Design and Evaluation of a Low-Cost Wearable Patch for Relative Hydration Monitoring in Remote Patient Care
Advisor: Allendoerfer
Internship Location: Cornell Engineering
Onsite Mentor: Carter Shen, BME Master's Student, Cornell Engineering
Dehydration is common and often underdetected in older adults, yet most remote patient monitoring (RPM) systems do not include hydration tracking. This project investigates the feasibility of a low-cost, non-invasive wearable patch designed to detect relative changes in hydration using skin-adjacent temperature and relative humidity sensors. The device is controlled by an Arduino-based microcontroller and emphasizes accessibility through inexpensive, open-source components.
A proof-of-concept study is conducted using bench-top models and self-testing to avoid regulatory constraints. Sensor outputs are evaluated under controlled conditions simulating varying hydration levels and compared against surrogate measures. Expected results include small, consistent decreases in skin temperature and larger decreases in relative humidity as hydration declines. While not intended for clinical diagnosis, this project demonstrates the potential for integrating simple hydration sensing into affordable RPM systems.
Project Blog
Week 1: Introduction & Literature Review
Welcome to my blog!
Dehydration is a common but is an often overlooked health issue that can result in fatigue, headaches, and more serious complications if not addressed. My project will focus on developing a low-cost and wearable hydration monitoring patch to detect early signs of dehydration using simple physiological signals like skin temperature and humidity. This device will be meant for everyday users, specifically older populations who might not recognize dehydration symptoms. By combining accessible sensors with a compact, user-friendly design, I aim to create a prototype of an affordable tool that can make hydration tracking more practical.
This week, I started off with my literature review, focusing on existing research and hydration-monitoring technologies, which will help me refine my prototype to ensure that it can address a gap in both research and the current market.
Over the past five years, wearable, patch-like hydration monitoring technologies have become more common and more advanced. Many research studies that I have read describe flexible smart patches that are made using printed electronics, microfluidic channels, and electrochemical sensors to measure various dehydration indicators like sodium, lactate, glucose, and pH in sweat. For example, one study created a fully printed patch to monitor pH and hydration levels in wounds to detect possible signs of infection (Iversen et al., 2021). Other researchers developed wearable patches, which were validated through lab calibrations and exercise trials done on humans and compared their results to gold-standard hydration measurements, like weight loss (Zhu et al., 2022). One study even combined a smart mouthguard to measure saliva osmolality with a chest patch that tracks ECG, skin temperature, and movement (Kim et al., 2024). To sum it up, most prior research focuses on high-performance sensing systems tested in controlled athletic environments.
Current products on the market also demonstrate a similar trend. Companies like Nix Biosensors, hDrop, MX3 Diagnostics, and the Gx Sweat Patch are mainly designed for athletes. These devices all typically measure sodium concentration in sweat and the rate at which athletes are sweating to help athletes optimize their performance. However, many of them are costly (over $100) and are either single-use patches or require disposable cartridges. The current ones on the market are functional and impressive, but they are built for high-performance athletes rather than the average person.
My project will take a different approach: instead of focusing on elite athletes or utilizing expensive electrochemical sensing systems, I will be designing a low-cost wearable patch for early dehydration detection in older adults. Rather than using sodium levels in sweat as an indication of dehydration, my device will combine multiple physiological signals to identify potential dehydration levels.
The materials that I plan to use are a DS18B20 temperature sensor, a DHT22 humidity sensor, and an Arduino Uno, to start off. The DS18B20 will measure skin temperature, as skin temperature will increase when losing fluids. The DHT22 will measure the microclimate humidity near the skin, as increased humidity near the skin’s surface would be an indication of fluid loss. So, instead of looking at just one metric, I will combine these two different signals to detect patterns that will indicate dehydration before severe symptoms occur.
