Problem & Motivation
With hypertension affecting 48.1% of American adults and contributing to over 685,000 deaths in 2022, blood pressure measurement is a critical healthcare tool. While automated blood pressure monitors are widely available, understanding their underlying mechanics and signal processing principles requires building one from the ground up. Our project focused on developing an automated blood pressure monitoring system that implements both traditional oscillometric measurement techniques and the signal processing necessary to extract meaningful cardiovascular information from pressure oscillations.
By designing our own system, we gained deep insights into how blood pressure signals are acquired and processed - from the physical principles of pressure transduction to the nuances of filtering and analyzing oscillometric waveforms. This hands-on approach allowed us to understand the relationship between cuff pressure changes and the physiological parameters they represent, while also tackling practical challenges in analog signal conditioning and digital processing.
Circuit system, more details available in presentation linked here.
Technical Implementation
Our system's design centered around precise signal acquisition and processing. The MPX5100GP pressure sensor serves as the core sensing element, using a piezoresistive transducer to convert cuff pressure variations into analog voltage signals. These signals contain two key components: the low-frequency cuff pressure and the higher-frequency oscillations from arterial pulses. To separate and analyze these components, we implemented a dual-path signal processing chain.
The analog front-end features carefully designed filters: a low-pass filter (cutoff 1.06 Hz) for measuring absolute cuff pressure, and a band-pass filter (10.6-40 Hz) for isolating the subtle arterial pulse oscillations. The system's mechanical control uses an Arduino microcontroller coordinating two pumps and a solenoid valve through MOSFET switches, allowing both automated measurement sequences and manual control through a user interface with an OLED display.
Results & Validation
We validated our device against both gold standard and consumer-grade blood pressure monitors across various physiological conditions. Testing with eight subjects under different scenarios (exercise, held breath, and box breathing) demonstrated the system's reliability. During exercise testing, our device showed strong correlation with gold standard measurements, with mean differences of +5.34/+2.31 mmHg for systolic/diastolic pressures. Similar accuracy was observed during breath holding tests (+4.85/+2.14 mmHg difference) and box breathing exercises (+2.19/+3.02 mmHg difference), with the latter showing no statistically significant difference from commercial devices.
While the current prototype demonstrates promising results, we identified several areas for future improvement. These include replacing the Arduino with an nRF Connect Nordic board for better voltage handling, implementing automatic calibration for varying cuff conditions, and improving the physical design for better air seal integrity. These enhancements (currently ongoing) will further improve the device's accuracy and usability in real-world settings.