normal corresponding logic of the species and the functional gene under the component according to the component type in the pet diet of the time slice, if not, marking the species function association as abnormal, aiming at the functional gene layer and the gene sequence layer, analyzing the association between the sequence characteristic of a certain functional gene and the expression abundance of the gene, judging whether the association accords with the normal corresponding logic of the functional gene and the sequence characteristic under the environment according to the environmental state of the acquisition part of the time slice, and if not, marking the species function association as abnormal; Marking species function association abnormality and gene sequence association abnormality as preliminary abnormality signals, recording the abnormality type of each preliminary abnormality signal as species function association abnormality or gene sequence association abnormality, the frequency as the occurrence frequency of the abnormality in a time slice and the severity as the degree of association deviation from normal logic; And distributing weights according to logic deviation difficulty corresponding to the abnormal type of each preliminary abnormal signal, occurrence frequency corresponding to the frequency and deviation depth corresponding to the severity, and integrating the preliminary abnormal signals with the weights with scene context information of the corresponding time slices to form an abnormal mark data set with the weights.
  4. 4. The pet health monitoring system based on microbiome gene detection of claim 3, wherein the constructing a dynamic baseline model based on real-time gene data flow and anomaly marker data set, calibrating the baseline model using the results of cross-level correlation analysis, and updating dynamic tuning parameters with data, specifically comprises: Firstly extracting species layer microorganism species abundance distribution, functional gene layer functional gene expression abundance distribution and gene sequence layer gene sequence characteristic distribution of each time slice from a real-time gene data stream, and extracting preliminary abnormal signal weight and scene context information of each time slice from an abnormal mark data set with weight; The method comprises the steps of using the extracted abundance distribution of a species layer, the expression abundance distribution of a functional gene layer and the characteristic distribution of a gene sequence layer as basic characteristics, using the weight of an abnormal signal in an abnormal marking dataset with weight as a correction characteristic, constructing an initial dynamic baseline model, using a basic characteristic training model to learn the stable characteristics of the distribution of each layer in a normal state, and using the correction characteristic training model to identify the boundary between the normal distribution and the abnormal distribution; And (3) inputting the basic characteristics of the time slice into an initial dynamic baseline model to obtain basic predicted values of each layer distribution of the time slice, adjusting the basic predicted values by using the abnormal signal weights in the time slice with the weight abnormal marked data set, taking the adjusted result as parameters of the time slice baseline model, updating the integral parameters of the dynamic baseline model by using the parameters of the new time slice, and simultaneously identifying the threshold values of normal and abnormal boundaries according to the distribution adjustment model of the abnormal signal weights of the new time slice to realize dynamic parameter adjustment along with data update.
  5. 5. The pet health monitoring system based on microbiome gene detection of claim 4, wherein comparing the real-time collected data with a dynamic baseline model, finding out deviating abnormal activities, marking deviating abnormal behaviors, and submitting the deviating recognition results to sample metadata analysis to obtain abnormal background information, and specifically comprising: Dividing pet gene data acquired in real time into time slices consistent with a dynamic baseline model according to acquisition time, respectively inputting species layer abundance, functional gene layer expression abundance and gene sequence layer characteristics in gene sequence data of each time slice into the dynamic baseline model, and outputting the normal range of distribution of each layer of the time slices by the model; Comparing the distribution of each layer of real-time data with the normal range output by the model, and judging potential abnormal activity if the abundance of a species layer of a certain time slice exceeds the normal range or the expression abundance of a functional gene layer exceeds the normal range or the characteristics of a gene sequence layer exceeds the normal range; Recording the deviation degree of the abnormal behavior, which is the difference between the actual distribution and the normal range, of the determined abnormal behavior with the potential abnormal activity mark, wherein the relevant context information is the scene context information of the time slice; evaluating the severity of the deviation behavior according to the magnitude of the deviation degree and the occurrence frequency of the abnormal activities, wherein the greater the deviation degree is, the higher the frequency is, the higher the abnormal grade is, and each abnormal activity is allocated with an abnormal grade; Extracting acquisition time, acquisition position and recent diet record of the pet from the related context information of the abnormal behavior, searching sample metadata by using the extracted information, and reading the abnormal background information such as the environment state of the sample corresponding to the abnormal behavior during acquisition, specific components in the diet of the pet and specific types of recent contact articles.
