Volatile Organic Compound Sensor
For Use In Future Glucose Monitoring
The Team
Our team consists of three senior electrical engineering students: Anthony Arjona Pech, Jonathan Cervantes, and Jessica Mellor.
To find information on how to contact us for any questions or inquiries, please visit the contact us section.
About Us
The three of us are all students at SSU studying Electrical Engineering. Anthony and Jonathan are transfer students from Santa Rosa Junior College, while Jessica has spent her entire undergraduate experience here at SSU! When we're not working on this project we all enjoy leisure activities including but not limited to playing video games, social distancing, and staying healthy! All of us are passionate about Electrical Engineering and are excited to almost be out the door, and into the work force as engineers. Anthony's favorite aspect of EE is working with the sophisticated testing equipment, while Jonathan's is learning about how our technologic world works, for Jessica, there's too many aspects to choose a favorite from. Overall, we're excited to be working on this project and are grateful for our two advisors.
Our Advisors
Additionally, we were assisted by our advisors: Mr. Mark Thoren, an engineer at Analog Devices, and Dr. Sudhir Shrestha, a professor here at Sonoma State University who also doubled as our customer.
Mr. Mark Thoren (Industry Advisor)
Mark.Thoren@analog.com
Dr. Sudhir Shrestha (Faculty Advisor)
sudhir.shrestha@sonoma.edu
Problem Statement
The problem we’re investigating addresses the need of researchers who are developing noninvasive health monitoring solutions by using volatile organic compounds (VOCs) found in a person’s breath. They are trying to test their trained algorithms with patients. They feel extremely frustrated because current methods of testing their algorithms with large numbers of patients is time consuming and inefficient.
Value Proposition
Our VOC Profiler helps researchers who want to develop breath VOC based health monitoring solutions by reducing the time and cost needed to test their algorithms with large numbers of patients unlike lab-based tests which require bringing patients to the lab and the use of expensive equipment.
Proposed Solution
Our solution to address the issues associated with current VOC analysis methods is a small handheld device capable of accepting a trained ML algorithm, displaying results on a user friendly GUI, and saving data on an external memory card. All enclosed in a 3D printed case. The device consists of an STM32 MCU for running ML algorithms, an LCD display for our GUI, three sockets for holding VOC sensors, a rechargeable battery, and a micro-USB port for programming the device and charging the battery. The GUI was designed with ease of use in mind, so no special training is required to begin conducting tests, this also means the GUI was designed with readability and color contrasts in mind. For the purpose of our demonstration, we used three Figaro metal oxide semiconductors (MOS) sensors, our algorithm, Profilemaster, trained and tested using data collected with samples of Equate hand sanitizer and Germ-X hand sanitizer, and implemented the algorithm on the device and conducted a real-time analysis with the hand sanitizers. We acquired hand sanitizer VOC data by placing the sensors in a mason jar with a specific brand of hand sanitizer, recording the resistance values from the sensors, and using this data to train a Keras based machine learning algorithm. Once we had collected and trained the algorithm, we then imported it to the STM32, and tested to see if it was able to distinguish between the hand sanitizers when a test is initiated via the UI, the results would also be shown on the UI, and be saved onto an SD card, which it was able to do. The algorithm we developed is intended to be used as a demo for our product showing that our device is capable of profiling VOCs, and overall, is a suitable replacement that addresses the issues with current methods with regards to efficiency, expense, and mobility. Researchers are free to implement their own algorithms, to best suit their needs, and have patients test their samples outside of a lab environment without the concern of transporting the samples or data elsewhere for analysis.
