1. Introduction
Health monitoring technologies have drawn significant attention from consumers and the scientific community in recent years [
1,
2,
3]. These include heart rate, calorie count and daily exercise tracking, currently available in wearable fitness devices. Recently, hydration is emerging as a health indicator that requires non-invasive monitoring [
3]. Hydration monitoring benefits athletes [
4], active people, workers in hot environments, military personnel and older adults [
5]. Hydration monitoring has the potential to improve personal health and contribute to health care cost reduction.
Dehydration can be defined as 1% or greater loss of body mass due to water loss [
6,
7]. Dehydration can result in impaired cognitive function, reduced physical performance, headaches and fatigue symptoms [
6]. If dehydration becomes severe (loss of 8% of body weight), it may become fatal as reported in [
8]. Dehydration can be classified in three major categories: hypertonic, hypotonic and isotonic, based on changes in water and sodium levels [
9]. Hypertonic describes states where tissue loses more water than sodium, resulting in higher salt concentration [
9,
10]. This can be caused by inadequate fluid intake, sweating and vomiting [
7]. If sodium is lost at a higher rate than water with the use of medications [
6,
7] or a genetic disorder [
11], the body may experience hypotonic dehydration [
12]. If the body lacks both water and sodium, isotonic dehydration occurs [
7,
9]. This loss is caused through perspiration, urine or diarrhea [
7]. Dehydration can be monitored through weight changes [
7], blood pressure [
7], skin (stratum corneum) [
9,
13], saliva [
14], urine and blood tests, as well as analysis of sweat [
15].
Dehydrated persons produce sufficient sweat to be assessed. Sweat of dehydrated people, including athletes, was monitored after extensive exercise or exposure to excess heat in [
15,
16,
17,
18,
19,
20,
21,
22]. Quantitatively, 15–25% lower sweat rates are observed in a group of older subjects (age 52–71 years) compared to a younger group (age 20–30 years) subjected to similar environmental conditions and exercise [
23,
24,
25,
26,
27]. Specifically, the average sweat rate for the younger group is reported as 0.4–1.2 μL/min/cm
2 compared to 0.3–0.58 μL/min/cm
2 for the older subjects with and without fluid replacement during exercise or subjected to heat. An average rate of 1 μL/min/cm
2 of local sweat on the forearm would result in 0.4 L of estimated sweat loss after 20 min of exercise for an adult with weight of 70 kg, height 1.7 m and age 25 years [
28,
29]. For older age groups (above 80 years), measurement of axillary sweating or moisture is recommended in [
30,
31] to assess dehydration. However the exact amount of absorbed axillary sweating was not reported for older patients.
Human sweat carries rich physiological biomarkers that make it an attractive fluid for non-invasive hydration monitoring. It includes electrolytes (sodium, chloride, potassium, magnesium, and calcium), metabolites, proteins and amino acids [
15,
16,
17,
18,
19,
32]. Beside traditional laboratory practices [
7,
15], several new technologies are reported in [
16,
17,
19,
20,
21,
22,
33] to analyze sweat biomarkers for non-invasive monitoring. This includes non-selective technologies to monitor overall sweat properties, such as measuring sweat pH levels using bio-textiles sensors [
16,
17] or conductivity [
19]. Other selective technologies record potential difference or impedance across sodium (Na
+) selective electrodes (ISE) [
20,
33], or multi-biomarker sensing selective patches [
21,
22]. Most of the reported technologies require either a conditioning period, large sample size, additional electronic circuits or special sweat sampling mechanisms. A common theme in the different sensing works reported in [
9,
15,
17,
19,
20,
21,
22,
33] is that, among different sweat electrolytes, sodium (Na
+) or sodium chloride (NaCl) results in the largest recorded variations with changing hydration states.
Recently, several studies reported assessment of human dehydration using microwave signals [
9,
34,
35,
36]. Body fluids with different electrolytic concentrations can result in different losses and hence absorption of electromagnetic waves [
9]. The detected reflected or transmitted signals carry information on electrolytic concentrations, motivating development of microwave-based sensors for non-invasive hydration monitoring. Most of the reported studies used reflection or transmission measurements to assess water content in blood plasma [
34,
35,
36]. On the other hand, none of these studies [
34,
35,
36] analyzed sweat content or used microwaves to track electrolytes concentrations directly. However, Brendtke et al. developed a patch antenna in the range of 7.0–9.5 GHz for monitoring hydration status through assessing water content in skin tissues [
9]. This work focuses on developing artificial skin equivalents incubated in a range of test liquids. Among these tested liquids, NaCl solutions with concentrations from 0% to 20% were used to alter skin hydration levels. The changes in Return Loss (RL) and frequency of the local minimum (ƒ
min) were recorded for different skin equivalents, demonstrating RL magnitude changes of 3.4 dB and ƒ
min changes of 90 MHz with changing NaCl concentrations from 0% to 20%. This sensitivity is recorded when placing samples in a bioreactor chamber incorporating the antenna, and may be altered by changing measurement procedures, such as integration into a wearable module or when it is placed in direct contact on real skin. Moreover, the work in [
9] does not include testing with other major electrolytes in body fluids.
