Predicting Hydration Status Using Machine Learning Models From Physiological and Sweat Biomarkers During Endurance Exercise: A Single Case Study

Abstract

Improper hydration routines can reduce athletic performance. Recent studies show that data from noninvasive biomarker recordings can help to evaluate the hydration status of subjects during endurance exercise. These studies are usually carried out on multiple subjects. In this work, we present the first study on predicting hydration status using machine learning models from single-subject experiments, which involve 32 exercise sessions of constant moderate intensity performed with and without fluid intake. During exercise, we measured four noninvasive physiological and sweat biomarkers including heart rate, core temperature, sweat sodium concentration, and whole-body sweat rate. Sweat sodium concentration was measured from six body regions using absorbent patches. We used three machine learning models to determine the percentage of body weight loss as an indicator of dehydration with these biomarkers and compared the prediction accuracy. The results on this single subject show that these models gave similar mean absolute errors, while in general the nonlinear models slightly outperformed the linear model in most of the experiments. The prediction accuracy of using the whole-body sweat rate or heart rate was higher than using core temperature or sweat sodium concentration. In addition, the model trained on the sweat sodium concentration collected from the arms gave slightly better accuracy than from the other five body regions. This exploratory work paves the way for the use of these machine learning models to develop personalized health monitoring together with emerging, noninvasive wearable sensor devices.

Real-time smart multisensing wearable platform for monitoring sweat biomarkers during exercise

“…For all models, WBSR and HR gave lower MAEs than the other biomarkers. This is in line with the recent findings where four biomarkers collected from a single subject were used to predict %BWL [12]. We also found that these three models gave similar MAEs.…”

“…MAE is defined as: MAE(y, ŷ) = N i=1 |y i − ŷi |/N , where N is the number of samples, ŷi is the i-th ground truth %BWL, and y i is the predicted %BWL. Details of the implementation can be found in [12].…”

“…Subjects were asked to arrive well hydrated and their hydration status were controlled using USG measurements as described in [12]. Subjects then entered a chamber with a controlled environment (27.0±1.5 °C, 45±9% relative humidity) and started cycling with water intake (ad libitum) on an ergocycle (CycleOps Pro 400, CycleOps, Madison, USA).…”

“…As a large variation in sweat [Na + ] has been reported among different body regions [15], sweat samples were collected using the absorbent patch technique [16] from the forehead, arm (tricep), and mid back. The procedure to collect sweat with absorbent patches was similar to the description in [12] with the following modifications. The compress size was 7.5×7.5 cm 2 and the duration of sweat collection was 30 mins corresponding to the main phase.…”

Person identification using deep neural networks on physiological biomarkers during exercise

“…There is increasing interest in wearable sweat devices that provide continuous real-time monitoring of sweat biomarkers [5], [6], because biomarkers such as sweat sodium concentration and sweat rate are linked to electrolyte loss and fluid balance [7]. In [8] the authors used sweat and physiological biomarkers for predicting the hydration status of an individual, and demonstrated the potential of providing a personalized hydration strategy using these biomarkers.…”