Pragya Sharma

Pragya Sharma

PhD, School of Electrical and Computer Engineering, Cornell University, Prof. Edwin Kan Group

Noninvasive Cardiorespiratory Monitoring using Wearable RF Near-Field Coherent Sensing

Cornell University, Graduate Research Assistant

2016-2020

This project developed and validated a low-cost wearable radio-frequency (RF) near-field coherent sensing (NCS) system for continuous cardiopulmonary monitoring and respiratory disorder detection.

Problem

Continuous monitoring of respiratory and cardiac physiology traditionally relies on skin-contact sensors such as ECG leads and chest belts. These systems limit long-term comfort, are prone to motion artifacts, and are not well suited for unobtrusive or home-based monitoring.

Non-contact RF sensing approaches showed promise but were historically limited by low signal-to-noise ratio, sensitivity to environmental motion, and limited validation in realistic clinical scenarios.

Approach

Subject wearing RF NCS and reference sensors. Bottom-left shows respiration waveforms from RF NCS vs BIOPAC chest belts. Bottom-right shows extracted heartbeat waveform from NCS vs ECG and estimated HR.

I designed and implemented a wearable near-field RF sensing platform capable of extracting detailed respiratory and heartbeat waveforms through clothing using a software-defined radio architecture.

Key components of the work included:

  • Designing and executing an IRB-approved human subject study (N = 30) involving simulated breathing disorders, apnea events, and attention/vigilance tasks
  • Developing a real-time, over-clothing cardiopulmonary sensing system with high signal fidelity
  • Implementing a nearly tuning-free peak-detection algorithm for respiratory disorder identification across a broad frequency range (2–40 breaths per minute)
  • Achieving high respiratory rate accuracy (94.8%, RMSE 2.9 BPM) and respiratory volume estimation performance (RMSE 0.11 L) across irregular breathing patterns
  • Designing a semi-supervised SVM-based framework for motion artifact detection (91% accuracy)
  • Engineering a bed-integrated RF system for sleep apnea detection in collaboration with the Weill Cornell Medical Sleep Center, including optimized antenna design to separately capture thoracic and abdominal motion
  • Developing and programming an attention-detection protocol (Mackworth Clock Task) to study relationships between vigilance and cardiopulmonary dynamics

Impact

This work represents one of the early demonstrations of near-field RF sensing for simultaneous respiratory and cardiac monitoring with strong signal robustness and clinical validation.

By extending evaluation beyond laboratory conditions into clinical sleep settings, this project advanced the feasibility of RF-based monitoring as a scalable, low-burden alternative to conventional contact sensors for both hospital and home health applications.