ABOUT
STEM II starts with a hiring process.
Everyone has to apply for two roles (CEO, CIO, CMO, or CTO) and submit
their applications to Dr. C. After Dr. C selects the CEOs, the CEOs begin
interviewing everyone else and hiring people for the other three roles.
Once teams are formed, they start brainstorming a project. My group wanted to
focus on women’s health, and we ended up deciding to build a bracelet that
predicts a menstrual cycle. This device can be useful for people with communication
difficulties, irregular periods, or conditions like endometriosis that make
periods more difficult. Many organizations (including the White House) are also
recommending that people avoid tracking their periods with apps because their data
can be sold. By tracking vital signs associated with the menstrual cycle (heart
rate, temperature, and skin conductivity) and storing that data locally, we can
predict most periods using a machine learning model. Keep reading to learn more
about our project!
FUTURE BUSINESS FRONTIERS (FBF)
My Team
CEO: PALAK YADAV
CIO: KEIRA REID
CMO: CLAIRE NEWCOM
CTO: CECILIA CARBONELL (ME)
DESIGN
Iontophoresis and Ion-Selective Electrodes
Skin conductivity varies with the phases of the menstrual cycle. The first approach that we considered used iontophoresis and ion-selective electrodes to track skin conductivity. While we didn’t use this approach in our final design (click here to skip to our final design), I still think it was really interesting, so I’ll briefly cover it here. I’d recommend reading Cystic Fibrosis and Glucose Monitoring using a Fully Integrated Wearable Platform by Emaminejad et al. in 2017 to learn more.
Iontophoresis is a method of transdermal drug delivery. It’s used a lot in eye medication because it’s not painful. We wanted to use it to generate sweat and then use ion-selective electrodes to analyze that sweat. Iontophoresis uses an electric current between two metal plates on the skin (see image to right). The plates are oppositely charged, and one plate has a similarly charged medication on it (for our project this would be a sweat stimulant). When the plates are turned on, the like charges of the plate and medication repel and the medication is attracted to the oppositely charged plate, traveling through the skin. As it travels through the skin, the medication is administered. At this phase, the sweat generation would be done and the sweat would be collected by either gauze or hydrogel.
After that, we would analyze the sweat using ion-selective electrodes. We would need one reference electrode for sodium, one for chloride, and a shared reference. The paper by Emaminejad et al. does a really good job of explaining how they made these electrodes, so I’d recommend reading that if you’re interested in a more in-depth explanation. Ion-selective electrodes are commonly used in laboratories to test for the concentration of a specific ion in a solution. Here, by testing for sodium and chloride, we can find skin conductance and compare it to previous data.
Heart Rate Monitor
To track heart rate, we used the MAX30102 heart rate sensor that Mr. Loven recommended. We had some difficulty with this heart rate monitor but after Mr. Loven met with us to troubleshoot, it started working properly.
Once the heart rate monitor was up and running, we got up and started running. This was an important part of our testing process, because when we ran up and down stairs we expected our heart rates to go up. We compared our heart rates measured by the sensor before and after running with the values from Sasha Nagireddy’s Apple Watch to ensure that our values were close.
Temperature
Body temperature was the second vital sign that we tracked. To track body temperature, we used the temperature sensor from Mr. Loven’s Creative Engineering elective. This sensor was a lot easier to work with, so after calibrating with a thermometer, we were able to perform some design studies and move on to the final vital sign: skin conductivity.
Electrodermal Activity (EDA)
Electrodermal activity, or skin conductivity, is a measure of how much someone’s skin conducts electricity. Because we didn’t have a sensor for this, we had to make our own electrode system.
The final circuit that we designed is very similar to a voltage divider (this article from Khan Academy is decent, but this is also a circuit you can make at home with very few supplies) because we wanted to measure resistance. We used a couple of 18-gauge wires taped to coins for electrodes (they get the job done) and used the analog input pins on the Arduino UNO to measure the input voltage. Then, we used the equation (R1*Vin)/
Prototype
This is the final prototype of our device. In
addition to the sensors I mentioned earlier, this prototype uses an Arduino
(for running the sensors) and a Raspberry Pi (for running our machine learning
model).
Poster