Works Cited:
Iversen, M., Monisha, M., & Agarwala, S. (2021). Flexible, wearable and fully-printed smart patch for pH and hydration sensing in wounds. International Journal of Bioprinting, 8(1), 447. https://doi.org/10.18063/ijb.v8i1.447
Kim, K. R., Kang, T. W., Kim, H., Lee, Y. J., Lee, S. H., Yi, H., Kim, H. S., Kim, H., Min, J., Ready, J., Millard‐Stafford, M., & Yeo, W. (2024). All‐in‐One, wireless, Multi‐Sensor integrated Athlete Health Monitor for Real‐Time continuous detection of dehydration and physiological stress. Advanced Science, 11(33), e2403238. https://doi.org/10.1002/advs.202403238
Zhu, Y., Haghniaz, R., Hartel, M. C., Guan, S., Bahari, J., Li, Z., Baidya, A., Cao, K., Gao, X., Li, J., Wu, Z., Cheng, X., Li, B., Emaminejad, S., Weiss, P. S., & Khademhosseini, A. (2022). A breathable, Passive‐Cooling, Non‐Inflammatory, and biodegradable Aerogel electronic skin for wearable Physical‐Electrophysiological‐Chemical analysis. Advanced Materials, 35(10), e2209300. https://doi.org/10.1002/adma.202209300
Week 2: Putting Everything All Together
Welcome back to my blog! As I talked about in my previous blog, I have already ordered the materials and sensors that I need for my device. This week, I will be focusing on the assembly/design of the device. To start, I created a circuit schematic for my device on EasyEDA, which I have annotated and shown below:
This circuit schematic shows how the Arduino Uno (which is essentially the “brain” of the system) connects to the two sensors that I plan on using to collect data for the hydration-monitoring device. The DS18B20 sensor measures skin temperature, while the DHT22 sensor measures humidity and temperature from the surrounding environment. Both sensors will send their data to the Arduino, so the measurements can be processed and output neatly on the Arduino console. The resistors that are labeled as 4.7kΩ and 10kΩ are called “pull-up resistors,” and they help the sensors communicate clearly with the Arduino. A pull-up resistor helps keep the signal wire at a stable default voltage, so the signal wire doesn’t have random changes in its voltage when no data is being sent. The resistor keeps the signal wire at a high voltage until the sensor intentionally changes this voltage to send data, and when the sensor sends data, it will temporarily pull the voltage low, allowing the Arduino to detect this change. In other words, the pull-up resistors simply make sure that the sensor signals stay stable and readable so the Arduino can accurately collect data from the two sensors.
Then, because I ordered parts already, I have also started to build the device, connecting the parts together, using the circuit schematic as my guide. This week, I mainly focused on getting the humidity sensor (DHT22) and the skin temperature sensor (DS18B20) to work, which included connecting them to a breadboard and the Arduino Uno and coding the sensors in the Arduino Integrated Development Environment (Arduino IDE—a computer program that allows me to write C++ code and upload it to the Arduino). Since I am not very familiar with coding, especially C++, there was a learning process involved as well. I watched some YouTube videos to learn and get familiar with this new coding language. Here is the physical diagram of the device:
This is the completed, physical circuit diagram that was made based on the circuit schematic shown above. Everything was able to connect successfully, and I was able to code each sensor so that I could get the desired output for each (i.e., temperature, humidity levels). As you may notice in the physical diagram, though, I did not have a 10kΩ pull-up resistor for the DHT22 humidity sensor. This is because it was already built into the DHT22 humidity sensor that I have, so there was no need to include it again.
To sum everything up, I successfully connected and programmed both sensors! Next week, even though I will be on the Disneyland trip, I will work on coding the sensors so their signals can be paired to detect dehydration. I will also be researching testing methods, as that will be the next step in making my device. See you guys in the next one!
Week 3: Getting Ready for Testing!
Hey everyone, and welcome back to my blog!
