Parsing wireless electrocardiogram signals with context free grammar conditional random fields

Nguyen, Nguyen; Adams,; NATARAJAN, .; Marlin,

doi:10.1109/wh.2016.7764570

Cited by 3 publications

(3 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the modeling side, segmentation based models have been successfully used for a wide variety of activity recognition tasks [7, 41, 53, 54]. For example, Tang et al, [54] and Sung et al, [53] use conditional segmentation models for labeling and segmenting activities in video streams.…”

Section: Background and Related Workmentioning

confidence: 99%

“…Adams et al, [7] use a hierarchical segmentation model to label and segment smoking activities in respiration data. Most closely related to our approach, [41] use a CRF-CFG model for ECG morphology extraction. In this work, we develop a grammar for a CRF-CFG model to detect conversation episodes, which has different characteristics than prior works on ECG morphology or smoking, demonstrating wider applicability for the CRF-CFG approach.…”

Section: Background and Related Workmentioning

confidence: 99%

“…In this section, we begin by reviewing the CRF-CFG model [16] and then describe how we apply it to speech detection and conversation episode segmentation. The CRF-CFG model was first used in mHealth to extract heart-beat signal morphology (QRS complex) in ECG time-series data [41]. To the best of our knowledge, ours is the first work to apply CRF-CFG model for detecting conversation episodes on respiration time-series data.…”

Section: Speech Detection Using Conditional Random Field-context Freementioning

confidence: 99%

See 2 more Smart Citations

rConverse

Bari

Adams

Rahman

et al. 2018

Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.

View full text Add to dashboard Cite

Monitoring of in-person conversations has largely been done using acoustic sensors. In this paper, we propose a new method to detect moment-by-moment conversation episodes by analyzing breathing patterns captured by a mobile respiration sensor. Since breathing is affected by physical and cognitive activities, we develop a comprehensive method for cleaning, screening, and analyzing noisy respiration data captured in the field environment at individual breath cycle level. Using training data collected from a speech dynamics lab study with 12 participants, we show that our algorithm can identify each respiration cycle with 96.34% accuracy even in presence of walking. We present a Conditional Random Field, Context-Free Grammar (CRF-CFG) based conversation model, called rConverse, to classify respiration cycles into speech or non-speech, and subsequently infer conversation episodes. Our model achieves 82.7% accuracy for speech/non-speech classification and it identifies conversation episodes with 95.9% accuracy on lab data using a leave-one-subject-out cross-validation. Finally, the system is validated against audio ground-truth in a field study with 32 participants. rConverse identifies conversation episodes with 71.7% accuracy on 254 hours of field data. For comparison, the accuracy from a high-quality audio-recorder on the same data is 71.9%.

show abstract

Section: Background and Related Workmentioning

confidence: 99%

Section: Background and Related Workmentioning

confidence: 99%

Section: Speech Detection Using Conditional Random Field-context Freementioning

confidence: 99%

See 1 more Smart Citation

rConverse

Bari

Adams

Rahman

et al. 2018

Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.

View full text Add to dashboard Cite

show abstract

A Time-Limited Scoping Review Investigating the Use of Wearable Biosensors in the Substance Use Field

Jeyadevan,

Grigg

2024

Curr Addict Rep

View full text Add to dashboard Cite

Purpose of Review The purpose of this time-limited scoping review was to investigate applications of wearables under investigation in the substance use field and consider the feasibility of these devices as an overdose intervention tool. Recent Findings Wearables were used for drug use detection, overdose detection and/or opioid overdose detection paired with naloxone delivery. Highest accuracy was associated with wrist-worn devices interpreted with machine learning algorithms. Challenges included difficulty in continuous monitoring, optimising algorithms, variability in device accuracy, and differences between lab and community settings. Qualitative studies investigated willingness to wear biosensors, and identified key themes of comfort, security, and aesthetics. Summary Findings represent promising first steps towards use of wearables as a harm reduction tool for people who use substances. However, few studies investigated overdose detection and real-time intervention, and devices were limited in their biosensing capacities. As technologies advance, systematic reviews are required, and devices should be evaluated and validated to ensure they are fit-for-purpose.

show abstract

Computational Complexity of Algorithms for Optimization of Multi-Hybrid Renewable Energy Systems

Igbinovia

Křupka

2018

2018 International Conference on Power System Technology (POWERCON)

View full text Add to dashboard Cite

Parsing wireless electrocardiogram signals with context free grammar conditional random fields

Cited by 3 publications

References 21 publications

rConverse

rConverse

A Time-Limited Scoping Review Investigating the Use of Wearable Biosensors in the Substance Use Field

Computational Complexity of Algorithms for Optimization of Multi-Hybrid Renewable Energy Systems

Contact Info

Product

Resources

About