Non-invasive Chinese sturgeon health state prediction method

Abstract

The invention relates to the technical field of fish health monitoring, and discloses a non-invasive Chinese sturgeon health state prediction method, which comprises the following steps of: acquiring excrement samples of a plurality of Chinese sturgeons in a suction manner, respectively performing physical characteristic detection and biochemical characteristic detection on the excrement samples, and constructing a data set; dividing respective proportions of physical features and biochemical features through expert analysis, then analyzing detection results of the physical features, making an expert score table to obtain corresponding scores and weights, and training a random forest model by using detection results of the biochemical features; and predicting the health state of the Chinese sturgeons to be detected by using the formulated expert score table and the trained random forest model.

Description

Technical Field

[0001] The invention belongs to the technical field of fish health monitoring, and in particular relates to a non-invasive method for predicting the health status of Chinese sturgeon. Background Art

[0002] Current non-invasive technologies for fish health monitoring mainly focus on behavioral monitoring and environmental factor analysis. By monitoring the activity patterns of fish and the chemical composition of the surrounding environment, researchers can infer the health status of fish. However, these technologies often rely on indirect environmental factors (such as water quality) and cannot directly reflect the internal health status of fish.

[0003] In terms of animal feces testing, existing studies have used fecal analysis to infer animal health, but especially in aquatic environments, feces collection and analysis is more difficult and is easily affected by dilution in the water, resulting in reduced accuracy of test data.

[0004] Therefore, a practical prediction method is urgently needed to obtain the health status of Chinese sturgeon. Summary of the invention

[0005] The present invention provides a non-invasive method for predicting the health status of Chinese sturgeon, which solves the technical problems in existing feces monitoring methods, such as difficulty in feces collection, low accuracy of detection data, and inability to directly reflect the intrinsic health of fish.

[0006] The present invention can be achieved through the following technical solutions:

[0007] A non-invasive method for predicting the health status of Chinese sturgeon, comprising the following methods:

[0008] Step 1: Obtain multiple fecal samples of Chinese sturgeon by suction, perform physical property detection and biochemical characteristic detection on these fecal samples, and construct a data set;

[0009] Step 2: Through expert analysis, the weight of physical characteristics and biochemical characteristics is divided, and then the test results of physical characteristics are analyzed, and an expert scoring table is formulated to obtain the corresponding scores and weights. At the same time, the test results of biochemical characteristics are used to train the random forest model;

[0010] Step 3: Use the prepared expert scoring table and the trained random forest model to predict the health status of the Chinese sturgeon to be inspected.

[0011] Furthermore, the trained random forest model was used to predict the biochemical characteristics of the Chinese sturgeon to be inspected, and the scores and weights corresponding to each biochemical characteristic were obtained. Then, they were accumulated with the scores and weights corresponding to the physical characteristics to obtain the total score. Finally, the health status of the Chinese sturgeon to be inspected was judged according to the score range in which the total score fell.

[0012] Furthermore, a suction mechanism is used to perform suction on the anus of the Chinese sturgeon to obtain a fecal sample. The suction mechanism includes a syringe, the needle of the syringe is connected to a hose, a suction cup is provided at the free end of the hose, and a centrifuge tube is connected to the barrel of the syringe, and the centrifuge tube is used to carry the fecal sample.

[0013] Furthermore, the physical characteristics are obtained through appearance inspection and microscopic inspection, including color characteristics, texture characteristics, shape characteristics of the feces appearance, and the composition of pathogenic microorganisms and parasite types that may be contained inside; the biochemical characteristics include metabolic / immune enzyme activity indicators and microbial community structure indicators, wherein the metabolic / immune enzyme activity indicators are obtained through biochemical analysis, and the microbial community structure indicators are sequenced using the Illumina MiSeq high-throughput sequencing platform to obtain 16SrRNA gene sequence data, which is then analyzed.