This week, I focused on having the sensors in my device work together to potentially detect a dehydration risk, and I will do this through code. To make both sensors work in the same program, I used two Arduino libraries: DHT.h for the DHT22 sensor and DallasTemperature.h with OneWire.h for the DS18B20 sensor. In my code, the Arduino will read humidity and temperature from the DHT22 and will request the DS18B20’s temperature reading every cycle. These readings will then be printed to the serial monitor so I can see them in real time. I also added a simple decision system: if the DS18B20 temperature reading rises above a certain threshold and the humidity is low, the program will print a warning to indicate possible dehydration. If only one of these factors is outside the safe range, the program will print a mild risk message, and if both are within normal ranges, it will report normal conditions. I will need to do more research on what “normal conditions” for people are, but this code will allow the two sensors to work together instead of independently, which will help the device identify patterns associated with dehydration rather than just relying on one measurement. Currently, this is the output that the sensors give me:
Now, to test the device, I plan to use artificial skin substitutes, which I’ve researched in the past week. Many wearable biosensor papers describe using hydrogel-based materials to simulate human skin when testing sensors. Researchers have also compared synthetic materials, like silicone or polyurethane, with hydrogels that better mimic the water content, softness, and thermal properties of skin. Since my device relies on detecting temperature and environmental conditions near the skin surface, it is important that I test it on a surface that behaves similarly to real skin. Based on the literature, I chose to focus on gelatin and agar as testing materials.
Gelatin is derived from collagen, which is one of the main structural proteins in human skin. Because of this, gelatin hydrogels can replicate the elasticity, softness, and moisture retention of biological tissue. Agar, derived from seaweed, forms stable gels that can also hold large amounts of water and maintain their structure well during experiments. Studies on wearable biosensors frequently use gelatin or agar gels because they allow researchers to simulate skin hydration and thermal behavior in a controlled way. Both materials are also inexpensive, easy to prepare, and safe to use, which makes them practical for repeated testing.
Using gelatin and agar will allow me to create a realistic testing surface for my hydration-monitoring patch without needing to test directly on human skin in the early stages of development. By placing my sensors against these hydrogels and adjusting temperature and humidity conditions, I will be able to observe how the sensors respond to different simulated hydration environments. This will help me evaluate whether the device can reliably detect patterns associated with dehydration.
Next week, I will start creating gelatin/agar gels and testing my device. Stay tuned!
Week 4: First Bump in the Road...
Hey everyone! As the title of the blog post would suggest, I have encountered the first bump in the road: making gelatin hydrogels, or at least getting them to simulate different “hydration levels.” The process of making hydrogels was actually pretty easy: I mixed store-bought gelatin powder with water, warmed it up, stirred until the gelatin was dissolved, and then poured the solution into molds to cool and solidify. After letting them sit at room temperature, I was able to get stable, flexible gel samples that looked and felt similar to soft tissue:
Originally, I thought that I could simulate different hydration levels by simply changing the temperature of the gelatin. The idea was that heating the gel could make it “less hydrated,” and cooling it could make it “more hydrated,” since I could be using my temperature sensors as indicators of dehydration as well. However, when I tested this, it didn’t work the way I expected. While the temperature changes affected the surface, they didn’t change the actual water content inside the gel. The gelatin stayed at basically the same hydration level regardless of temperature when measuring with the temperature sensor. This is a big problem because my device is meant to detect dehydration, and I needed clear differences in hydration levels to test it properly.
So instead, I decided not to focus too much on the temperature of the gelatin, and instead focus on the amount of water present in the gelatin hydrogels by making multiple gelatin samples with different water-to-gelatin ratios, which essentially made some “more hydrated” than others. I haven’t gotten down to which different gelatin-to-water ratios I will be using yet, since most of what I was doing was experimental to see if I could get clear differences between the samples. But I will include my different ratios in the next blog. I also plan on leaving some samples exposed to air to allow for evaporation, which will gradually reduce the samples’ water content and simulate dehydration over time. This approach will work better since it will create real, measurable differences in hydration levels.
In addition to testing on hydrogels, I am also planning on doing some self-testing. Based on two research studies, the elbow region is one of the most reliable areas for detecting hydration changes because of its skin properties and consistent signal readings. I will place my sensors on my elbow and record data with my Arduino setup.
Even though my first idea didn’t work, this process helped me better understand both my materials and sensors. Instead of focusing too much on trying to simulate temperature readings, I now have another way to simulate dehydration as a whole. Combining this with self-testing, in the future, gives me a stronger foundation for improving my device. Next week, I plan on building a data recording system by writing code/creating a platform that logs sensor readings over time, so I can collect and analyze trends. I will also get down my gelatin-to-water ratios and use my device on them to start data collection. See you guys in the next one!
