1
|
Kam K, Rapoport DM, Parekh A, Ayappa I, Varga AW. WaveSleepNet: An interpretable deep convolutional neural network for the continuous classification of mouse sleep and wake. J Neurosci Methods 2021; 360:109224. [PMID: 34052291 DOI: 10.1016/j.jneumeth.2021.109224] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/25/2021] [Revised: 04/28/2021] [Accepted: 05/17/2021] [Indexed: 12/24/2022]
Abstract
BACKGROUND Recent advancement in deep learning provides a pivotal opportunity to potentially supplement or supplant the limiting step of manual sleep scoring. NEW METHOD In this paper, we characterize the WaveSleepNet (WSN), a deep convolutional neural network (CNN) that uses wavelet transformed images of mouse EEG/EMG signals to autoscore sleep and wake. RESULTS WSN achieves an epoch by epoch mean accuracy of 0.86 and mean F1 score of 0.82 compared to manual scoring by a human expert. In mice experiencing mechanically induced sleep fragmentation, an overall epoch by epoch mean accuracy of 0.80 is achieved by WSN and classification of non-REM (NREM) sleep is not compromised, but the high level of sleep fragmentation results in WSN having greater difficulty differentiating REM from NREM sleep. We also find that WSN achieves similar levels of accuracy on an independent dataset of externally acquired EEG/EMG recordings with an overall epoch by epoch accuracy of 0.91. We also compared conventional summary sleep metrics in mice sleeping ad libitum. WSN systematically biases sleep fragmentation metrics of bout number and bout length leading to an overestimated degree of sleep fragmentation. COMPARISON WITH EXISTING METHODS In a cross-validation, WSN has a greater macro and stage-specific accuracy compared to a conventional random forest classifier. Examining the WSN, we find that it automatically learns spectral features consistent with manual scoring criteria that are used to define each class. CONCLUSION These results suggest to us that WSN is capable of learning visually agreeable features and may be useful as a supplement to human manual scoring.
Collapse
Affiliation(s)
- Korey Kam
- Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, USA.
| | - David M Rapoport
- Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, USA
| | - Ankit Parekh
- Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, USA
| | - Indu Ayappa
- Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, USA
| | - Andrew W Varga
- Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, USA.
| |
Collapse
|
2
|
Bogáthy E, Papp N, Vas S, Bagdy G, Tóthfalusi L. AM-251, A Cannabinoid Antagonist, Modifies the Dynamics of Sleep-Wake Cycles in Rats. Front Pharmacol 2019; 10:831. [PMID: 31404291 PMCID: PMC6675864 DOI: 10.3389/fphar.2019.00831] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/02/2019] [Accepted: 06/28/2019] [Indexed: 01/01/2023] Open
Abstract
Study Objectives: (a) To describe the microarchitecture of wakefulness and sleep following administrations of 5- and 10-mg/kg AM-251 in rats. (b) To develop a new statistical method to follow bout-to-bout dynamics. Method: Wistar rats (n = 6) had been equipped with electroencephalography (EEG) and electromyography (EMG) electrodes. Following their recovery and habituation after the surgery, the animals were injected with vehicle and 5- and 10-mg/kg AM-251 intraperitoneally and EEG, EMG, and motor activity were analyzed for the subsequent 3 h. Results: AM-251 induced a dose- and time-dependent increase in the number of bouts in active wake (AW), and it decreased this number in all other vigilance states except in passive wake (PW). In contrast, the bout duration in PW compensatory decreased. The effect of AM-251 on the sleep transition dynamics was monitored with a new tool we call "transition heatmap." The analysis of bout trajectories with transition heatmaps reveals a highly organized pattern. Conclusion: AM-251 selectively influences the frequency of vigilance state transitions, but it has no direct impact on the state lengths. AM-251 markedly changed the state transition dynamics, which was visualized with the help of state transition heatmaps.
