Sleep Stage Classification Using Unsupervised Feature Learning

Längkvist, Martin; Karlsson, Lars; Loutfi, Amy

doi:10.1155/2012/107046

Cited by 210 publications

(160 citation statements)

References 26 publications

Supporting

Mentioning

159

Contrasting

Unclassified

Order By: Relevance

“…Vural and Yildiz [95] used the principal component analysis [96] for the classification of hybrid features and reported an accuracy of 69.98%. Langkvist et al [97] performed sleep stage classification using deep belief nets, an unsupervised feature learning approach.…”

Section: Discussionmentioning

confidence: 99%

Nonlinear Dynamics Measures for Automated EEG-Based Sleep Stage Detection

et al. 2015

View full text Add to dashboard Cite

Background: The brain's continuous neural activity during sleep can be monitored by electroencephalogram (EEG) signals. The EEG wave pattern and frequency vary during five stages of sleep. These subtle variations in sleep EEG signals cannot be easily detected through visual inspection. Summary: A range of time, frequency, time-frequency and nonlinear analysis methods can be applied to understand the complex physiological signals and their chaotic behavior. This paper presents a comprehensive comparative review and analysis of 29 nonlinear dynamics measures for EEG-based sleep stage detection. Key Messages: The characteristic ranges of these features are reported for the five different sleep stages. All nonlinear measures produce clinically significant results, that is, they can discriminate the individual sleep stages. Feature ranking based on the statistical F-value, however, shows that the third order cumulant of higher order spectra yields the most discriminative result. The distinct value ranges for each sleep stage and the discriminative power of the features can be used for sleep disorder diagnosis, treatment monitoring, and drug efficacy assessment.

show abstract

Section: Discussionmentioning

confidence: 99%

Nonlinear Dynamics Measures for Automated EEG-Based Sleep Stage Detection

et al. 2015

View full text Add to dashboard Cite

show abstract

“…Längkvist et al [16] applied DBN in time series classification and obtained a remarkable result. The standard DBN optimizes the posterior probability ( | ) of the class labels given the current input .…”

Section: Cycle_dbn For Time Series Classificationmentioning

confidence: 99%

“…This data set records 18 activities performed by 9 subjects wearing 3 IMUs and a HR-monitor. Each of data contains 54 columns per row and the columns contain the following data: timestamp (1), activityID (2), heart rate (3), IMU hand (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20), IMU chest (21-37), and IMU ankle (38-54). In our experiment, we only select 7 activities which are "lying (1)," "sitting (2)," "standing (3)," "walking (4)," "running (5)," "cycling (6)," "Nordic walking (7)."…”

Section: 21mentioning

confidence: 99%

“…Chao et al [15] evaluated the DBN performance as a forecasting tool on predicting exchange rate. Längkvist et al [16] applied DBN for sleep stage classification and evaluated the performance. The result illustrated that DBN either with features (feat-DBN) or using the raw 2 Mathematical Problems in Engineering data (raw-DBN) performed better than the feat-GOHMM.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Cycle Deep Belief Network Model for Multivariate Time Series Classification

Wang

Hua

Hao

et al. 2017

Mathematical Problems in Engineering

View full text Add to dashboard Cite

Multivariate time series (MTS) data is an important class of temporal data objects and it can be easily obtained. However, the MTS classification is a very difficult process because of the complexity of the data type. In this paper, we proposed a Cycle Deep Belief Network model to classify MTS and compared its performance with DBN and KNN. This model utilizes the presentation learning ability of DBN and the correlation between the time series data. The experimental results showed that this model outperforms other four algorithms: DBN, KNN_ED, KNN_DTW, and RNN.

show abstract

“…Längkvist in [38] have used RBMs to classify sleep stage data with a greater accuracy than conventional learning systems based on Gaussian mixture and hidden Markov models. Wulsin et al [39] have attempted to model electroencephalography (EEG) signals using Deep Belief Networks composed of RBMs for fast classification and anomaly measurement in a semi-supervised 21 http://googleblog.blogspot.com/2012/06/using-large-scale-brain-simulations-for.html.…”

Section: Future Vision-cognitive Computing and Artificial Intelligencementioning

confidence: 99%

Future of Healthcare—Sensor Data-Driven Prognosis

Pal

Mukherjee

Dey

2016

Wireless World in 2050 and Beyond: A Window Into the Future!

View full text Add to dashboard Cite

Connected Wellness/Healthcare is all about retrieving people's physiological parameters through sensors and performing analysis. Individually, such analytics can help a patient to maintain a wellness regime, or to decide when to see a doctor, and can assist doctors in diagnosis. Collectively, such analytics, if performed over a long time over large set of patients, has the potential to discover new disease diagnostic and treatment protocols. In this chapter, we first discuss how advances in sensing and analytics can take us from a reactive illness-driven healthcare system to a proactive wellness-driven system. We introduce an IoT driven architecture and discuss how non-invasive, affordable, unobtrusive sensing using mobile phones, wearables and nearables is making physiological and pathological data collection from human body possible in thus far unimaginable ways. We also introduce breakthrough technologies in form of exosomes and 3D organ printing that has the potential to disrupt the future healthcare landscape.Assuming various kinds of physiological and pathological measurements becoming available 24 × 7 from patients, a whole set of analytics opportunities open up for alerting, diagnostics, prognostics and diagnostic rules discovery. We discuss how big data analytics techniques like anomaly detection, stream reasoning, evidence based learning moving towards AI techniques like deep learning and cognitive computing will help in moving from the current art of diagnostics to future science of prognostics. Finally we take a peek into a hypothetical healthcare system of future where with help of all the above technologies, we describe a day in the life of a person in 2050.

show abstract

Sleep Stage Classification Using Unsupervised Feature Learning

Cited by 210 publications

References 26 publications

Nonlinear Dynamics Measures for Automated EEG-Based Sleep Stage Detection

Nonlinear Dynamics Measures for Automated EEG-Based Sleep Stage Detection

A Cycle Deep Belief Network Model for Multivariate Time Series Classification

Future of Healthcare—Sensor Data-Driven Prognosis

Contact Info

Product

Resources

About