  6. 6. The pet health monitoring system based on microbiome genetic testing of claim 5, wherein the analysis of sample metadata based on the anomaly recognition result of the dynamic baseline model finds potential deep health threat patterns and anomaly signals, trace the threat patterns and anomaly signals to determine root causes and anomaly paths, and specifically comprises: Firstly, collecting sample metadata matched with abnormal background information in a deviation recognition result, wherein the sample metadata comprises an acquisition environment state of a time slice corresponding to the abnormal behavior, specific components of pet diet, recent contact article types and historical health records of the pet, and acquiring complete scene information when the abnormal behavior occurs from the sample metadata; Using complete scene information as an analysis framework, carrying out association analysis on the acquired environmental states, specific dietary components, contact object types and historical health records in sample metadata, finding out abnormal dietary components, abnormal environmental states or abnormal contact objects which occur simultaneously with abnormal behaviors, combining abnormal hierarchical association in gene data, identifying potential deep health threat modes as species function association abnormality caused by the dietary components, genetic sequence association abnormality caused by the environmental states or cross-layer association abnormality caused by the contact objects, and identifying abnormal signals as specific data expressions corresponding to the threat modes; Extracting health clues related to abnormal signals from sample metadata, wherein the health clues are diet, environment or contact object conditions when similar abnormality occurs in the historical health records, and identifying threat types as digestive health threats, skin health threats or immune health threats according to the matching degree of the health clues and the current abnormal signals; Tracing the identified abnormal signal from the current time slice to the historical time slice, firstly finding that the direct source of the abnormal signal of the current time slice is the intake of certain dietary components or the change of certain environmental states or the contact of certain contact objects of the previous time slice, and then continuing tracing the source of the previous time slice until the time slice which initially causes the abnormality and the corresponding factors are found, and determining the initial source of the abnormality; And integrating the threat mode, threat type and abnormal path corresponding to the abnormal signal with the original source to generate a combined report.
  7. 7. The pet health monitoring system based on microbiome gene detection of claim 2, wherein the analyzing the gene sequence data is serially connected according to the collection time, and the sample metadata of the corresponding time point is attached to the gene sequence data serially connected at the same time to form a real-time gene data stream, which specifically comprises: Dividing time units for the analyzed gene sequence data according to the hours of the acquisition time, connecting the gene sequence data in each time unit in series according to the acquisition time from the morning to the evening, attaching the sample metadata acquired in the same time unit to the tail of the gene sequence data connected in series according to the sequence of the acquisition time point, and attaching corresponding metadata according to the position priority if a plurality of position samples are acquired at the same time point, so as to form a real-time gene data stream connected in series according to the time units and attached to the metadata.
  8. 8. The pet health monitoring system based on microbiome gene detection of claim 3, wherein the extracting context information of the environmental status of the collection site, the type of components in the pet diet, the type of recently contacted items from each time slice data specifically comprises: Searching the environmental temperature and humidity record of the collecting part and the cleanliness record of the collecting part of the time slice corresponding to the collecting time point from the sample metadata to serve as environmental states, searching protein sources, carbohydrate sources and additive types in the diet of the pet twenty-four hours before the time slice to serve as component types in the diet, and searching toy materials, nest cleanliness and external plant types contacted by the pet seven-twelve hours before the time slice to serve as recently contacted article types.
  9. 9. The pet health monitoring system based on microbiome gene detection of claim 4, wherein updating the overall parameters of the dynamic baseline model with the parameters of the new time slice while adjusting the model to identify thresholds for normal and abnormal boundaries based on the distribution of new time slice abnormal signal weights comprises: The parameters of the new time slice comprise the basic predicted value of each layer of distribution of the time slice and the corrected parameters, the basic predicted value of the new time slice is added into a historical basic feature library of the model, the average value of the historical basic feature library is used for updating the basic weight of each layer of distribution of the model, the corrected parameters of the new time slice are added into a historical correction feature library of the model, and the weight of the correction feature of the model is updated by the median of the historical correction feature library; And counting the distribution of the weight of the abnormal signal of the new time slice, if the duty ratio of the abnormal signal with high weight is increased, increasing the threshold value of the normal and abnormal boundary identification model, so that the abnormal is more strictly identified by the model, and if the duty ratio of the abnormal signal with low weight is increased, decreasing the threshold value, so that the abnormal is more loosely identified by the model, and the boundary threshold value is adjusted along with the distribution of the weight of the abnormal signal.
