Abstract
Cardiovascular diseases claim over 10 million lives annually, highlighting the critical need for long-term monitoring and early detection of cardiac abnormalities. Existing techniques like electrocardiograms (ECG) and Holter are accurate but suffer from discomfort caused by body-attached electrodes. While wearable devices using photoplethysmography offer more convenience, they sacrifice accuracy and are susceptible to environmental interference. Here we present a radio frequency (RF)-based (60 to 64 GHz) sensing system that monitors long-term heart rate variability (HRV) with clinical-grade accuracy. Our system successfully overcomes the orders-larger interference from respiration motion in far-field conditions without any model training. By identifying previously undiscovered frequency ranges (beyond 10-order heartbeat harmonics) where heartbeat information predominates over other motions, we generate prominent heartbeat patterns with harmonics typically considered detrimental. Extensive evaluations, including a large-scale outpatient setting involving 6,222 eligible participants and a long-term daily life scenario, where sleep data was collected over 5 separate random nights over two months and a continuous 21-night period, demonstrate that our system can monitor HRV and identify abnormalities with comparable performance to clinical-grade ECG-based systems. This RF-based HRV sensing system has the potential to support active self-assessment and revolutionize medical prevention with long-term and precise health monitoring.
Introduction
Cardiovascular diseases (CVDs) remain the leading cause of global mortality. Early detection through long-term cardiac monitoring is essential for improving patient outcomes. However, the current gold standards—ECG and Holter monitors—require skin-attached electrodes that are uncomfortable for prolonged use. While consumer wearables like smartwatches utilize photoplethysmography (PPG), they often struggle with accuracy and are sensitive to factors like ambient light and skin tone.
Contactless radio frequency (RF) sensing using millimeter-wave (mmWave) radar has emerged as a promising alternative. However, it faces a significant physical challenge: respiration interference. In a far-field setting, chest movements from breathing (centimeter-scale) are orders of magnitude larger than those from heartbeats (sub-millimeter). This disparity typically drowns out cardiac signals, making precise heart rate variability (HRV) measurement difficult without invasive setups or person-specific machine learning training.
Methodology: Leveraging High-Order Harmonics
The researchers developed RF-HRV, a system utilizing a 60–64 GHz mmWave radar to achieve clinical-grade accuracy without model training. The core innovation lies in a physics-based signal processing pipeline that treats harmonics as a signal source rather than noise.
The system employs three primary techniques:
- High-Order Harmonic Leveraging: The researchers discovered that respiratory harmonics decay much faster than heartbeat harmonics. By focusing on frequency ranges beyond the 10th order, the system isolates a "cleaner" heartbeat signal where respiratory interference is naturally attenuated.
- Variational Mode Decomposition (VMD): Raw RF signals are decomposed into Intrinsic Mode Functions (IMFs). High-frequency IMFs are selected to capture the rich vibrations of the heartbeat.
- The Beat Frequency Effect: To reconstruct the heart rate, the system utilizes the "beat" pattern generated by superimposing adjacent harmonics. The frequency of this beat corresponds to the fundamental heart rate, allowing for precise Inter-Beat Interval (IBI) estimation.
Experimental Validation
The RF-HRV system was validated through two extensive studies to ensure its robustness in both clinical and real-world environments.
Large-Scale Outpatient Study
The system was tested on 6,222 eligible participants (ages 3 to 94) at a tertiary hospital, using 12-lead ECGs as the ground truth. RF-HRV achieved a median real-time IBI error of 26.1 ms, representing a 10x improvement over previous RF-based benchmarks. It successfully identified Bradycardia and Tachycardia with over 92% accuracy.
Long-Term Daily Life Monitoring
To test long-term reliability, sleep data was collected from a subject over 5 random nights across two months and a continuous 21-night period. Using Polysomnography (PSG) as a reference, the system maintained high performance despite changes in sleeping positions, demonstrating its potential for home-based health monitoring.
Discussion and Future Work
While RF-HRV excels at rate-based monitoring, it faced challenges with complex arrhythmia detection, achieving 47.6% accuracy. This suggests that while the timing of beats is captured accurately, diagnosing specific irregular rhythms may require more sophisticated feature extraction or hybrid machine learning models.
Current limitations also include multi-person interference and sensitivity to vigorous motion artifacts. Future research will focus on advanced beamforming to isolate signals in multi-occupant environments and improving signal-to-noise ratios in dynamic settings. Despite these challenges, RF-HRV represents a significant step toward non-invasive, long-term, and precise cardiac health monitoring.
Summary updated