Jugular Venous Pulse Waveform Extraction From a Wearable Radio Frequency Sensor.

Many prevalent heart diseases can be indicated by the features of the jugular venous pulse (JVP), an efficacious indicator of right heart health. However, JVP dynamics are not widely utilized in clinical settings as its observation and sensing remain cumbersome. Non-invasive measures of cardiac behavior, including the JVP, are of growing interest to enable continuous and at-home monitoring of cardiac disorders. In this work, we propose a wearable near-field radio-frequency (RF) sensor affixed with a neck collar on the clavicle over the internal jugular vein to enable non-invasive JVP sensing. We employed a complex vector injection signal processing method to extract repeatable JVP waveform features in multiple postures. With a 21-subject human study, we demonstrated morphologically consistent JVP sensing with consistent a-, c-, and v-wave feature timings, benchmarked by synchronous electrocardiogram and phonocardiogram. Further, inter-postural experiments demonstrated the capability of the proposed system to quantify morphological changes to the JVP which are present in many cardiac disorders. The results of this work suggest the proposed near-field RF sensor is capable of non-invasive JVP monitoring, potentially enabling improved sensing in both clinical and ambulatory environments.

Systolic Time Interval Extraction in Hypertensive and Hypotensive Pig Models Using Wearable Near-Field Radio-Frequency Sensors.

Screening and monitoring for cardiovascular diseases (CVDs) can be enabled by analyzing systolic time intervals (STIs). As CVDs have a strong causal correlation with hypertension, it is important to validate STI sensor accuracy in hypertensive hearts to ensure consistent performance in this prevalent cardiac disease state. This work presents STI extraction using a non-invasive near-field radio-frequency (RF) sensor during normotension, hypertension, and hypotension in a pig model. Waveform features of semilunar and atrioventricular valve dynamics during systole were extracted to derive isovolumic contraction time (ICT) and left ventricular ejection time (LVET), benchmarked by a phonocardiogram and aortic catheterization. Study-wide mean relative ICT and LVET errors were -4.4ms and -3.6ms, respectively, demonstrating high accuracy during both normal and abnormal systemic pressures.Clinical relevance- This work demonstrates accurate STI extraction with relative error less than 5 ms from a non-invasive near-field RF sensor during normotensive, hypotensive, and hypertensive systemic pressures, validating the sensor's accuracy as a screening tool during this disease state.

Detection and Prediction of Sleep Disorders by Covert Bed-Integrated RF Sensors.

OBJECTIVE: Respiratory disturbances during sleep are a prevalent health condition that affects a large adult population. The gold standard to evaluate sleep disorders including apnea is overnight polysomnography, which requires a trained technician for live monitoring and post-processing scoring. Currently, the disorder events can hardly be predicted using the respiratory waveforms preceding the events. The objective of this paper is to develop an autonomous system to detect and predict respiratory events reliably based on real-time covert sensing. METHODS: A bed-integrated radio-frequency (RF) sensor by near-field coherent sensing (NCS) was employed to retrieve continuous respiratory waveforms without user's awareness. Overnight recordings were collected from 27 patients in the Weill Cornell Center for Sleep Medicine. We extracted respiratory features to feed into the random-forest machine learning model for disorder detection and prediction. The technician annotation, derived from observation by polysomnography, was used as the ground truth during the supervised learning. RESULTS: Apneic event detection achieved a sensitivity and specificity up to 88.6% and 89.0% for k-fold validation, and 83.1% and 91.6% for subject-independent validation. Prediction of forthcoming apneic events could be made up to 90 s in advance. Apneic event prediction achieved a sensitivity and specificity up to 81.3% and 82.1% for k-fold validation, and 80.5% and 82.4% for subject-independent validation. The most important features for event detection and prediction can be assessed in the learning model. CONCLUSION: A bed-integrated RF sensor can covertly and reliably detect and predict apneic events. SIGNIFICANCE: Predictive warning of the sleep disorders in advance can intervene serious apnea, especially for infants, servicemen, and patients with chronic conditions.

Deduced Respiratory Scores on COVID-19 Patients Learning from Exertion-Induced Dyspnea.

Dyspnea is one of the most common symptoms of many respiratory diseases, including COVID-19. Clinical assessment of dyspnea relies mainly on self-reporting, which contains subjective biases and is problematic for frequent inquiries. This study aims to determine if a respiratory score in COVID-19 patients can be assessed using a wearable sensor and if this score can be deduced from a learning model based on physiologically induced dyspnea in healthy subjects. Noninvasive wearable respiratory sensors were employed to retrieve continuous respiratory characteristics with user comfort and convenience. Overnight respiratory waveforms were collected on 12 COVID-19 patients, and a benchmark on 13 healthy subjects with exertion-induced dyspnea was also performed for blind comparison. The learning model was built from the self-reported respiratory features of 32 healthy subjects under exertion and airway blockage. A high similarity between respiratory features in COVID-19 patients and physiologically induced dyspnea in healthy subjects was observed. Learning from our previous dyspnea model of healthy subjects, we deduced that COVID-19 patients have consistently highly correlated respiratory scores in comparison with normal breathing of healthy subjects. We also performed a continuous assessment of the patient's respiratory scores for 12-16 h. This study offers a useful system for the symptomatic evaluation of patients with active or chronic respiratory disorders, especially the patient population that refuses to cooperate or cannot communicate due to deterioration or loss of cognitive functions. The proposed system can help identify dyspneic exacerbation, leading to early intervention and possible outcome improvement. Our approach can be potentially applied to other pulmonary disorders, such as asthma, emphysema, and other types of pneumonia.