Week 5: More Bumps in the Road...
Hello and welcome back to my blog! Last week, if you may recall, I encountered my first obstacle. This week, I have unfortunately encountered my second obstacle…
As mentioned in my previous blog post, I planned on focusing more on detecting hydration levels through humidity. The idea was that as something becomes more “dehydrated,” the surrounding humidity would decrease, and my DHT22 sensor would be able to pick up on that change. However, after testing this, I realized that this approach didn’t work the way I expected. The humidity levels were not changing enough or fast enough to be useful. Even when I tried to create “drier” conditions, the humidity readings stayed relatively stable and never dropped close to my dehydration threshold of 40% humidity. Instead of showing clear differences, the values changed very slowly over time, making it difficult to use humidity alone as a reliable indicator of hydration.
To further test this, I tried using gelatin hydrogels with different water concentrations. I made one sample using a normal ratio of 1 tablespoon of gelatin to 1/4 cup of water, and another “dehydrated” version using 1.5 tablespoons of gelatin to 1/4 cup of water. My expectation was that the second sample would behave like a less hydrated surface and produce different humidity readings. However, when I tested both samples, there were no noticeable differences in the sensor output. The humidity readings remained almost identical, which showed that simply increasing the gelatin concentration does not significantly change the measurable humidity at the surface.
This was definitely frustrating because it meant that my original plan of detecting hydration through humidity alone wasn’t going to work as intended. It also showed me that hydration is more complex than I initially thought—it’s not just about how much water is inside a material, but also how that water interacts with the environment and evaporates over time.
For next week, I plan to adjust my approach by focusing on how humidity changes over time rather than instant measurements. Specifically, I want to try covering the gelatin with Saran Wrap with small holes poked into it to better simulate sweating and controlled evaporation. By doing this, I can create a more realistic system where moisture is slowly released, similar to how sweat leaves the skin. Instead of looking for immediate changes, I will measure humidity over several days (likely across the entire week) to see if gradual trends appear.
Hopefully, this new setup will allow me to capture meaningful changes in humidity that correlate more closely with hydration levels. Even though this week didn’t go as planned, it helped me better understand the limitations of my sensors and pushed me to think more critically about how to simulate real-world conditions.
Weeks 6-7: Some Results!
Hello everyone, and welcome back! Last week was Spring Break, hence, no post, so I combined the two weeks into this blog post. This week, I continued working on testing my hydration-monitoring prototype by improving my gelatin hydrogel setup. I made a new “normal” gelatin sample using a ratio of 1 tablespoon of gelatin to 1/4 cup of water, which serves as my baseline for comparison. To better simulate how moisture might be released from skin, I covered the gelatin with Saran Wrap and poked small holes into it, allowing limited evaporation while still keeping the system somewhat controlled. After preparing the sample, I placed my sensors into position and took an initial reading of both temperature and humidity.
Instead of just taking quick measurements, I let the gelatin sit and recorded daily readings over the course of a week. My goal was to see if gradual changes in hydration would show up in the data over time. However, the results were not what I expected. The temperature readings stayed around room temperature, although there was a slight drop during the week because it got colder over spring break. More importantly, the humidity levels remained almost constant throughout the entire testing period. This was likely because the Saran Wrap covering limited how much moisture could escape, preventing any noticeable decrease in humidity.
Even though the results didn’t show dehydration trends, I was still able to make an important improvement to my code. I updated the logic for the DHT22 humidity sensor by adding an upper threshold of 70% humidity. Originally, I only had a lower threshold (40%) to indicate dehydration risk, but I realized that I can look at causes of dehydration, one being excessive sweating (high humidity). After adding this upper threshold, my Arduino was able to classify conditions where humidity was too high as a “mild risk”, which makes the system more realistic and responsive to different scenarios.
For next week, I plan to modify my experiment by removing the Saran Wrap covering entirely. This should allow the gelatin to lose moisture more freely and hopefully cause the humidity readings to drop closer to my lower threshold. I also want to begin doing more self-testing to compare how real skin behaves versus the gelatin model. Overall, even though this week didn’t produce the results I was hoping for, it helped me refine both my testing setup and my code, bringing me one step closer to a more accurate hydration-monitoring system.
Week 8: Tweaking
Hi everyone, and welcome back! This week, I modified my experiment by removing the Saran Wrap covering from my gelatin hydrogel setup and running a new round of tests. Previously, the covering limited evaporation, and my humidity readings barely changed. By leaving the gelatin uncovered, I wanted to see if it would lose moisture more freely and create clearer changes that my DHT22 sensor could detect.
I used the same baseline gelatin mixture (1 tbsp gelatin to 1/4 cup water) and placed my sensors the same way as before to keep the setup consistent. I recorded daily humidity and temperature readings over the course of the week. This time, I did see a difference. The humidity readings slowly decreased over time, which suggests that the gelatin was actually losing water through evaporation. The change wasn’t dramatic, but it was more noticeable than before. The values still didn’t consistently drop all the way to my 40% lower threshold, but I did observe a clear downward trend, which is something I wasn’t able to capture in earlier trials. The temperature readings stayed close to room temperature, with some small fluctuations depending on the weather and time of day.
In addition to the gelatin tests, I also began doing self-testing to compare how real skin behaves versus the hydrogel model. I placed the sensors on my elbow, since research suggests that this area provides more consistent hydration-related readings. When I measured my own data, I noticed that the humidity and temperature values were more dynamic compared to the gelatin. Small changes in movement, environment, or even just sitting still caused slight fluctuations in the readings. However, overall, my values stayed within the “normal” range, which makes sense since I was properly hydrated during testing.
Comparing the two, I realized that while gelatin is useful for controlled, gradual testing, it doesn’t fully capture how real skin behaves in real time. The gelatin changes slowly, while human skin responds more quickly to environmental and physiological factors. This helped me better understand the limitations of my model and the importance of combining both types of testing.
Overall, removing the Saran Wrap was definitely an improvement. It allowed me to finally see measurable changes in humidity over time, even if they weren’t as large as I initially expected. Combined with the self-testing, this experiment gave me a much clearer picture of how my sensors behave in both controlled and real-world conditions. Moving forward, I plan to keep refining both setups so I can better detect meaningful hydration trends and improve the accuracy of my device.
Week 9: Data!!
This week, I continued improving my hydration-monitoring experiment by removing the Saran Wrap covering and running a full set of tests on both gelatin hydrogels and real skin. My goal was not only to observe changes over time, but also to directly compare how a simulated model behaves versus actual human skin.
For the first part of the experiment, I used my standard gelatin mixture (1 tbsp gelatin to 1/4 cup water) with no covering at all. I placed my sensors in the same position as before and recorded humidity and temperature readings over several days. The results are shown in my first graph (gelatin over time). From this data, I observed a slight downward trend in humidity, which suggests that the gelatin was slowly losing moisture through evaporation. However, the change was still gradual and not large enough to consistently reach my dehydration threshold of 40% humidity. The temperature stayed close to room temperature, with minor fluctuations depending on the environment. Overall, this graph showed that removing the covering did improve my setup, but the gelatin still changes slowly compared to what I expected.
The second graph shows data collected from my skin (on my elbow) before and after exercise. This was part of my self-testing, since research suggests the elbow is a good region for detecting hydration-related signals. Unlike the gelatin, the skin data showed much more noticeable variation. After exercise, I saw a clear increase in humidity, likely due to sweating, and a slight increase in temperature. This created a much more dynamic pattern compared to the gelatin graph, with greater differences for us to see.
Comparing the two graphs really helped me understand the differences between my model and real conditions. The gelatin provides a controlled, slow-changing system, which is useful for observing gradual trends. However, real skin behaves in a much more responsive and dynamic way, especially when influenced by factors like movement and sweating. This explains why my humidity-based detection wasn’t working well at first—the changes I was trying to detect simply weren’t happening quickly enough in the gelatin alone.
Even though the gelatin experiment didn’t fully reach my expected thresholds, it still provided valuable insight. It confirmed that my sensors can detect small changes over time, while the self-testing showed that the system is capable of picking up larger, real-world fluctuations. Together, these two graphs give me a more complete understanding of how my device performs.
