Facebook-f Linkedin-in Instagram Youtube

A Non-Invasive Approach to Estimate Lactate Based on Data from Wearable Sensors

Facebook
Instagram
LinkedIn

Hamed Darbandi, Carolien Munsters & Paul Havinga (2023)

Background

Managing fatigue in horses is vital for their performance and well-being. Recognizing physical demands and actively monitoring fatigue helps extending longevity, health, and success of equine athletes while upholding ethical standards and public opinion. Traditional blood lactate (LA) tests are invasive, heart rate (HR), however, though informative, can be influenced by numerous factors. This study investigates using inertial measurement units (IMUs) and HR monitors to estimate lactate levels and assess fatigue in sport horses.

The team of equine Integration and Utrecht University is checking the IMU and heart rate sensors on a horse's sacrum and cannon bones for a study. The dressage horse and rider are waiting patiently for all sensors to be prepared and synchronised before training.

Methodology

The study collected data from 21 eventing horses during a standard exercise test using two ProMove-mini IMUs on sacrum and limb, along with HR and LA measurements. IMUs generated two three-dimensional acceleration and angular velocity signals. The sacrum IMU three axes of rotation were x, y, and z, defined in the order as longitudinal, mediolateral, and vertical axes. For limb, x-axis was aligned to the cannon bone, while y- and z-axis were set as longitudinal axis (abduction/ adduction) and mediolateral axis (protraction/retraction). The researchers processed this data to extract movement parameters and used machine learning to estimate lactate levels non-invasively.

Results

The study found that deep learning models using motion sensors and heart rate data effectively estimated lactate levels, with the best model achieving a root mean squared error of 0.11 mmol/L. Sensors on the sacrum provided more accurate fatigue information than limb sensors. The addition of HR data led to a notable accuracy improvement. Specific movement features, like pelvic roll and limb angles, were significant indicators of fatigue. This approach offers a promising non-invasive method for monitoring equine fatigue and enhancing training and welfare.

The study developed a non-invasive model for successful LA and fatigue estimation using both machine learning with input of wearable IMUs ad HR monitors.

Read the full article!
Recente berichten
Categorieën