Attention Detection by Heartbeat and Respiratory Features from Radio-Frequency Sensor.

This work presents a study on users' attention detection with reference to a relaxed inattentive state using an over-the-clothes radio-frequency (RF) sensor. This sensor couples strongly to the internal heart, lung, and diaphragm motion based on the RF near-field coherent sensing principle, without requiring a tension chest belt or skin-contact electrocardiogram. We use cardiac and respiratory features to distinguish attention-engaging vigilance tasks from a relaxed, inattentive baseline state. We demonstrate high-quality vitals from the RF sensor compared to the reference electrocardiogram and respiratory tension belts, as well as similar performance for attention detection, while improving user comfort. Furthermore, we observed a higher vigilance-attention detection accuracy using respiratory features rather than heartbeat features. A high influence of the user's baseline emotional and arousal levels on the learning model was noted; thus, individual models with personalized prediction were designed for the 20 participants, leading to an average accuracy of 83.2% over unseen test data with a high sensitivity and specificity of 85.0% and 79.8%, respectively.

Physiological Features of Cardiac Ventricle and Valve Dynamics from Wearable Radio-Frequency Sensors.

Early detection of cardiovascular diseases via non-invasive, convenient, and continuous monitoring is crucial to reducing preventable deaths. This paper illustrates such monitoring using wearable near-field radio-frequency sensors to analyze ventricle and valve transients, which can be used as indicators of myriad cardiac disorders. We applied a novel vector injection signal processing method to improve timing consistency in ventricular contraction, ventricular relaxation, and valve opening extraction. The median relative timing error in valve opening detection was 14.7ms and 37.8ms for semilunar and atrioventricular valves, respectively, as benchmarked by the S1 and S2 heart sounds from a synchronous phonocardiogram. Clinical Relevance- No wearable sensor currently exists to conveniently and reliably evaluate ventricular and valvular dynamics, specifically valvular opening. Beyond extraction of the heart rate and its variation, the method in this paper has the potential to enable non-invasive measurements of detailed cardiac cycle timing features including valve openings, isovolumetric contraction/relaxation times, and ejection periods, improving the monitoring of patient health away from clinical healthcare centers.

Wearable RF Near-Field Cough Monitoring by Frequency-Time Deep Learning.

Coughing is a common symptom for many respiratory disorders, and can spread droplets of various sizes containing bacterial and viral pathogens. Mild coughs are usually overlooked in the early stage, not only because they are barely noticeable by the person and the people around, but also because the present recording method is not comfortable, private, or reliable for long-term monitoring. In this paper, a wearable radio-frequency (RF) sensor is presented to recognize the mild cough signal directly from the local trachea vibration characteristics, and can isolate interferences from nearby people. The sensor operates at the ultra-high-frequency band, and can couple the RF energy to the upper respiratory track by the near field of the sensing antenna. The retrieved tissue vibration caused by the cough airflow burst can then be analyzed by a convolutional neural network trained on the frequency-time spectra. The sensing antenna design is analyzed for performance improvement. During the human study of 5 participants over 100 minutes of prescribed routines, the overall recognition ratio is above 90% and the false positive ratio during other routines is below 2.09%.

Wearable radio-frequency sensing of respiratory rate, respiratory volume, and heart rate.

Many health diagnostic systems demand noninvasive sensing of respiratory rate, respiratory volume, and heart rate with high user comfort. Previous methods often require multiple sensors, including skin-touch electrodes, tension belts, or nearby off-the-body readers, and hence are uncomfortable or inconvenient. This paper presents an over-clothing wearable radio-frequency sensor study, conducted on 20 healthy participants (14 females) performing voluntary breathing exercises in various postures. Two prototype sensors were placed on the participants, one close to the heart and the other below the xiphoid process to couple to the motion from heart, lungs and diaphragm, by the near-field coherent sensing principle. We can achieve a satisfactory correlation of our sensor with the reference devices for the three vital signs: heart rate (r = 0.95), respiratory rate (r = 0.93) and respiratory volume (r = 0.84). We also detected voluntary breath-hold periods with an accuracy of 96%. Further, the participants performed a breathing exercise by contracting abdomen inwards while holding breath, leading to paradoxical outward thorax motion under the isovolumetric condition, which was detected with an accuracy of 83%.