Advanced Energy-Efficient System for Precision Electrodermal Activity Monitoring in Stress Detection

Ruoyu Zhang, Ruijie Fang, Elahe Hosseini, Chongzhou Fang, Ning Miao, Houman Homayoun

Introduction

Stress has emerged as a significant global health challenge, necessitating reliable long-term monitoring solutions. Electrodermal Activity (EDA), which quantifies changes in skin conductance resulting from sympathetic nervous system-induced sweat gland activity, serves as a robust physiological marker for stress. While the fingers and toes offer the strongest EDA signals, they are unsuitable for continuous daily monitoring. Transitioning sensors to the wrist enhances user comfort but introduces a "placement paradox": the signals are subtler, requiring a vastly higher dynamic range and resolution than traditional sensors provide. Current wearable solutions often compromise between accuracy and battery life. This paper presents a novel, ultra-low-power wearable system developed at UC Davis that utilizes an adaptive gain mechanism to achieve high-precision EDA monitoring in a compact form factor.

System Architecture and Adaptive Gain Selection

The proposed system is built around the nRF52832 microcontroller, featuring an ARM Cortex-M4 processor and integrated Bluetooth Low Energy (BLE). The core innovation lies in the Analog Front End (AFE), which implements an Adaptive Gain Selection Mechanism.

To accommodate the wide dynamic range of skin resistance (10 MΩ to 25 kΩ), the AFE employs two multiplexers (MUXs) to dynamically select from a ladder of precision resistors. This configuration allows the system to adjust its gain in real-time based on the input signal level, preventing amplifier saturation while maintaining high sensitivity across the entire measurement range.

System architecture showing the power management unit, MCU, and the Analog Front End (AFE) with adaptive gain selection.

Power efficiency is achieved through aggressive management strategies. The firmware utilizes an "enable" function to activate the MUXs and operational amplifiers only during the 125ms sampling windows, corresponding to an 8Hz sampling rate. Furthermore, high-efficiency DC-DC converters step down the battery voltage to 1.8V to minimize power consumption.

Experimental Evaluation

The system's performance was validated through three distinct phases:

  1. Simulation: EDA data was generated using a specialized Python library to verify the adaptive gain logic across theoretical conductivity ranges.
  2. On-Board Accuracy: Testing with standard metal film resistors (27kΩ to 10MΩ) demonstrated an absolute error rate of less than 1% across all resistance levels.
  3. Real-World Benchmarking: The prototype was compared against the Empatica E4, a medical-grade wearable, during a 10-minute stress protocol involving deep breathing and speaking tasks. The system achieved a 0.73 Pearson correlation with the gold-standard device, successfully capturing identical peaks and trends.

Discussion of Key Insights

The primary advantage of the adaptive gain approach is the significant improvement in measurement resolution. By dynamically adjusting the gain, the system maintains a resolution of 31 Ω/bit at low resistances and 4.6 kΩ/bit at high resistances. In contrast, fixed-gain systems suffer from rapidly degrading sensitivity as skin resistance increases, often losing the ability to detect subtle phasic changes.

Comparison of resolution and output voltage between circuits with and without the adaptive gain mechanism.

A critical finding in power management was the "ignition effect." The researchers observed that frequent power-cycling of the sensing circuitry consumed more energy (3.6mA) than continuous operation due to the startup current of the converters. This led to a refined strategy that balances sampling frequency with power-up overhead, resulting in a total system current draw of just 700µA. This efficiency enables over 30 hours of continuous data acquisition on a small 30mAh battery.

Conclusion and Future Directions

This work demonstrates that sophisticated analog design can enable clinical-grade EDA monitoring in a consumer-ready wearable format. Future iterations will focus on replacing ±1% tolerance resistors with high-precision thin-film components (±0.01%) to eliminate individual calibration requirements. Additionally, the team aims to optimize the BLE communication protocol for better stability and integrate the electronics into flexible substrates for enhanced user comfort in "smart clothing" applications.

Summary updated

BibTeX
@inproceedings{zhang2024advanced,
  author = {Ruoyu Zhang and Ruijie Fang and Elahe Hosseini and Chongzhou Fang and Ning Miao and Houman Homayoun},
  title = {{Advanced Energy-Efficient System for Precision Electrodermal Activity Monitoring in Stress Detection}},
  booktitle = {BSN},
  year = {2024},
  doi = {10.1109/bsn63547.2024.10780551},
}
Figure