This research group focuses on advancing contactless home health monitoring for neurological and chronic physiological conditions. Their work leverages multimodal sensing technologies like mmWave radar, depth camera, and imu sensors, etc. to capture fine-grained human biomarkers and micro-movements that were previously difficult to track outside a clinical setting. These innovations enable the precise quantification of motor dysfunctions and physiological indicators—from the involuntary tremors of Parkinson’s disease to minute eye-blink kinematics and arterial pulse waves—facilitating early screening and personalized disease management in daily life.

Specific systems for multi-level human-centric sensing include mmTremor, a practical monitoring solution that detect pathological tremors (Parkinson’s and Essential Tremor) from daily activities, providing objective data for neurological assessment. Blinic and SDE pioneered contactless dry eye assessment techniques with micro blink kinematics and clinical knowledge distillation. WaveBP enables continuous arterial blood pressure monitoring, capturing tiny pulse waveforms via radar to support cardiovascular health management. Collectively, these AI-driven solutions demonstrate robust performance in home environments, offering a comprehensive approach to managing complex health conditions through invisible sensing.

Related publications:

• [ACM MobiCom'25] Q. Hu, Y. Zhou, J. Wang, Z. Huang, G. Li, Q. Xu, Q. Zhang, “mmTremor: Practical Tremor Monitoring for Parkinson’s Disease and Essential Tremor in Daily Life”

  
  

  Introduction: This study introduces mmTremor, the first privacy-preserving contactless system designed for practical tremor monitoring in patients with Parkinson’s Disease and Essential Tremor. Addressing the limitations of adherence and motion interference in existing at-home solutions, the system employs a multimodal spatiotemporal deep learning pipeline to robustly detect tremors during activities of daily living (ADL). Validated across diverse real-world environments with 37 participants, mmTremor achieves a high macro-F1 score of 0.877, demonstrating superior discriminability even with unseen users. The research highlights the potential of contactless sensing to transform daily disease management by providing objective, continuous symptom tracking without wearable burdens.

  
  

• [ACM MobiCom'25] M. Xue, W. Xie, Z. Yi, Z. Zhang, S. Wu, Y. Zhu, Q. Zhang, C. Chen, “Home-based Dry Eye Assessment via Blink Kinematics Using mmWave and Clinical Knowledge Distillation”

  
  

  Introduction: This study developed Blinic, a contactless home monitoring system designed to assess Dry Eye Disease (DED) by predicting Tear Film Break-Up Time (TBUT) using commercial millimeter-wave radar. To replace costly clinical procedures, the system utilizes a specialized antenna-coded MIMO radar design to capture minute blink kinematics with high precision. A key technical innovation is the integration of a teacher-student learning framework and a fine-tuned Large Language Model (DryEye-LLM), which transfer professional medical insights from clinical records directly to the radar-based model. Validated with 192 participants, Blinic achieved a mean absolute error of 2.73 seconds for TBUT and 90.54% accuracy for severity grading, offering a practical, accurate solution for managing eye health in daily life.

  
  

• [ACM IMWUT 2024] Q. Hu, Q. Zhang, H. Lu, S. Wu, Y. Zhou, Q. Huang, H. Chen,Y. Chen, N. Zhao, “Contactless Arterial Blood Pressure Waveform Monitoring with mmWave Radar” (Distinguished Paper Award🎉🎉🎉)

  
  

  Introduction: This study introduces WaveBP, the first contactless system capable of monitoring Arterial Blood Pressure Waveforms (ABPW) using commercial millimeter-wave radar. Addressing the limitations of invasive or wearable devices, the system leverages a novel hybrid Transformer model, mmFormer, which extracts precise cardiac information from radar reflections based on hemodynamic principles. To ensure clinical-grade accuracy without physical contact, WaveBP incorporates a cross-modality knowledge transfer framework that learns from traditional ECG/PPG signals, alongside a beamforming-based data augmentation technique to enhance signal robustness. Evaluated on 43 subjects, the system achieved a high waveform correlation of 0.903 and a low measurement error of -0.14±7.48 mmHg, demonstrating its potential for continuous, non-intrusive cardiovascular health tracking and cardiac abnormality detection..

  
  

• [ACM IMWUT 2024] M. Xue, Y. Zeng, S. Gu, Q. Zhang, B. Tian, and C. Chen. “SDE: Early Screening for Dry Eye Disease with Wireless Signals”, in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2024.

  
  

  Introduction: This study presents SDE, a contactless screening system designed to identify Dry Eye Disease (DED) using ubiquitous radio frequency (RF) signals, bypassing the need for specialized in-hospital equipment. To capture subtle physiological indicators, the system extracts fine-grained biomarkers from signal variance, specifically focusing on spontaneous blinking actions that characterize eye health. A critical technical advancement is the integration of a deep learning-based unsupervised domain adaptation model, which enables the system to generalize to new users and varying environments—such as homes, clinics, and offices—by aligning features in both local and global spaces. Validated with 54 volunteers across four distinct scenarios, SDE demonstrates the capability to accurately screen for DED in real-world settings, offering a convenient and accessible tool for early intervention and population-level eye health monitoring.

  
  

Demos:

• mmTremor:

• WaveBP:

• Blinic:

• SDE: