I explored the possibility of adopting a human step-detecting algorithm to count steps in elephants using wearable technology like the Daily Diary tag. With the peak detection method for step counting, I was able to achieve an average accuracy of 96.0% in my step-counting algorithm (lowest: 92.8%; highest: 100%). Despite being one of the most common methods of step counting, the peak detection method often overestimates the step count which was also observed in this study.
Additionally, I found a significant difference in the mean VeDBA measured in the indoor and the outdoor enclosures (Dunn test, p<0.001), with the VeDBA being higher for the indoor enclosure where the surface was covered with sand compared to both outdoor enclosures where the surface was harder and mud-covered. VeDBA did not differ between the two outdoor enclosures. This indicates that a difference in acceleration associated with a step may be observed due to differences in the substrate over which the movement occurs. Though the test was performed only using a 15-minute subset for each enclosure, it demonstrates that data from biosensors may be used to distinguish movement patterns based on subtle differences, which are often indistinguishable to the naked eye, in the signals logged by the sensors.
Nevertheless, there were some caveats in this study that are important to keep in mind. The results presented in this study are based on just 17 hours of data from one female elephant and are thus not only limited quantitatively but also do not account for variability among individuals. The data were also collected during a period when access to outdoor enclosures was reduced due to severe weather conditions, which limited the variability in data collected from outdoor enclosures.
However, the implications of this study are important for furthering our understanding of activity and movement patterns in elephants. Consistent activity patterns are an indicator of the good welfare of animals housed under human care. Deviations from these activity patterns, often observed by the caretakers in a qualitative manner, are taken as signs of stress and reduced welfare. Body-worn technology that continuously tracks the activity of an animal provides a more consistent estimate of the stress and welfare of animals. Therefore, the development of such technologies will provide us with a quantitative and reliable measure of activity that can be easily analysed to evaluate the well-being of the animal.
This study is a first step towards the development of continuous activity tracking of animals. The high accuracy of the proposed algorithm may be used as a proof of concept for continuous activity logging in a detailed manner and hopefully, be used to further activity tracking of animals in a manner at par with what is observed in humans. Activity monitoring in humans has proved beneficial to the health and well-being of millions of individuals and could produce similar results for animals. This is especially true for endangered species like the Asian elephant about whom there is a notable lack of information regarding their activity budgets and movement patterns. Activity monitoring in animals can not only improve the welfare of animals housed under human care but also provide us with detailed information regarding the physiology and movement of both captive and free-living animals. Detailed information regarding movement patterns as well as daily and seasonal activity patterns may aid in conservation efforts for the Asian elephant.