Prediction of poor health in small ruminants and companion animals with accelerometers and machine learning

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Global warming is one of the biggest challenge of our times, and significant efforts are being undertaken by academics, industries and other actors to tackle the problem. In the agricultural field precision farming is part of the solution to environmental sustainability and has been researched increasingly in recent years. Indeed, it has the potential to effectively increase livestock yield and decrease production carbon footprint while maintaining welfare.
The thesis begins with a review of developments in automated animal monitoring and then moves on to a case study of a health monitoring system for small-ruminant in South Africa.
As a demonstration and validation of the potential use case of the system, the method we propose is then applied to another study which aims to study feline health. Lower and Middle Income countries will be strongly affected by the changing climate and its impacts. We devise our method based on two South African small scale sheep and goat farms where assessment of the health status of individual animals is a key step in the timely and targeted treatment of infections, which is critical in the fight against anthelmintic and antimicrobial resistance. The FAMACHA scoring system has been used successfully to detect anaemia caused by infection with the parasitic nematode Haemonchus contortus in small ruminants and is an effective way to identify individuals in need of treatment. However, assessing FAMACHA is labour-intensive and costly as individuals must be manually examined at frequent intervals. Here, we used accelerometers to measure the individual activity of extensively grazed small ruminants exposed to natural Haemonchus contortus worm infection in southern Africa over long time scales (13+ months). When combined with machine learning for missing data imputation and classification, we find that this activity data can predict poorer health as well as those individuals that respond to treatment, with precision up to 80%. We demonstrate that these classifiers remain robust over time. Interpretation of trained classifiers reveals that poorer health can be predicted mainly by the night-time activity levels in the sheep. Our study reveals behavioural patterns across two small ruminant species, which low-cost biologgers can exploit to detect subtle changes in animal health and enable timely and targeted intervention. This has real potential to improve economic outcomes and animal welfare as well as limit the use of anthelmintic drugs and diminish pressures on anthelmintic resistance in both commercial and resource-poor communal farming.
The validation of the proposed techniques with a different study group will be discussed in the latter part of the thesis. We used the accelerometry data of indoor cats equipped with wearable accelerometers in conjunction with their health status to detect signs of degenerative joint disease, and adapted our machine-learning pipeline to analyse bursts of high activity in the cats. We were able to classify high-activity events with precision up to 70% despite the relatively small dataset adding further evidence to the viability of animal health monitoring with accelerometers.
Date of Award5 Dec 2023
Original languageEnglish
Awarding Institution
  • University of Bristol
SupervisorChristos C Ioannou (Supervisor), Andrew Dowsey (Supervisor) & Tilo Burghardt (Supervisor)


  • Haemonchus contortus
  • Anthelmintic Resistance
  • Precision Livestock Farming
  • Accelerometers
  • Machine learning

Cite this