AI-Welfare: An intelligent system for assessing sheep welfare under climate challenges

Abstract

The article presents the AI-Welfare concept as an intelligent analytical system for the comprehensive assessment of sheep welfare, designed to integrate different groups of indicators and provide an overall characterization of the animal’s condition. The proposed approach is based on the combination of physiological, physical, behavioural, clinical and environmental indicators, enabling welfare to be considered as a dynamic and multifaceted state rather than as a set of individual parameters. Particular attention is given to the need to simultaneously consider the functional condition of the animal, its behavioural activity, housing conditions and potential risk factors. The main groups of indicators suitable for the comprehensive assessment of sheep welfare are identified, and the feasibility of integrating them into a unified analytical system is substantiated. Physiological indicators include heart rate, respiratory rate and body temperature; physical indicators include body condition and other characteristics of physical status; behavioural indicators include locomotor activity, resting, feeding and social behaviour; while clinical indicators encompass signs of disease, injury, discomfort and stress. The findings confirm the feasibility of simultaneously using several complementary indicators, since an individual parameter does not always adequately reflect the overall welfare status of an animal. Comprehensive assessment makes it possible to consider the relationships between physiological responses, physical condition, behaviour and environmental conditions. The developed AI-Welfare model provides for a sequential process of sensor-based data acquisition, collection of primary signals, data cleaning and synchronization, data transmission and integration with the results of professional assessment. The subsequent stage involves AI-based analysis aimed at identifying patterns, changes in indicator dynamics, deviations from individual or group profiles, and potential welfare risks. Its practical significance lies in the possibility of integrating data from different sources into a unified monitoring system that can be used to improve the objectivity of welfare assessment, enhance flock management, and support the principles of One Welfare and contemporary approaches to responsible livestock management.