Moving forward, I plan to keep refining both testing methods. The gelatin will help me control variables and simulate long-term dehydration, while self-testing will allow me to observe real physiological responses. In my next blog, I will include a graph of my skin over time, as it will hopefully allow me to reach my lower threshold level for humidity. Combining both approaches will be key to improving the accuracy of my hydration-monitoring patch and making it more reliable in real-world use.
Week 10: Data (contd.)
Hello and welcome back! This week, I continued working with my gelatin hydrogel model, but I also expanded my testing to include longer-term measurements on real skin, which ended up being one of the most important steps in my project so far. My goal was to see if my device could actually detect real dehydration and rehydration patterns, not just controlled changes in a lab setup.
First, I continued testing with my standard gelatin mixture (1 tbsp gelatin to 1/4 cup water) with no covering. I recorded humidity and temperature over multiple days to track how the gelatin changed over time. As expected, the gelatin showed a slow decrease in humidity, confirming that it was gradually losing moisture through evaporation. However, even over longer periods, the humidity still did not consistently drop to my lower threshold of 40%, which I use as an indicator for dehydration.
This graph shows that while there is a downward trend, the changes are very gradual and limited, which reinforces what I observed before: gelatin is useful for controlled testing, but it does not fully replicate the dynamic changes seen in real human skin.
To better understand how my device performs in real conditions, I conducted a longer self-testing experiment on my elbow. Over a period of time, I intentionally allowed myself to become slightly dehydrated (for example, by limiting water intake and monitoring over several hours). During this period, I recorded my sensor data continuously. For the first time, I observed that my skin surface humidity decreased significantly, eventually reaching my 40% lower threshold, which triggered the dehydration warning in my code. This was a major milestone because it showed that my system could actually detect dehydration in a real-world scenario.
After that, I rehydrated by drinking water and continued recording data. As expected, my skin surface humidity began to increase again, moving back into the normal range. This created a clear pattern of dehydration followed by recovery, which was much more noticeable than anything I saw with gelatin alone.
This graph shows a much more dynamic response compared to the gelatin graph. The drop in humidity during dehydration and the increase after rehydration are both clearly visible, which suggests that my sensors are capable of detecting meaningful physiological changes.
Comparing these two sets of data helped me understand an important point: gelatin is useful for slow, controlled testing, but real skin provides the variation needed to truly validate the device. The gelatin experiments helped confirm that my sensors can detect gradual changes, while the self-testing proved that the system can respond to real dehydration conditions.
Overall, this week was a big step forward. I was able to move beyond just testing materials and actually demonstrate that my device can detect real changes in hydration status over time. Moving forward, I plan to continue combining both methods (gelatin for controlled experiments and self-testing for real-world validation) to improve the accuracy and reliability of my hydration-monitoring patch.
Week 11: Refine, Refine, Refine!
Hello and welcome back!! I can't believe we've reached the end of my project, but I still have one (maybe two) more blog post to give. This week, I continued testing my hydration-monitoring prototype on my own skin to see whether the results I observed previously were actually reproducible. In earlier trials, I noticed that my humidity readings decreased when I became dehydrated and increased again after rehydrating. While this was exciting, I wanted to make sure it wasn’t just a one-time occurrence or caused by random environmental factors. To do this, I repeated the experiment multiple times over several days while keeping the setup as consistent as possible.
For each trial, I placed the sensors on my elbow, since research papers I previously reviewed suggested that the elbow region provides more stable hydration-related readings due to its skin properties. I measured both humidity and temperature over time while going through normal daily activities. In some trials, I intentionally waited longer before drinking water or exercised beforehand to encourage mild dehydration, while in others I tested after already being hydrated. Throughout the experiments, I recorded the sensor readings and compared how they changed over time.
One thing I noticed was that the overall trends were actually fairly consistent. When I was more hydrated, the skin surface humidity readings tended to stay higher and more stable. When I became dehydrated or exercised without immediately rehydrating, the humidity values gradually decreased and sometimes approached my lower threshold of 40%. After drinking water, the readings usually began to increase again. Even though the exact values varied slightly from day to day, the general pattern stayed similar across trials, which suggests that the device may actually be detecting meaningful hydration-related changes rather than random noise.