[0014] Furthermore, the total score is set at 100 points, where a score of 80-100 represents complete health, a score of 70-79 indicates a minor health problem, a score of 60-69 indicates a potential health risk, and a score below 60 indicates a serious health problem.

[0015] Calculate the total score using the following formula:

[0016] Total score = ∑ (indicator score × corresponding weight) + expert score.

[0017] Furthermore, the health status report is output in combination with the historical health status data.

[0018] The beneficial technical effects of the present invention are:

[0019] 1. Combining the fecal data of healthy parent fish of Chinese sturgeon, the random forest model is used in combination with the expert evaluation form for real-time calculation and analysis, which can realize non-invasive and instant health monitoring of Chinese sturgeon: Monitoring through fecal data does not need to directly interfere with the physiological state of the fish, which is a non-invasive and relatively harmless detection method. By collecting fecal data of Chinese sturgeon, a "snapshot" of the internal health of the fish can be collected, such as the nutritional status, digestive function, immune response, etc. of the fish, which will affect the composition and quality of its feces. These data are very representative as a proxy for health status. Then, when a machine learning model such as the random forest model is used for real-time analysis, the analysis method based on real-time data stream can monitor the health status of Chinese sturgeon in real time.

[0020] Since the model can quickly calculate and output health assessment results through real-time input of fecal data, this automated and immediate feedback mechanism makes health monitoring more efficient and flexible. Especially during the broodstock breeding period, the system can detect potential health problems in advance, thereby effectively ensuring the breeding success rate.

[0021] 2. The health prediction method of the present invention can significantly reduce the interference caused by traditional methods, improve the accuracy and efficiency of health assessment, and provide a more scientific and efficient solution for the breeding and conservation of Chinese sturgeon. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the overall process of the method of the present invention;

[0023] Figure 2 It is a schematic structural diagram of the suction mechanism of the present invention. DETAILED DESCRIPTION

[0024] The specific implementation of the present invention is described in detail below with reference to the accompanying drawings and preferred embodiments.

[0025] like Figure 1 As shown, the present invention provides a non-invasive method for predicting the health status of Chinese sturgeon. By online detection of the excrement of Chinese sturgeon, such as feces, biochemical characteristic data and physical characteristic data are obtained, an expert scoring table is formulated using the physical characteristic data, and a random forest model is trained using the biochemical characteristic data to establish a corresponding health prediction model to evaluate the health status of Chinese sturgeon, especially broodstock, during the breeding stage.

[0026] The details are as follows:

[0027] Step 1: Use suction to obtain multiple fecal samples of Chinese sturgeon, perform physical and biochemical characteristic tests on these fecal samples, and construct a data set.

[0028] First, use the suction mechanism to suction the anus of the Chinese sturgeon to obtain a fecal sample, such as Figure 2 As shown, the suction mechanism includes a syringe, the needle of which is connected to a hose, a suction cup is provided at the free end of the hose, and a centrifuge tube is connected to the syringe barrel of the syringe, which is used to carry the fecal sample. When collecting feces, the suction cup can be placed close to the anus surface of the Chinese sturgeon, and a professional can gently press its abdomen while using a syringe to suction, and then push the syringe to transfer the collected feces to the sealed centrifuge tube through the hose, thereby completing the safe collection of fecal samples. Then, the physical and biochemical characteristics of the fecal sample are extracted.

[0029] The collection operation should be performed by trained professionals to ensure that the force is moderate and does not cause excessive squeezing or irritation to the fish body. At the same time, use a suction cup made of soft material to reduce friction with the anal skin. The connection between the hose and the syringe should be smooth to avoid strong negative pressure during suction and damage to the anal tissue.

[0030] 1. Sample pretreatment: including sample dilution, filtration and preservation for subsequent analysis.

[0031] Take 200-500 mg of stool sample, put it into a sterile 2 ml centrifuge tube, add 1xPBS solution, shake and mix, then centrifuge at 10,000 rpm for 3 minutes, discard the upper liquid, and keep the precipitate.

[0032] 2. Obtain physical characteristic data through visual inspection and microscopic inspection.

