ИСТИНА

The stability of the microbial community is associated with microecologicalinteractions. Antagonism is one these, that determines the benefi ts ofprobiotics for the host. Mathematical models allow to predict the quantitativemeasure of intrapopulation relationships. The aim was to create a predictivemodel for bacterial contamination outcome depending on the probioticsantagonism and prebiotics concentration. It should allow to improve thescreening of synbiotics composition for gut microbial infections prevention.The functional model (fermentation) simulated the distal colon section (N , pH6.8, fl ow rate 0.04 h ). The strains of Bif. adolescentis ATCC 15703 and Bac.cereus ATCC 9634 were chosen as model probiotics and pathogen.Oligofructose Orafti P95 (OF) was used as prebiotics at 5, 7, 10, 12 and 15 gper 1 L of the medium. At the fi rst stage the system was inoculated withbifi dobacteria and dynamic equilibrium (bifi dobacteria count, lactic and aceticacids) was achieved. Then the system was contaminated with bacilli previouslyincubated for 50 h (sporulation). The microbial counts, concentration of acidsand residual carbohydrates were measured. Bacilli mono culture was studiedas a control. The stationary count of bacilli in mono culture was markedlyhigher, while the diff erences were insignifi cant for all co-cultures variants. Theincrease (up to 8 h) of the lag phase (probably, associated with the sporegermination) was observed for higher prebiotics concentration. The specifi cgrowth rate at exponential phase varied at diff erent OF concentrations. Thus,the OF concentration infl uence to two key events of bacterial infection, whichwill together determine when the maximal pathogen count will be reached.This time delay increases the chains for microbiota and host immunity toprevent the infection. The mathematical model was created using artifi cialneural networks (ANN). For the fi rst stage, three ANN were considered, the (https://www.worldmicrobeforum.org/) My Credit Cart Boris Karetkin 2 -1 inputs of which were the OF concentrations, initial bifi dobacteria count, andtime, and the output was bifi dobacteria count, concentrations of lactic andacetic acids. The resulting output variables, OF concentration and the initialbacilli count were as inputs for the contamination stage 3 ANNs. First ANN hastwo outputs (probiotics and bacilli counts) and two others have one output(one of the microbe count). The predictive value of the ANN with two outputsfor the dynamics of bacilli count was high (error less than 5%). A model can beredesigned for three-stage continuous fecal fermentation in the future.