Abstract
Gram-negative bacterial infections, particularly Salmonella and Escherichia coli (E. coli) represent significant threats to poultry production worldwide, causing substantial economic losses and public health concerns. Therefore, this study was carried out to investigate the prevalence of these pathogens in some broiler farms and assessing associated risk factors using machine learning. A cross-sectional study was conducted from August 2021 to March 2023, collecting 764 samples (284 environmental and 480 non-environmental) from Sharkia, Ismailia, Damietta, and Mansoura Governorates, Egypt, from broiler farms with two different housing systems (deep litter and battery systems).Bacterial isolation using selective media including MacConkey agar, XLD, and EMB and identification using conventional biochemical tests, Vitek2c2 compact system and confirmed via serological typing. Machine learning including artificial neural networks (ANN) and decision trees were applied to identify risk factors and develop predictive models to be applied in farms. The results revealed that in the environmental samples, E. coli was most prevalent in air (10.61%) and litter (6.82%), while Salmonella spp. was detected at lower rates in air (2.27%) and litter (1.52%).In the non-environmental samples, E. coli was most prevalent in cloacal swabs (9.11%) and liver (7.38%), whereas Salmonella spp. showed higher prevalence in liver (5.86%), yolk sac (3.47%), and heart (3.25%).There was a strong relationship between Salmonella and E.coli existence which was confirmed with hierarchical cluster analysis, The artificial neural networks (ANN) model demonstrated excellent predictive performance with area under the curve (AUC) values of 0.926 for Salmonella and 1.0 for E. coli. Decision tree analysis revealed that antibiotic type was the primary predictor for both pathogens, followed by bird age, housing system, and environmental factors including temperature and humidity. Multivariate analysis identified several critical risk factors: single antibiotic use (versus combination therapy), specific housing systems (semi-closed deep litter for Salmonella, all systems for E. coli), bird age (≤4 days for Salmonella, all ages for E. coli), and environmental conditions including temperature (>28.5°C) and humidity (>44.5%). Breed type also significantly influenced pathogen prevalence, with certain breeds showing higher susceptibility.

Key words: Broilers, Salmonella, Escherichia coli, risk assessment, machine learning.