Towards wearable, non-invasive and real-time hydration tracking

Overview of the 1st prototype

HydroTrack is the first prototype I made by integrating a low-cost optical sensor into an open-source programmable watch and turning it into a hydration monitor.

  • Hardware – SparkFun AS7265x Triad spectroscopy sensor embedded in a DSTIKE Watch V4 (ESP32).
  • Experiments – Calibration with lab-grade spectrometer (Hach DR3900), electrolyte-solution benchmarking, skin-hydration tests during fasting & exercise.
  • Deployment – TinyML model runs fully on-device; LEDs indicate hydration state (green = hydrated, red = dehydrated).

Methods Summary

Hardware Integration

The AS7265x Triad sensor captures 18 wavelengths (410–940 nm).
Signals are read via I²C on an ESP32 microcontroller and processed on the watch itself.

Machine-Learning Pipeline

  • Preprocessing: noise filtering + Eulerian Video Magnification to amplify subtle optical variations.
  • Models: Random Forest and XGBoost for 3 hydration classes: hydrated, mid, dehydrated.
  • Deployment: converted to TinyML header files and executed locally on the watch.

Experiments

  1. Electrolyte calibration with reference solutions: To validate the sensor, we tested several hydration drinks and lab-made sodium and potassium solutions at different concentrations. Their light absorbance was first measured with a professional lab spectrometer and then compared with readings from our wearable optical sensor to assess accuracy
  2. Tracking hydration change during workout: We measured changes in skin absorbance before, during, and after workout sessions to observe how dehydration affects the body’s optical response.

Beyond the Prototype

After developing the first HydroTrack prototype, I conducted an extensive review of hydration-monitoring technologies. This helped benchmark HydroTrack against all existing methods and identify the gaps in real-world hydration sensing and what are the most promising aprroaches.

Next-gen Hydration Research Study:

I am currently leading an IRB-approved study at UMass Lowell to validate the next-generation HydroTrack device during treadmill exercise.

What the study involves

  • 1–2 hour laboratory session
  • 30–60 minutes of treadmill running at a moderate pace
  • Wearing the wristband to measure hydration levels
  • Providing saliva samples pre- and post-workout stages

Call for Participants

Hydration Study Flyer

Publications

  • Belabbaci N.A., Alam M.A.U., HydroTrack: Spectroscopic Analysis Prototype Enabling Real-Time Hydration Monitoring in Wearables, IEEE EMBC 2024.
    → Read on IEEE Xplore

  • Belabbaci N.A., Wearable Hydration Monitoring Technologies: A Comprehensive Review of 100+ Studies, JMIR mHealth and uHealth 2025; 13:e60569.
    → Read on JMIR

  • Master’s Thesis – A Prototype Smartwatch Utilizing Spectroscopy for Real-Time Monitoring of Hydration Status, UMass Lowell 2024.
    → Download PDF