Acute hepatopancreatic necrosis disease (AHPND) has a wide prevalence, rapid onset, and high mortality for shrimp culture, which is an important limiting factor to affect shrimp aquaculture development in recent years and bring massive economic losses to the industry in worldwide. Systematic study of which factors are significantly correlated with the occurrence of AHPND, and further establishment of a prediction model for the occurrence of shrimp AHPND are important for the prevention and control of the diseases. In this study, the Litopenaeus vannamei in pond culture were preliminarily analyzed the coupling relationship between the occurrence and prevalence of AHPND in shrimps and pathogens, environmental and host autoimmune factors by collecting environmental factors, pathogen abundance, and the host health indicators under AHPND incidence. Then the mathematical early warning model of AHPND occurrence in pond cultured L. vannamei was constructed by Deep Forest Algorithm. In order to explore the relationship between the occurrence of AHPND and environment, pathogen and shrimp immunity, the occurrence of AHPND and its environment, pathogen and shrimp immunity factors in pond cultured L. vannamei were continuously monitored. A total of 18 parameters were collected, including the air temperature, water temperature, salinity, pH, dissolved oxygen (DO), ammonia nitrogen (NH4-N) and nitrite (NO2-N) concentration, culturable bacteria and Vibrio in water, culturable bacteria and Vibrio in shrimp hepatopancreas, the proportion of Vibrio in water and shrimp hepatopancreas, and the activities of ACP, AKP, SOD, LZM, PO in shrimp muscles. The parameter simulation prediction data based on the L. vannamei AHPND occurrence related factor sequence (environmental factor, microbial factor and shrimp health indicator) are constructed for the first time, the one-dimensional sequence is mapped into the three-dimensional space, different kernel functions are selected in combination with the actual classification problem to compare the model fitting accuracy, and the parameters in the model are optimized by the test algorithm. A total of 140 groups of relevant data were collected under the same mode, and the groups of additional pouring of exogenous inputs in the process of breeding were eliminated. By deleting invalid data, there were 100 groups of classified monitoring data, including 25 groups of morbidity data and 75 groups of health data. And due to the dimensional and quantitative differences among different factors, the model will be affected. In order to improve the speed of subsequent experimental training and the accuracy of prediction, the 100 groups of training test data were processed by mapminmax function are normalized for data processing. The relationship between 18 parameters and the occurrence of AHPND in L. vannamei was analyzed by Pearson correlation, and the main influencing factors were further screened by pairwise analysis between the factors. Pearson correlation analysis showed that the incidence of AHPND was positively correlated (P<0.05) with the salinity, the number of culturable bacteria and Vibrio in shrimp hepatopancreas, the proportion of Vibrio in shrimp hepatopancreas, the number of culturable bacteria and Vibrio in water, the activities of LZM, ACP and PO in shrimp muscles, the correlation coefficients were 0.3501, 0.5741, 0.5211, 0.3911, 0.3747, 0.2383, 0.4382, 0.2571 and 0.2289, respectively. which indicating that the AHPND was more likely to occur with the increase of these parameters under a certain range. On the other hand, the incidence of AHPND was negatively correlated with water temperature (P<0.05), and the correlation coefficient was -0.2279. And the water temperature, pH, DO, NH4-N concentration, NO2-N concentration, Vibrio proportion in water, AKP and SOD had little correlation with the incidence of AHPND (P>0.05). Furthermore, in the process of model construction, parameters are removed according to the correlation between parameters and factors. The occurrence of AHPND in L. vannamei was directly and significantly correlated with 7 parameters including the total number of shrimp bacteria, the total number of shrimp Vibrios, LZM, the proportion of shrimp Vibrio, the total number of water bacteria, salinity, and water Vibrios. The prediction performance of three popular integrated learning method algorithms based on decision tree, Deep Forest, LightGBM, and XGBoost, was evaluated by Python language programming, and finally a 4-dimensional vector early warning prediction model based on Deep Forest algorithm for the total number of shrimp bacteria, the proportion of Vibrio shrimp, water bacteria, and salinity was established (accuracy 89.00%). Although the prediction performance of the deep forest model decreased somewhat compared with the support vector machine model (accuracy 100.00%) established in this study, the algorithm was gradually screened out based on the correlation relationship between factors, including the effects of all factors. It is proved that the deep forest model established in this study is the most ideal prediction model for predicting the occurrence of AHPND in L. vannamei among the ten algorithm models tried, and the superiority of the deep forest algorithm is also further verified. The relevant research results provide basic data and technical support for shrimp AHPND disease prediction and health prevention and control, and lay a theoretical foundation for further establishment of aquaculture animal disease early warning theory. |