[0033] Visual examination: Record the color, texture, shape and other appearance characteristics of the stool sample, which may be related to the health of the digestive system.

[0034] Microscopic examination: The possible pathogenic microorganisms and parasites in the feces are observed under a microscope. Microscopic examination can directly observe biological factors such as microorganisms, parasite eggs, larvae, etc. in fecal samples, and provide basic data for early detection of infection and disease.

[0035] Based on the above data, an expert scoring table is formulated through expert analysis to prepare for the subsequent formation of specific scoring standards.

[0036] Note: If any parasite is found in fecal microscopy, there is no need to score using microorganisms or other indicators and the individual is directly defined as an unhealthy individual, as follows (but not limited to the following pathogens).

[0037] Major categories of pathogens Pathogen Trematoda Fasciola hepatica Schistosoma spp. Tapeworms (Cestoda) Spirometra mansoni Nematodes Anisakis spp. Intestinal nematodes (Capillaria spp.) Acanthocephala Pomphorhynchus spp. Flagellates Giardia spp. Flagellates (Hexamita spp.) Trypanosoma spp. Protozoa Ovulate worms (Eimeria spp.) Cryptosporidium spp. Microsporidia Microsporidia spp. Monogenea Gyrodactylus spp. Ciliophora. Apiosoma spp. Ichthyophthirius multifiliis

[0038] 3. Obtain biochemical characteristic data through biochemical analysis

[0039] Ⅰ. Metabolic / immune enzyme activity indicators obtained through biochemical analysis

[0040] Detect enzyme activity in stool to assess digestive system function.

[0041] II. Microbial community structure indicators were sequenced using the Illumina MiSeq high-throughput sequencing platform to obtain 16SrRNA gene sequence data, which were then analyzed and obtained as follows:

[0042] First, nucleic acid was extracted using a commercial kit (OMEGA kit EZNA TMMag-Bind Soil DNA Kit is used as an example to extract total DNA from stool samples. During the extraction process, attention should be paid to the degree of sample dissolution. For cells that are difficult to dissolve, such as G+ bacteria, it may be necessary to increase the lysis temperature or use additional physical or chemical methods to increase the efficiency of cell lysis.

[0043] The DNA is then quality tested: the concentration and purity of the DNA is tested using a spectrophotometer to ensure that the A260 / A280 ratio of the DNA is between 1.8-2.0 and the A260 / A230 ratio is above 2.0 to ensure that the DNA is suitable for subsequent PCR amplification.

[0044] PCR amplification: PCR amplification of the V3-V4 region of the 16S rRNA gene is performed using specific primers. These primers can cover a wide range of bacteria while providing sufficient resolution to distinguish different bacterial species. After successful extraction, two rounds of PCR amplification are performed: The first round of PCR amplification system includes: 15μl 2x Robust PCR Master Mix, 1 μl Bar-PCR Primer F, 1 μl Primer R, (Nobar_341F CCTACGGGNGGCWGCAG) (Nobar_805RGACTACHVGGGTATCTAATCC).

[0045] 10-20ng PCR product, 9-12μl H2O; Illumina bridge PCR compatible primers were introduced in the second round of amplification. System includes: 15μl 2x Robust PCR Master Mix, 1 μl primer F, 1 μl Index-PCR primer R, 10-20 ng PCR product, 9-12 μl H2O. The library size was detected by 2% agarose gel electrophoresis, and the library concentration was determined using Qubit3.0 fluorescence quantification instrument. All samples were mixed in a 1:1 ratio.

[0046] Sequencing library construction: The purified PCR products are used to construct a sequencing library, which will be used for high-throughput sequencing.

