A wearable that listens to the skin. HydroSense records skin conductance, heart rate, oxygen, and temperature in real time — turning an unpredictable hyperhidrosis episode into data you can actually study.
Hyperhidrosis affects 4–5% of people, yet most never get a formal diagnosis. Episodes happen unpredictably, and without real data, treatment stays guesswork.
Skin conductance — the signal tied to sweat activity — isn't tracked continuously by any consumer device.
Heart rate, oxygen, and environment aren't captured together in one package.
Data isn't stored in a structure fit for machine learning.
Patients and doctors have no tool to spot patterns and triggers over time.
Built today on a microcontroller and breadboard, moving next to a custom PCB for size and reliability.
Shows recording status and live vitals.
Reads fingertip skin conductance — rises with sweat activity.
Tracks heart rate and blood oxygen via optical sensing.
One reads skin temp, one reads ambient air temp.
Reads humidity at the skin and in the surrounding air.
Central board reading every sensor and managing data flow.
Nothing below is locked in. The plan is to get a clean, reliable sensor stream working, then start shaping it into an actual machine learning pipeline — likely to get tweaked as 2026–2027 goes on.
Five channels logged with synced timestamps.
CSV / SQLite storage, ready for training.
Hoping to sort episode vs. no-episode from vitals + environment.
Eventually, a personal pattern report to share with a doctor.
This is what the project needs right now — not a fixed headcount. You don't need to be a professional or an upperclassman to apply — if you can offer something toward a role, or you're willing to learn and build up what it takes, you're more than welcome. Everyone starts somewhere.
Owns the embedded code, wires every sensor into one clean data stream, and designs the CAD housing.
Turns the breadboard into a custom PCB, picks components, and runs electrical tests.
Builds and iterates the fabricated case so the device is actually comfortable to wear.
Builds the pipeline that turns raw sensor data into a working ML model.
The four roles above aren't the whole project. There's real, unfilled need for planning, research, and ideas nobody's scoped out yet — if you've got something to bring, this is where it goes.
Most club projects get handed off to 30-plus people and lean on upperclassmen to run the show — so even if you land a role, you're doing one tiny slice of something you can barely call your own. HydroSense is built to stay a very small team on purpose. The work doesn't get diluted, so what you contribute is real, visible, and something you can genuinely contribute to and grow with.
Open brainstorming on direction and design, then finalize roles, source parts, and set up shared docs.
Get every sensor reading cleanly and confirm logging works.
Design the custom PCB and build v1 of the enclosure.
Assemble the PCB, fit the enclosure, and run full tests.
Collect real data, run a first ML pass, and present results.
Because the team stays small, the work doesn't get spread thin — real pieces of this project get delegated to you, not a shared credit on someone else's build. You'll walk away with actual project experience: something you built, can explain end-to-end, and put on a resume as your own.
Real practice turning early ideas into an actual working plan.
A working prototype on a custom PCB, in a fabricated enclosure.
Firmware reading all five sensor channels at once.
A structured dataset from real wear sessions.
A baseline ML model with documented accuracy.
A full technical report on hardware, software, and data.
A presentation delivered to EPHSO.
Hyperhidrosis is underdiagnosed and often dismissed as minor — yet it affects tens of millions of people. HydroSense sits at the intersection of embedded systems, mechanical design, electrical engineering, and machine learning: real enough to matter on a resume, focused enough to finish in one semester.
Apply Now →