Collapse
Affiliation(s)
- Emese Bogáthy
- Department of Pharmacodynamics, Semmelweis University, Budapest, Hungary
| | - Noémi Papp
- Department of Pharmacodynamics, Semmelweis University, Budapest, Hungary
| | - Szilvia Vas
- Department of Pharmacodynamics, Semmelweis University, Budapest, Hungary.,MTA-SE, Neuropsychopharmacology and Neurochemistry Research Group, Budapest, Hungary.,Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge, United Kingdom
| | - György Bagdy
- Department of Pharmacodynamics, Semmelweis University, Budapest, Hungary.,MTA-SE, Neuropsychopharmacology and Neurochemistry Research Group, Budapest, Hungary.,NAP-A-SE, New Antidepressant Target Research Group, Budapest, Hungary.,NAP-2-SE, New Antidepressant Target Research Group, Budapest, Hungary
| | - László Tóthfalusi
- Department of Pharmacodynamics, Semmelweis University, Budapest, Hungary
| |
Collapse
|
3
|
Duan Q, Liu J, Zhao D. A fast algorithm for short term electric load forecasting by a hidden semi-markov process. J STAT COMPUT SIM 2019. [DOI: 10.1080/00949655.2019.1573241] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/27/2022]
Affiliation(s)
- Qihong Duan
- School of Mathematics and Statistics, Xi'an Jiaotong University, Xi'an, People's Republic of China
| | - Junrong Liu
- School of Mathematics, Northwest University, Xi'an, People's Republic of China
| | - Dengfu Zhao
- School of Electrical Engineering, Xi'an Jiaotong University, Xi'an, People's Republic of China
| |
Collapse
|
4
|
Allocca G, Ma S, Martelli D, Cerri M, Del Vecchio F, Bastianini S, Zoccoli G, Amici R, Morairty SR, Aulsebrook AE, Blackburn S, Lesku JA, Rattenborg NC, Vyssotski AL, Wams E, Porcheret K, Wulff K, Foster R, Chan JKM, Nicholas CL, Freestone DR, Johnston LA, Gundlach AL. Validation of 'Somnivore', a Machine Learning Algorithm for Automated Scoring and Analysis of Polysomnography Data. Front Neurosci 2019; 13:207. [PMID: 30936820 PMCID: PMC6431640 DOI: 10.3389/fnins.2019.00207] [Citation(s) in RCA: 34] [Impact Index Per Article: 6.8] [Reference Citation Analysis] [Abstract] [Key Words] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/04/2018] [Accepted: 02/22/2019] [Indexed: 12/14/2022] Open
Abstract
Manual scoring of polysomnography data is labor-intensive and time-consuming, and most existing software does not account for subjective differences and user variability. Therefore, we evaluated a supervised machine learning algorithm, SomnivoreTM, for automated wake–sleep stage classification. We designed an algorithm that extracts features from various input channels, following a brief session of manual scoring, and provides automated wake-sleep stage classification for each recording. For algorithm validation, polysomnography data was obtained from independent laboratories, and include normal, cognitively-impaired, and alcohol-treated human subjects (total n = 52), narcoleptic mice and drug-treated rats (total n = 56), and pigeons (n = 5). Training and testing sets for validation were previously scored manually by 1–2 trained sleep technologists from each laboratory. F-measure was used to assess precision and sensitivity for statistical analysis of classifier output and human scorer agreement. The algorithm gave high concordance with manual visual scoring across all human data (wake 0.91 ± 0.01; N1 0.57 ± 0.01; N2 0.81 ± 0.01; N3 0.86 ± 0.01; REM 0.87 ± 0.01), which was comparable to manual inter-scorer agreement on all stages. Similarly, high concordance was observed across all rodent (wake 0.95 ± 0.01; NREM 0.94 ± 0.01; REM 0.91 ± 0.01) and pigeon (wake 0.96 ± 0.006; NREM 0.97 ± 0.01; REM 0.86 ± 0.02) data. Effects of classifier learning from single signal inputs, simple stage reclassification, automated removal of transition epochs, and training set size were also examined. In summary, we have developed a polysomnography analysis program for automated sleep-stage classification of data from diverse species. Somnivore enables flexible, accurate, and high-throughput analysis of experimental and clinical sleep studies.
