Evaluation of the sleeping pattern in horses: etogram development and use of an innovative non-invasive method for clinical monitoring of stabled horses

Abstract

The sleep in humans is much studied and generates useful and applicable information in improving the welfare of people and athletic performance of athletes. Appropriate rest is considered one of the pillars for success in therapeutic approaches and as part of training for athletes to achieve victories. In the sleep horses, there are about thirty NREM phases all day long, in this case the horse can rest standing. However, to achieve the REM sleep it must be on recumbency position. Complete rest is achieved by cyclic alternation of this phases. During hospital care in horses it is possible to identify a lot of behavioral changes due to hospitalization. However, the rest influence in hospitalized and sick animals has never been studied in this species. It is unknown how much hospitalization can change the physiological parameters of animals during the therapeutic approach, influencing both their recovery and hospitalization time, as well as the occurrence of clinical complications. In addition, it is speculated that interference with resting time is related to difficulty in treating and/or handling horses, increasing stress in the hospital environment. The aim of this study is to assess whether factors such as hospitalization, type of injury, affected system and exercise influence the resting time (increase or decrease). Our hypothesis is these factors interfere with the resting time, emotional and physical state of the horses. These data are expected to demonstrate the need for intervention to improve rest and minimize time of hospitalization, as well as its deleterious effects. For this study it will be used horses that are hospitalized for at least five days at the Veterinary Hospital of the School of Veterinary Medicine and Animal Science of University of São Paulo and at the Teaching Hospital - Texas A&M University - EUA. Horses in training of the Texas A&M University, in their natural environment and when hospitalized, will be used in the study. They will be monitored with recordings of resting time (using etogram), clinical parameters and heart rate variability. The data obtained regarding resting times will be labeled and used for the development of an automated reading system, involving artificial intelligence using the Deep Learning method, through the training of a convolutional neural network (CNN). After reading and labeling the data, a correlation will be made between resting times, physiological parameters and heart rate variability with hospitalization time, type of diagnosed lesion, affected system and exercise. This method may in the future be incorporated into the concept of "smart hospital", where artificial intelligence will be used as an aid to veterinarians in detecting behavioral and clinical changes in the hospitalized horse. (AU)