Effects of a complex feed additive on productivity and blood parameters of laying hens using stochastic fractal-based neural network model

Neural networks (NNs) benefit biomedicine and agriculture, especially when relying on the specificity and implementation of stochastic fractal-supported models. In the poultry industry, a particular challenge is the search for an ideal sorbent-based complex additive to minimize the loss of valuable feed components that can be tailored to groups of gastrointestinal microorganisms. The aim of this study was thus to develop and apply a mathematical model and Gaussian NN to analyze productivity and blood parameters of laying hens when administering a complex feed additive from the mineral shungite sorbent, plus a nutritive supplement of brown seaweed meal. We developed and built a computational NN that modelled the stochastic ManyToOne relationship of an array of hens’ main blood parameters and performance traits. The results presented herein were that the artificial computational stochastic fractal-based NN (EuclidNN) first effectively analyzed the profiles of operational taxonomic units (OTUs) of the physiological/biochemical blood parameters. Also, correlation coefficients were highly positive in relation to certain zootechnical indicators, suggesting that feed additive intake may have led to changes in these performance traits. Calculations suggested that when implementing the feed additive, the values of the Cognitive Salience Index (CSI) vector vCSI2 declined. Hereby, this vector correlates with, and affects the egg production trait. Moreover, there was a certain relationship between the feed additive intake and feed and water consumption. Further, EuclidNN computed the respective bioconsolidation indices of hens and, simultaneously, processed several profiles of OTUs for all experimental variants. It also contributed to the calculation of bioconsolidation index values for each variant, i.e., a quantitative assessment of the physiological/biochemical blood descriptors, depending on diet. Collectively, the poultry productivity prediction based on the developed model and NN is pivotal as an initial step for future improvements of economically important traits in chickens when using novel and efficient complex feed additives.