Collapse
Affiliation(s)
- Giancarlo Allocca
- The Florey Institute of Neuroscience and Mental Health, Parkville, VIC, Australia.,Somnivore Pty. Ltd., Parkville, VIC, Australia
| | - Sherie Ma
- The Florey Institute of Neuroscience and Mental Health, Parkville, VIC, Australia.,Somnivore Pty. Ltd., Parkville, VIC, Australia
| | - Davide Martelli
- The Florey Institute of Neuroscience and Mental Health, Parkville, VIC, Australia.,Laboratory of Autonomic and Behavioral Physiology, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
| | - Matteo Cerri
- Laboratory of Autonomic and Behavioral Physiology, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
| | - Flavia Del Vecchio
- Laboratory of Autonomic and Behavioral Physiology, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
| | - Stefano Bastianini
- PRISM Laboratory, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
| | - Giovanna Zoccoli
- PRISM Laboratory, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
| | - Roberto Amici
- Laboratory of Autonomic and Behavioral Physiology, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
| | - Stephen R Morairty
- Center for Neuroscience, Biosciences Division, SRI International, Menlo Park, CA, United States
| | - Anne E Aulsebrook
- School of BioSciences, The University of Melbourne, Parkville, VIC, Australia
| | - Shaun Blackburn
- School of Life Sciences, La Trobe University, Bundoora, VIC, Australia
| | - John A Lesku
- School of Life Sciences, La Trobe University, Bundoora, VIC, Australia.,Avian Sleep Group, Max Planck Institute for Ornithology, Seewiesen, Germany
| | - Niels C Rattenborg
- Avian Sleep Group, Max Planck Institute for Ornithology, Seewiesen, Germany
| | - Alexei L Vyssotski
- Institute of Neuroinformatics, University of Zurich and ETH Zurich, Zurich, Switzerland
| | - Emma Wams
- The Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom
| | - Kate Porcheret
- The Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom
| | - Katharina Wulff
- The Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom
| | - Russell Foster
- The Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom
| | - Julia K M Chan
- Melbourne School of Psychological Sciences, The University of Melbourne, Parkville, VIC, Australia
| | - Christian L Nicholas
- Melbourne School of Psychological Sciences, The University of Melbourne, Parkville, VIC, Australia.,Institute of Breathing and Sleep, Austin Health, Heidelberg, VIC, Australia
| | - Dean R Freestone
- Department of Medicine, St. Vincent's Hospital, The University of Melbourne, Fitzroy, VIC, Australia
| | - Leigh A Johnston
- Biomedical Engineering, The University of Melbourne, Parkville, VIC, Australia.,Florey Department of Neuroscience and Mental Health, The University of Melbourne, Parkville, VIC, Australia
| | - Andrew L Gundlach
- The Florey Institute of Neuroscience and Mental Health, Parkville, VIC, Australia.,Somnivore Pty. Ltd., Parkville, VIC, Australia.,Florey Department of Neuroscience and Mental Health, The University of Melbourne, Parkville, VIC, Australia
| |
Collapse
|
5
|
Ray EL, Sasaki JE, Freedson PS, Staudenmayer J. Physical activity classification with dynamic discriminative methods. Biometrics 2018; 74:1502-1511. [PMID: 29921026 DOI: 10.1111/biom.12892] [Citation(s) in RCA: 4] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/01/2017] [Revised: 01/01/2018] [Accepted: 03/01/2018] [Indexed: 11/28/2022]
Abstract
A person's physical activity has important health implications, so it is important to be able to measure aspects of physical activity objectively. One approach to doing that is to use data from an accelerometer to classify physical activity according to activity type (e.g., lying down, sitting, standing, or walking) or intensity (e.g., sedentary, light, moderate, or vigorous). This can be formulated as a labeled classification problem, where the model relates a feature vector summarizing the accelerometer signal in a window of time to the activity type or intensity in that window. These data exhibit two key characteristics: (1) the activity classes in different time windows are not independent, and (2) the accelerometer features have moderately high dimension and follow complex distributions. Through a simulation study and applications to three datasets, we demonstrate that a model's classification performance is related to how it addresses these aspects of the data. Dynamic methods that account for temporal dependence achieve better performance than static methods that do not. Generative methods that explicitly model the distribution of the accelerometer signal features do not perform as well as methods that take a discriminative approach to establishing the relationship between the accelerometer signal and the activity class. Specifically, Conditional Random Fields consistently have better performance than commonly employed methods that ignore temporal dependence or attempt to model the accelerometer features.