  10. 10. The pet health monitoring system based on microbiome gene detection of claim 5, wherein the retrieving sample metadata with the extracted information reads the abnormal background information of the environmental state at the time of sample collection corresponding to the abnormal behavior, specific components in the pet diet, specific types of recently contacted articles, specifically comprises: The method comprises the steps of searching an environment temperature and humidity record and a collection position cleanliness record of the same collection time point in sample metadata by using the extracted collection time as an environment state, searching a diet component record of the position corresponding to the collection time in the sample metadata by using the extracted collection position, searching a diet specific component of the time range in the sample metadata by using a time range in the extracted recent diet record, and searching a contact object type record of the time range in the sample metadata by using the first seventy-two hours range of the extracted collection time.

Description

Pet health monitoring system based on microbiome gene detection Technical Field The invention relates to the technical field of pet health monitoring, in particular to a pet health monitoring system based on microbiome gene detection. Background Current pet health monitoring mainly relies on in vitro symptom observation and periodic biochemical examination, and microbiome detection technology has been applied to the pet field, but is often limited to single, static flora composition analysis or specific pathogen screening. In the prior art, the gene sequence data is usually treated as an isolated static snapshot, and the abnormality is judged by comparing a fixed database or a general health standard threshold. Such methods have significant drawbacks. Static analysis cannot capture the dynamic track of the microbiome of the individual pet along with age, diet and environmental changes, and subtle early anomalies deviating from the individual health baseline are difficult to discover in time. Meanwhile, the existing method focuses on analysis of single taxonomic level (such as species abundance), ignores inherent networking association and interaction among multi-level data such as gene functions, metabolic pathways and the like, and is insensitive to potential health threat identification caused by complex ecological relation imbalance and difficult to trace to source. Therefore, a technical scheme is needed, which can continuously and dynamically monitor the pet microbiome, penetrate through the data hierarchy barrier and analyze the internal association mode of the pet microbiome, so that early and accurate health risk early warning and root cause diagnosis are realized. Disclosure of Invention The invention aims to provide a pet health monitoring system based on microbiome gene detection so as to solve the problems in the background art. To achieve the above object, the present invention provides a pet health monitoring system based on microbiome gene detection, the system comprising: The data acquisition and fluidization module is used for determining a microbiome gene data acquisition scheme for pet health monitoring, acquiring gene sequence data and sample metadata from a pet biological sample according to the scheme, and concatenating the acquired data into a real-time gene data stream according to time sequence; The cross-level correlation and primary screening module is used for carrying out cross-level correlation analysis by taking the level logic of the real-time gene data flow as the context, combing different levels of correlation by combining the scene context generated by the data, finding out the correlation and abnormal interaction mode of the gene data among the levels, marking the found result as a preliminary abnormal signal, and collecting the preliminary abnormal signal into an abnormal marked data set with weight; The dynamic baseline modeling module is used for constructing a dynamic baseline model by depending on the real-time gene data flow and the abnormal marker data set, calibrating the baseline model by using the result of cross-level correlation analysis, and updating the dynamic parameter adjustment along with the data; the anomaly comparison and background analysis module is used for comparing the data acquired in real time with the dynamic baseline model, finding out deviated abnormal activities, marking the deviated abnormal behaviors, and transmitting the deviation recognition results to the sample metadata for analysis to obtain abnormal background information; and the deep threat tracing module is used for analyzing sample metadata based on an abnormal recognition result of the dynamic baseline model, finding out a potential deep health threat mode and an abnormal signal, tracing the threat mode and the abnormal signal, and determining a root cause and an abnormal path. Preferably, the determining the microbiome gene data acquisition scheme for pet health monitoring acquires gene sequence data and sample metadata from a pet biological sample according to the scheme, and concatenates the acquired data into a real-time gene data stream according to time sequence, which specifically comprises: Firstly, combining the age stage, daily diet type and recent health state of the pet, determining whether a biological sample to be acquired is a fecal sample, an oral sample or a skin sample, and determining that the gene data acquisition requirement corresponding to each type of sample is full-gene sequencing depth, functional gene targeting acquisition range and sample metadata must contain acquisition time, acquisition parts and recent diet records of the pet; performing gene data capture on biological samples collected in real time according to specific requirements, extracting intestinal microbial genes from fecal samples, extracting oral microbial genes from oral samples and extracting skin microbial genes from skin samples; Classifying the captured gene data according to sam