[0047] High-throughput sequencing: Sequencing is performed using Illumina MiSeq or other high-throughput sequencing platforms. During the sequencing process, 16S rRNA gene sequence data is obtained, and the obtained PEreads are first spliced ​​according to the overlap relationship. After distinguishing the samples, the sequence quality is quality controlled and filtered, and then OTU clustering (ASV denoising) analysis and species taxonomy analysis are performed. Based on the results of OTU clustering (ASV denoising) analysis, various diversity index analyses can be performed on OTU (ASV), as well as the detection of sequencing depth; based on taxonomic information, statistical analysis of community structure can be performed at various classification levels. Based on the above analysis, a series of in-depth statistical and visual analyses such as Beta diversity analysis, group test analysis, difference significance test, environmental factor association analysis, association and model prediction analysis, and functional prediction can be performed on the community composition and phylogenetic information of multiple samples.

[0048] Finally, the data analysis platform is used to analyze the microbial population: data analysis is automatically performed through bioinformatics tools such as QIIME2, Mothur, or the microbiome analysis pipeline SMURF.

[0049] The specific indicators are as follows:

[0050] (1) Microbial community structure: The microbial community structure in feces can directly reflect the health status of Chinese sturgeon. Compared with healthy Chinese sturgeon, the microbial diversity and richness of diseased individuals may change significantly, and some specific microorganisms may even become markers of disease. For example, pockmark disease can destroy the microecological balance of the skin and mucous membrane layer of Chinese sturgeon, causing its dominant flora to gradually change from Lactobacillus, Acinetobacter, etc. to Flavobacterium. The relative abundance of Flavobacterium can be used as an important indicator for assessing the risk of disease in Chinese sturgeon. By analyzing the biological significance of microorganisms classified by each genus, a corresponding weighted proportion can be assigned to each flora to more accurately assess the health status of Chinese sturgeon.

[0051] ①. Microbial diversity and richness: There may be significant differences in the microbial diversity and richness between healthy and diseased Chinese sturgeons.

[0052] ② Changes in dominant bacterial flora: At the genus classification level, the dominant genera in healthy Chinese sturgeon may include Lactobacillus, Acinetobacter, Pseudomonas, etc., while diseased Chinese sturgeon may have Flavobacterium as the main dominant genus. The relative abundance of these genus can be used to assess the disease risk of Chinese sturgeon.

[0053] ③. Presence of potential pathogens: Potential pathogens enriched in aquaculture water and mucus, such as Citrobacter, Acinetobacter, Flavobacterium, etc., may cause diseases under certain conditions.

[0054] ④. Ratio of Gram-positive and Gram-negative bacteria: The dominant flora in the intestinal contents may be Gram-positive bacteria, while the dominant flora in mucus and aquaculture water may be Gram-negative bacteria. This change in ratio may be related to the health status of Chinese sturgeon.

[0055] ⑤. Composition of microbial community structure: At the phylum level, mucus and intestinal contents may be dominated by Actinobacteria, Proteobacteria, and Firmicutes, while aquaculture water may be dominated by Proteobacteria, Bacteroidetes, and Firmicutes. Changes in these community structures can reflect the health status of Chinese sturgeons.

[0056] ⑥. Nutrients and metabolites: The detection of nutrients and metabolites, especially fat, protein and other components in feces, can be used to evaluate the nutritional status and digestion and absorption capacity of Chinese sturgeons, thereby indirectly reflecting their overall health status. Key detection indicators include lipase, protease and amylase activities, the levels of which can provide important information about the metabolic function of Chinese sturgeons.

[0057] ⑦. Pathogen detection: The detection of specific pathogens in feces, such as bacteria, viruses, parasites, etc., can directly indicate the risk of infectious diseases. The parasitic diseases that infect sturgeons mainly include spore diasis, Ichthyophthirius, Pseudocercosporidia, Trichodina, Trypanosomiasis, Trichodina, and Trichodina.

[0058] Step 2: Through expert analysis, the proportion of physical characteristics and biochemical characteristics is divided, and then the detection results of physical characteristics are analyzed, and an expert scoring table is formulated to obtain the corresponding scores and weights. At the same time, the detection results of biochemical characteristics are used to train the random forest model.