Collapse
Affiliation(s)
- Evan L Ray
- Department of Mathematics and Statistics, Mount Holyoke College, South Hadley, Massachusetts, U.S.A
| | - Jeffer E Sasaki
- Graduate Program in Physical Education, Universidade Federal do Triangulo Mineiro, Uberaba, Minas Gerais, Brazil
| | - Patty S Freedson
- Department of Kinesiology, University of Massachusetts, Amherst, Massachusetts, U.S.A
| | - John Staudenmayer
- Department of Mathematics and Statistics, University of Massachusetts, Amherst, Massachusetts, U.S.A
| |
Collapse
|
6
|
Brown LA, Hasan S, Foster RG, Peirson SN. COMPASS: Continuous Open Mouse Phenotyping of Activity and Sleep Status. Wellcome Open Res 2016; 1:2. [PMID: 27976750 PMCID: PMC5140024 DOI: 10.12688/wellcomeopenres.9892.2] [Citation(s) in RCA: 40] [Impact Index Per Article: 5.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/31/2022] Open
Abstract
Background: Disruption of rhythms in activity and rest occur in many diseases, and provide an important indicator of healthy physiology and behaviour. However, outside the field of sleep and circadian rhythm research, these rhythmic processes are rarely measured due to the requirement for specialised resources and expertise. Until recently, the primary approach to measuring activity in laboratory rodents has been based on voluntary running wheel activity. By contrast, measuring sleep requires the use of electroencephalography (EEG), which involves invasive surgical procedures and time-consuming data analysis. Methods: Here we describe a simple, non-invasive system to measure home cage activity in mice based upon passive infrared (PIR) motion sensors. Careful calibration of this system will allow users to simultaneously assess sleep status in mice. The use of open-source tools and simple sensors keeps the cost and the size of data-files down, in order to increase ease of use and uptake. Results: In addition to providing accurate data on circadian activity parameters, here we show that extended immobility of >40 seconds provides a reliable indicator of sleep, correlating well with EEG-defined sleep (Pearson’s r >0.95, 4 mice). Conclusions: Whilst any detailed analysis of sleep patterns in mice will require EEG, behaviourally-defined sleep provides a valuable non-invasive means of simultaneously phenotyping both circadian rhythms and sleep. Whilst previous approaches have relied upon analysis of video data, here we show that simple motion sensors provide a cheap and effective alternative, enabling real-time analysis and longitudinal studies extending over weeks or even months. The data files produced are small, enabling easy deposition and sharing. We have named this system COMPASS - Continuous Open Mouse Phenotyping of Activity and Sleep Status. This simple approach is of particular value in phenotyping screens as well as providing an ideal tool to assess activity and rest cycles for non-specialists.