This work provides a proof-of-concept for a novel approach to using microwave signals to detect hydration states using sweat. The microwave-based sensor is composed of an ultra-light weight conformal antenna printed on a paper substrate and operating in the range 2–4 GHz. The proposed sensor structure can be placed directly on the skin and has an overall height of 230 μm. The performance of the antenna changes when liquids are absorbed by the substrate, and we demonstrate that these changes are linked to the sodium chloride concentrations and amount of solution absorbed. This is realized by measuring location and magnitude of the resonant frequency of the reflected microwave signal. After presenting the antenna design in
Section 2.1, the approach to sensing is discussed in
Section 2.2. Experimental results and case studies, including NaCl solutions and artificial sweat, are presented in
Section 3, and concluding discussions are provided in
Section 4.
4. Discussion
This paper presents a demonstration of how microwave signals can be used to distinguish between diluted and dehydrated sweat. The conformal, low cost, disposable paper-based sensor uses reflection-based measurements to detect different hydration states using sweat. The sensing approach is based on the absorption of sweat/NaCl solution by the cellulose filter paper substrate. Different hydration states may result in sweat with different dielectric properties and conductivities due to changes in concentration of sodium chloride (and other electrolytes). These differences in sodium concentration are detected as different frequency shifts and magnitude levels in reflection coefficient measurements. Thus, different hydration states may be detected. Moreover, the sensor succeeds in detecting the amount of solution applied.
Several experiments have been conducted with the given sensor and solutions at different NaCl concentrations (0–10% or 0–1.7 mol/L) and quantities to validate sensor operation. Through the measurements, detected NaCl concentration ranges are classified as 0–0.5%, 0.5–2% and >2%. These ranges represent hyponatremia, moderate hypernatremia and severe hypernatremia hydration status. Microwave measurements show the success of the proposed sensor to distinguish between different concentrations of sodium chloride in the range from 0 to 1.7 mol/L (0–10%) with RL magnitude span of 8–18 dB and ƒmin changes of 430 MHz. The results show maximum variation of ±0.1 GHz at the first resonance and ±0.18 GHz at the second resonance for all concentrations. Moreover, the amount of solution is detected with accuracy of ±0.05 mL.
The proposed sensor was also tested with artificial sweat at different electrolyte concentrations and at different frequency bands. Microwave measurements have demonstrated the capability of the sensor to differentiate between diluted and dehydrated sweat with sodium chloride concentrations in the range from 0.01–0.2 mol/L (0.05–1%). The corresponding changes in RL magnitude span 2–10 dB and ƒmin changes span 170 MHz. The results show maximum variation of ±0.08 GHz for diluted sweat and ±0.1 GHz for dehydrated sweat. Moreover, the amount of absorbed artificial sweat is also detected with accuracy of ±0.05 mL. More significant changes in dielectric properties of artificial sweat are noted and measured with varying NaCl concentrations in comparison to property changes obtained when varying concentrations of other major sweat electrolytes (e.g. potassium chloride (KCl), urea and lactic acid). Therefore, the most significant dielectric property changes in sweat are likely to be related to changes in NaCl.
Compared to previously reported microwave-based hydration sensors [
9,
34,
35,
36], the proposed sensor demonstrates for the first time the capability of microwave signals to track sweat electrolytes for hydration monitoring. The proposed work compares the response and sensitivity of the proposed sensor as well as measurement variations at different operating frequencies (2 and 3.5 GHz) and at different tested solutions. The given sensor shows almost four times higher sensitivity (10 dB at different artificial sweat concentrations or 18 dB at different concentrations of NaCl solutions) than other reported microwave hydration sensors (3.4 dB at different concentrations of NaCl solutions [
9]). Such enhanced sensitivity is expected to provide detection accuracy and sensitivity when testing patients with low sweat secretion.
The proposed sensor has an overall size of 50 mm × 60 mm, which is appropriate for sampling sweat at different locations (e.g. armpit, forehead, forearm or chest). Operating at 3.5 GHz allows designing the sensor with adequate slot size incorporated for collecting sweat/salt solution. The cellulose filter paper substrate is highly conformal with overall height of 230 μm, low loss and environmentally friendly compared to existing flexible substrates. Microwave measurements show that, for the given sensor, no changes occur when absorbing more than 0.4–0.5 mL (8–10 drops), corresponding to covering the whole substrate with solution. The given sensor does not require conditioning time and provides instantaneous response. Detection time per measurement, including processing 2000 points, was less than 2 min. With the given substrate, using copper foil and no fabrication complexities, the overall sensor cost is estimated at 15–20 cents. This price point can place the proposed sensor in the disposable category.
5. Conclusion
This paper demonstrates for the first time the capability of microwave signals to track sweat electrolytes for hydration monitoring. The conformal, low cost, disposable, non-wearable, paper-based sensor uses reflection-based measurements to differentiate between diluted and dehydrated sweat. It also demonstrates high sensitivity to NaCl in the presence of other major sweat constituents: potassium chloride (KCl), urea and lactic acid. Thus, the NaCl concentration level in sweat mixtures could be also estimated. The given sensor shows almost 4 times higher sensitivity compared to other reported microwave hydration sensors. Such enhanced sensitivity is expected to provide detection accuracy and sensitivity when testing patients with low sweat secretion.
The proposed design can be directly placed on skin in the current form, absorbing sweat, without changing measurement procedures. Thus, the sensing decision is not influenced by contact with the skin or tissues and does not require proximity to the human body. The proposed sensor has the advantages of consistent performance, high sensitivity, simple sweat sampling, no fabrication complexities, as well as low price point which is appealing for integration in clinical practices.