[0059] ⅰ. Develop an expert scoring table

[0060] Through expert analysis of a large amount of experimental data, the weight of physical characteristics and biochemical characteristics can be divided according to a 25:75 ratio, and an expert scoring table is formulated as follows:

[0061]

[0062] ii. Training the Random Forest Model

[0063] Through the data transmission system, the collected fecal index information of Chinese sturgeon will be transmitted to the processing unit, and then the data will be parsed and preprocessed through a specific algorithm. Based on this, the system will build a fecal analysis model specifically for analyzing healthy Chinese sturgeon. The test results will be compared with known health standards. To ensure accurate analysis, the system will normalize each index so that the observed values ​​are proportionally adjusted within the healthy range. Next, the system will calculate each input indicator, analyze the degree of deviation from the healthy range, and assign different weights to different indicators according to their importance in physiology. For example, indicators such as proteases and lipases that directly affect digestion and metabolism will be given higher weights, while specific bacterial phyla in certain microbial communities may have less impact on health, so the weights are relatively low. Through weight and deviation analysis, the system will calculate a comprehensive score.

[0064] The calculation of weights is based on the feature importance of the random forest model and other ensemble learning models. Feature importance is a metric used to measure the contribution of a feature to the predictive power in a decision tree. In the random forest model, each decision tree split node is divided based on the value of a feature, and the importance of the feature is calculated by the contribution of the feature in the classification or regression task.

[0065] (1) Data preparation: Divide the data into features (X) and labels (Y), where X represents the observed data point and Y represents the healthy or unhealthy state. In order to ensure that the calculation standards of all features are consistent, the data needs to be standardized or normalized.

[0066] Standardized difference calculation formula:

[0067] (2) Use the RandomForestClassifier in the scikit-learn library to build and train the model. 100 labeled simulated samples are randomly generated for training to ensure that the model can capture the difference between healthy and unhealthy states.

[0068] (3) After training, the performance of the model was evaluated using a test set, including calculation of the model’s accuracy, confusion matrix, and other evaluation indicators to ensure that the model could accurately predict the health status of Chinese sturgeon.

[0069] (4) After training is complete, use the feature_importances_ attribute to extract the importance score of each feature. The importance of a feature is usually between 0 and 1, and the sum of the importance of all features is 1. The importance of a feature is calculated by accumulating the reduction in impurity for each feature on each decision tree.

[0070] Each time a decision tree selects a feature for data partitioning, the system calculates the reduction in impurity brought by the feature. After training, the model accumulates the contribution of each feature on each decision tree and takes the average to get the feature importance score.

[0071] (5) Use the joblib or pickle library in Python to save the model to a file for subsequent use.

[0072] (6) Once the model is saved, it can be loaded at any time and new stool test data can be input into the model for health prediction. The obtained weights are shown in Table 1.

[0073] Table 1: Evaluation indicators and weights of Chinese sturgeon feces

[0074]

[0075] Total score = ∑ (indicator score × corresponding weight) + expert score

[0076] The total score set by the model is 100 points, with 80-100 points representing complete health, 70-79 points indicating minor health problems, 60-69 points indicating potential health risks, and below 60 points indicating serious health problems. If individual important indicators do not reach the healthy range, the system will provide additional health status categories and add additional testing or monitoring suggestions for minor and potential problems.

[0077] In addition, the system also introduces online learning and semi-supervised learning mechanisms, which enable the model to continuously optimize and self-adjust based on the continuous accumulation of more data. Through this mechanism, the system will update and adjust the health assessment criteria in real time based on the latest test data to ensure that the assessment results are always accurate and reliable.

[0078] The system also has the function of generating detailed health test reports, including test results, health assessments, and possible health risks. The analysis results will be output through the terminal device, and users can view detailed test reports and health status predictions through display screens or other devices. At the same time, the online testing system is equipped with a data storage function that can record the data and analysis results of each test. This not only facilitates long-term trend analysis and historical data comparison, but also allows continuous monitoring of individual health changes.