At the same time, I also observed how sensitive the readings were to external conditions. Things like room temperature, weather, movement, and even how tightly the sensor rested against my skin could slightly affect the values. This reinforced something I had already started realizing earlier in the project: hydration detection is probably not going to depend on one exact number alone. Instead, it makes more sense to focus on patterns and trends over time. For example, a gradual drop in humidity combined with increased skin temperature may be more useful than just checking whether one value crosses a threshold.
Another important takeaway from this week was that real skin behaves very differently from my gelatin hydrogel models. The gelatin changes slowly and predictably, while human skin is much more dynamic. Sweat production, movement, environmental exposure, and even small physiological changes can all influence the readings. However, this also makes self-testing much more valuable because it allows me to see how the device performs under real-world conditions instead of only controlled laboratory-like setups.
Since this is the second-to-last blog post of the project, I’ve also been reflecting on how much the project evolved from the original idea. Although I was able to create a working proof-of-concept prototype and observe hydration-related trends, the device still requires significant refinement before it could become a truly reliable hydration-monitoring system. Future work could include adding more advanced sensors, improving the calibration process for different users and environments, collecting larger datasets, and using machine learning to better identify dehydration patterns. The physical design of the patch could also be improved to make it smaller, more comfortable, and more wearable for long-term use. Additionally, testing with more people and under more controlled conditions would help validate the accuracy of the system.
While my final blog post will mainly be a reflection on the overall experience rather than introducing new advancements, this week’s testing helped solidify many of the conclusions I reached throughout the project. Overall, this experience taught me that engineering research is rarely a straight path. Many of my original ideas did not work the way I expected, but each obstacle ultimately helped me better understand both the technology and the problem I was trying to solve.
Post 11: Concluding Thoughts
As this project comes to an end, I’ve spent a lot of time reflecting on everything I learned throughout the process of designing and testing my hydration-monitoring patch. When I first started this project, I honestly thought the biggest challenge would simply be building the device itself. However, I quickly realized that the real difficulty was figuring out how to accurately measure something as complex and dynamic as hydration in a low-cost and noninvasive way. What started as a simple idea gradually evolved into a much larger exploration of wearable biosensors, biomedical engineering, physiology, and experimental design.
One of the biggest things I learned from this project is that engineering and research are almost never linear processes. Many of my original ideas did not work the way I expected. For example, I initially believed that humidity changes would clearly and quickly reflect dehydration, but through experimentation I learned that humidity often changes much more slowly than expected and is heavily influenced by environmental conditions. I also learned that creating a good experimental model is difficult. My gelatin hydrogel tests were useful for creating controlled conditions, but they behaved very differently from real skin. At first this felt frustrating, but over time I realized that these “failures” were actually some of the most important learning experiences in the project because they forced me to rethink my assumptions and improve my methods.
I also gained a much stronger appreciation for how much testing and refinement real biomedical devices require before they become practical for everyday use. Through my self-testing experiments, I was able to observe reproducible trends where skin humidity decreased during dehydration and increased after rehydration. Even though the system is still only a proof of concept, seeing the sensors respond to real physiological changes was one of the most rewarding moments of the entire project. It showed me that the core idea behind the device has potential, even if it still needs significant improvement.
If I were to continue this project in the future, there are many things I would want to refine. I would like to incorporate more advanced sensors, improve calibration methods for different users and environments, and collect larger amounts of data to better validate the system. I also think machine learning could eventually help identify more accurate dehydration patterns by analyzing multiple signals at once instead of relying on simple thresholds. Additionally, the physical patch itself could become smaller, more comfortable, and more wearable for long-term use.
Overall, this project taught me much more than just how to build a circuit or write Arduino code. It taught me how to troubleshoot problems, adapt when experiments fail, analyze research literature, and think critically about how biomedical technology can impact real people. Most importantly, it showed me how challenging (but also rewarding) the engineering design process can be. Even though this project is ending, I feel like it has given me a much stronger foundation for future work in biomedical engineering and wearable health technology.