Collapse
Affiliation(s)
- Laurence A Brown
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| | - Sibah Hasan
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| | - Russell G Foster
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| | - Stuart N Peirson
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| |
Collapse
|
7
|
Brown LA, Hasan S, Foster RG, Peirson SN. COMPASS: Continuous Open Mouse Phenotyping of Activity and Sleep Status. Wellcome Open Res 2016. [PMID: 27976750 DOI: 10.12688/wellcomeopenres.9892.1] [Citation(s) in RCA: 25] [Impact Index Per Article: 3.1] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 01/12/2023] Open
Abstract
Background Disruption of rhythms in activity and rest occur in many diseases, and provide an important indicator of healthy physiology and behaviour. However, outside the field of sleep and circadian rhythm research, these rhythmic processes are rarely measured due to the requirement for specialised resources and expertise. Until recently, the primary approach to measuring activity in laboratory rodents has been based on voluntary running wheel activity. By contrast, measuring sleep requires the use of electroencephalography (EEG), which involves invasive surgical procedures and time-consuming data analysis. Methods Here we describe a simple, non-invasive system to measure home cage activity in mice based upon passive infrared (PIR) motion sensors. Careful calibration of this system will allow users to simultaneously assess sleep status in mice. The use of open-source tools and simple sensors keeps the cost and the size of data-files down, in order to increase ease of use and uptake. Results In addition to providing accurate data on circadian activity parameters, here we show that extended immobility of >40 seconds provides a reliable indicator of sleep, correlating well with EEG-defined sleep (Pearson's r >0.95, 4 mice). Conclusions Whilst any detailed analysis of sleep patterns in mice will require EEG, behaviourally-defined sleep provides a valuable non-invasive means of simultaneously phenotyping both circadian rhythms and sleep. Whilst previous approaches have relied upon analysis of video data, here we show that simple motion sensors provide a cheap and effective alternative, enabling real-time analysis and longitudinal studies extending over weeks or even months. The data files produced are small, enabling easy deposition and sharing. We have named this system COMPASS - Continuous Open Mouse Phenotyping of Activity and Sleep Status. This simple approach is of particular value in phenotyping screens as well as providing an ideal tool to assess activity and rest cycles for non-specialists.
Collapse
Affiliation(s)
- Laurence A Brown
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| | - Sibah Hasan
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| | - Russell G Foster
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| | - Stuart N Peirson
- Sleep and Circadian Neuroscience Institute (SCNi), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK
| |
Collapse
|
8
|
Yaghouby F, Sunderam S. SegWay: A simple framework for unsupervised sleep segmentation in experimental EEG recordings. MethodsX 2016; 3:144-55. [PMID: 27014592 PMCID: PMC4792881 DOI: 10.1016/j.mex.2016.02.003] [Citation(s) in RCA: 6] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/27/2015] [Accepted: 02/17/2016] [Indexed: 11/15/2022] Open
Abstract
Sleep analysis in animal models typically involves recording an electroencephalogram (EEG) and electromyogram (EMG) and scoring vigilance state in brief epochs of data as Wake, REM (rapid eye movement sleep) or NREM (non-REM) either manually or using a computer algorithm. Computerized methods usually estimate features from each epoch like the spectral power associated with distinctive cortical rhythms and dissect the feature space into regions associated with different states by applying thresholds, or by using supervised/unsupervised statistical classifiers; but there are some factors to consider when using them:Most classifiers require scored sample data, elaborate heuristics or computational steps not easily reproduced by the average sleep researcher, who is the targeted end user. Even when prediction is reasonably accurate, small errors can lead to large discrepancies in estimates of important sleep metrics such as the number of bouts or their duration. As we show here, besides partitioning the feature space by vigilance state, modeling transitions between the states can give more accurate scores and metrics.
An unsupervised sleep segmentation framework, “SegWay”, is demonstrated by applying the algorithm step-by-step to unlabeled EEG recordings in mice. The accuracy of sleep scoring and estimation of sleep metrics is validated against manual scores.
Collapse
Affiliation(s)
- Farid Yaghouby
- Department of Biomedical Engineering, University of Kentucky, Lexington, KY, USA
| | - Sridhar Sunderam
- Department of Biomedical Engineering, University of Kentucky, Lexington, KY, USA
| |
Collapse
|
9
|
Böckenholt U, McShane BB. Comments on: Latent Markov models: a review of the general framework for the analysis of longitudinal data with covariates. TEST-SPAIN 2014. [DOI: 10.1007/s11749-014-0388-0] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/24/2022]
|