[0079] Step 3: Use the prepared expert scoring table and the trained random forest model to predict the health status of the Chinese sturgeon to be inspected.

[0080] In order to verify the feasibility of the prediction method of the present invention, we conducted the following experiments:

[0081] Example: Health prediction composition of Chinese sturgeon A

[0082] 1. Sample collection

[0083] Collection object: Chinese sturgeon A, 4 years old, living in an artificial breeding environment.

[0084] Collection method: When the Chinese sturgeon is in a stable state, fecal samples are collected by suction mechanism. 250 mg of fecal samples are collected and placed in a sterilized 2 ml centrifuge tube for subsequent analysis.

[0085] 2. Sample Pretreatment

[0086] Dilution and filtration: Add 1 ml of PBS solution, shake and mix, then centrifuge at 10,000 rpm for 3 minutes, discard the supernatant liquid, and keep the precipitate for analysis.

[0087] 3. Appearance inspection

[0088] Results: The stool was dark brown in color, soft in texture, and irregular in shape. The overall appearance score was 7.

[0089] 4. Microscopic Examination

[0090] Results: No parasites were found, but a small amount of pathogenic microorganisms were present. Microscopic score was 4.

[0091] 5. Biochemical Analysis

[0092] A. Enzyme activity:

[0093] Lipase: 110U / g prot (healthy range: 103-160U / g), score 0.8, weight 0.02.

[0094] Amylase: 150 U / g prot (healthy range: 102-180 U / g), score 1.0, weight 0.03.

[0095] Protease: 190U / g prot (healthy range: 150-205U / g), score 1.0, weight 0.04.

[0096] B. Analysis of microbial community structure

[0097] The total DNA in the feces was extracted using the OMEGA kit, and the lysis and centrifugation steps were successful. The quality of the extracted DNA met the requirements of subsequent PCR amplification. The DNA was then sequenced using the Illumina MiSeq high-throughput sequencing platform to obtain 16SrRNA gene sequence data, which was then analyzed.

[0098] Chao1 Index: 900 (Health Range: 700-1300), Score 1.0, Weight 0.03

[0099] ACE Index: 1100 (Healthy Range: 700-1300), Score 1.0, Weight 0.05

[0100] Shannon Index: 3.8 (Healthy Range: 3.0-4.5), Score 1.0, Weight 0.04

[0101] Simpson Index: 0.92 (healthy range: 0.85-0.97), score 1.0, weight 0.06

[0102] Gram-negative bacteria: 55% (healthy range: 40%-60%), score 1.0, weight 0.06

[0103] Gram-positive bacteria: 45% (healthy range: 40%-60%), score 1.0, weight 0.03

[0104] Firmicutes: 20% (healthy range: 10%-30%), score 1.0, weight 0.05

[0105] Bacteroidetes: 15% (healthy range: 10%-20%), score 1.0, weight 0.03

[0106] Proteobacteria: 25% (healthy range: 10%-40%), score 1.0, weight 0.03

[0107] Actinobacteria: 8% (healthy range: 5%-10%), score 1.0, weight 0.03

[0108] Thermotogae: 2.5% (healthy range: 2%-3%), score 1.0, weight 0.03

[0109] Bacillariophyta: 6% (healthy range: 5%-10%), score 1.0, weight 0.03

[0110] Lactobacillus: 12% (healthy range: 5%-30%), score 1.0, weight 0.03

[0111] Bacteroides: 17% (healthy range: 10%-20%), score 1.0, weight 0.01

[0112] Prevotella: 7% (healthy range: 5%-10%), score 1.0, weight 0.01

[0113] Helicobacter (Helicobacter pylori): Not detected, score 1.0, weight 0.01

[0114] Citrobacter: 3% (healthy range: 0%-5%), score 1.0, weight 0.03

[0115] Acinetobacter: 5% (healthy range: 1%-10%), score 1.0, weight 0.03

[0116] Flavobacterium: 3% (healthy range: 0%-5%), score 1.0, weight 0.03

[0117] Pseudomonas: 4% (healthy range: 0%-5%), score 0.8, weight 0.03

[0118] 6. Total score calculation

[0119] Through the weight calculation of the health model, the comprehensive score of Chinese sturgeon A is obtained:

[0120] Total score = 84 points (based on the weight and score accumulation of each indicator)

[0121] 7. Health assessment and advice

[0122] Overall comment: Combined with the previous health records, the overall health of Chinese sturgeon A is good, all major indicators are within the normal range, and the lipase activity is slightly low but has no major impact.

