Intelligent methodologies: An integrated multi-modeling approach to predict adaptive mechanisms in farm animals

Abstract

“Welcome to the Machine” is the second song on Pink Floyd's 1975 album “Wish You Were Here”. Today, 49 years later, machine learning is a field of artificial intelligence that has been gaining prominence, especially in order to solve multifactorial problems such as animal adaptation, which involves thermoregulatory, morphological, hematological and immunological responses. For the first time, we seek to trace the adaptive profile, identify biomarkers of animal adaptation to thermal stress and investigate the potential effects underlying the complex relationships of animal adaptation with climatic variables, morphological, thermoregulatory, hematological, immunological and productive responses in the animal model using an intelligent predictive modeling as an auxiliary method. The database contains 7608 pieces of information. All variables were measured simultaneously on the animal. Machine learning techniques and path analyses were used. Our main results were: (i) we identified patterns that hematological variables represent important biomarkers to study animal adaptation; (ii) there are three adaptive profiles with thermoregulatory, hematological and immunological mechanisms varying according to the thickness of the coat; (iii) age, body condition score and lactation order are factors that influence adaptive responses; and (iv) environmental variables and thermal comfort indices influence thermoregulatory responses (primary mechanism), but do not influence hematological and immunological responses, while thermoregulatory variables influence hematological and immunological responses (secondary mechanism). Most of the variables under study and other indicators such as behavioral, biochemical, metabolic and physical related to the adaptive physiology in production animals are already measured by sensors. These variables can be used in a machine intelligence approach for assertive decision-making by farmers, as technology based on sensors in production animals is an innovative approach to maximize animal welfare and provide better alternatives to evaluate the animal response.