[0123] Recommendation: Continue to maintain the current breeding environment, regularly check fecal enzyme activity, and ensure normal digestive function. Through the system's self-analysis of health-related microbial indicators (such as the abundance of lactic acid bacteria, bifidobacteria, and nitrifying bacteria), as well as related enzyme activities (such as proteases), combined with the random forest model, the system can predict the health status of Chinese sturgeons and generate a health report.

[0124] Although specific embodiments of the present invention are described above, those skilled in the art should understand that these are merely examples and that various changes or modifications may be made to these embodiments without departing from the principles and essence of the present invention. Therefore, the scope of protection of the present invention is limited by the appended claims.

Claims

1. A non-invasive method for predicting the health status of Chinese sturgeon, characterized in that The following methods are included: Step 1: Obtain multiple fecal samples of Chinese sturgeon by suction, perform physical property detection and biochemical characteristic detection on these fecal samples, and construct a data set; Step 2: Through expert analysis, the weight of physical characteristics and biochemical characteristics is divided, and then the test results of physical characteristics are analyzed, and an expert scoring table is formulated to obtain the corresponding scores and weights. At the same time, the test results of biochemical characteristics are used to train the random forest model; Step 3: Use the prepared expert scoring table and the trained random forest model to predict the health status of the Chinese sturgeon to be inspected.

2. The non-invasive method for predicting the health status of Chinese sturgeon according to claim 1, characterized in that: The trained random forest model was used to predict the biochemical characteristics of the Chinese sturgeon to be inspected, and the scores and weights corresponding to each biochemical characteristic were obtained. Then, the scores and weights corresponding to the physical characteristics were added together to obtain the total score. Finally, the health status of the Chinese sturgeon to be inspected was judged according to the score range in which the total score fell.

3. The non-invasive method for predicting the health status of Chinese sturgeon according to claim 2, characterized in that: A suction mechanism is used to perform suction operation on the anus of the Chinese sturgeon to obtain a fecal sample. The suction mechanism includes a syringe, the needle of the syringe is connected to a hose, a suction cup is provided at the free end of the hose, and a centrifuge tube is connected to the barrel of the syringe, and the centrifuge tube is used to carry the fecal sample.

4. The non-invasive method for predicting the health status of Chinese sturgeon according to claim 3, characterized in that: The physical characteristics are obtained through appearance inspection and microscope inspection, including color characteristics, texture characteristics, shape characteristics of the feces appearance, and the composition of pathogenic microorganisms and parasite types that may be contained inside; the biochemical characteristics include metabolic / immune enzyme activity indicators and microbial community structure indicators, wherein the metabolic / immune enzyme activity indicators are obtained through biochemical analysis, and the microbial community structure indicators are sequenced using the Illumina MiSeq high-throughput sequencing platform to obtain 16SrRNA gene sequence data, which is then analyzed.

5. The non-invasive method for predicting the health status of Chinese sturgeon according to claim 2, characterized in that: The total score is set to 100 points, with scores of 80-100 representing complete health, scores of 70-79 indicating minor health problems, scores of 60-69 indicating potential health risks, and scores below 60 indicating serious health problems. Calculate the total score using the following formula: Total score = ∑ (indicator score × corresponding weight) + expert score.

6. The non-invasive method for predicting the health status of Chinese sturgeon according to claim 1, characterized in that: Combine historical health status data to output health status report.

Images