Abstract
Deep learning has been applied successfully in sequence understanding and translation problems, especially in univariate, unimodal contexts, where large number of supervision data are available. The effectiveness of deep learning in more complex (multi-modal, multi-variate) contexts, where supervision data is rare, however, is generally not satisfactory. In this paper, we focus on improving detection and prediction accuracy in precisely such contexts – in particular, we focus on the problem of predicting seizure onsets relying on multi-modal (EEG, ICP, ECG, and ABP) sensory data streams, some of which (such as EEG) are inherently multi-variate due to the placement of multiple sensors to capture spatial distribution of the relevant signals. In particular, we note that multi-variate time series often carry robust, spatio-temporally localized features that could help predict onset events. We further argue that such features can be used to support implementation of metadata supported multivariate attention (or MMA) mechanisms that help significantly improve the effectiveness of neural networks architectures. In this paper, we use the proposed MMA approach to develop a multi-modal LSTM-based neural network architecture to tackle seizure onset detection and prediction tasks relying on EEG, ICP, ECG, and ABP data streams. We experimentally evaluate the proposed architecture under different scenarios – the results illustrate the effectiveness of the proposed attention mechanism, especially compared against other metadata driven competitors.
Similar content being viewed by others
Notes
The EEG real world dataset is imbalanced, time steps having seizure-positive labels are \(7 \%\) of the total at best and \(<2\%\) at the worst (the labels are provided by expert physicians; see Sect. 3.1.1 for dataset details).
In experiments reported in Sect. 3, the descriptor vector length is 128.
Since the healthcare data is HIPAA protected, we make the code available. Also we have publicly available multi-modal COVID, traffic, Bitcoin and S &P index datasets and code for reproduction of results at https://rb.gy/umbzt8.
Provided by NSF testbed “Chameleon: A Large-Scale Re-configurable Experimental Environment for Cloud Research”.
References
Ali ÖG, Arıtürk U (2014) Dynamic churn prediction framework with more effective use of rare event data: The case of private banking. Expert Syst Appl 41(17):7889–7903
Alotaiby TN, Alshebeili SA, Alotaibi FM, Alrshoud SR (2017) Epileptic seizure prediction using CSP and LDA for scalp EEG signals. Computat Intell Neurosci. https://doi.org/10.1155/2017/1240323
Ando S, Huang CY (2017) Deep over-sampling framework for classifying imbalanced data. In: Joint european conference on machine learning and knowledge discovery in databases. Springer, pp. 770–785
Bahdanau D, Cho K, Bengio Y (2014) Neural machine translation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473
Bao F, Deng Y, Kong Y, Ren Z, Suo J, Dai Q (2019) Learning deep landmarks for imbalanced classification. IEEE Trans Neural Netw Learn Syst 31(8):2691–2704
Bogaerts T, Masegosa AD, Angarita-Zapata JS, Onieva E, Hellinckx P (2020) A graph CNN-LSTM neural network for short and long-term traffic forecasting based on trajectory data. Transport Res Part C: Emerg Technol 112:62–77
Cheng J, Dong L, Lapata M (2016) Long short-term memory-networks for machine reading. arXiv preprint arXiv:1601.06733
Cheon S-P, Kim S, Lee S-Y, Lee C-B (2009) Bayesian networks based rare event prediction with sensor data. Knowl-Based Syst 22(5):336–343
Ding K, Gupta PK, Diaz-Arrastia R (2016) Epilepsy after traumatic brain injury. Translational research in traumatic brain injury
Dong Y, Jiang X, Zhou H, Lin Y, Shi Q (2021) Sr2cnn: zero-shot learning for signal recognition. IEEE Trans Signal Process 69:2316–2329
Dumanis SB, French JA, Bernard C, Worrell GA, Fureman BE (2017) Seizure forecasting from idea to reality outcomes of the my seizure gauge epilepsy innovation institute workshop. Eneuro. https://doi.org/10.1523/ENEURO.0349-17.2017
England H (2018) Highways agency network journey time and traffic flow data. Guildford, UK, Highways England
Epilepsy foundation, michigan. https://epilepsymichigan.org/page.php?id=358
Fan M, Chou C-A (2018) Detecting abnormal pattern of epileptic seizures via temporal synchronization of EEG signals. IEEE Trans Biomed Eng 66(3):601–608
Freedman D, Pisani R, Purves R (2020) Statistics: Fourth international student edition. WW Nort Co Httpswww Amaz ComStatistics-Fourth-Int-Stud-Free Accessed 22
Garg Y, Candan KS (2021) Sdma: Saliency-driven mutual cross attention for multi-variate time series. In: 2020 25th International conference on pattern recognition (ICPR). IEEE, pp. 7242–7249
Garg Y, Candan KS (2021) Xm2a: Multi-scale multi-head attention with cross-talk for multi-variate time series analysis. In: 2021 IEEE 4th International Conference on Multimedia Information Processing and Retrieval (MIPR), pp. 151–157. IEEE
Graves A, Graves A (2012) Long short-term memory. Supervised sequence labelling with recurrent neural networks, 37–45
Hain D, Jurowetzki R (2020) Introduction to rare-event predictive modeling for inferential statisticians–a hands-on application in the prediction of breakthrough patents. arXiv preprint arXiv:2003.13441
Islam M, Tanaka T, Iimura Y, Mitsuhashi T, Sugano H, Wang D, Molla M, Islam K et al (2020) Statistical features in high-frequency bands of interictal ieeg work efficiently in identifying the seizure onset zone in patients with focal epilepsy. Entropy 22(12):1415
Jiang Y, Wu D, Deng Z, Qian P, Wang J, Wang G, Chung F-L, Choi K-S, Wang S (2017) Seizure classification from eeg signals using transfer learning, semi-supervised learning and tsk fuzzy system. IEEE Trans Neural Syst Rehabil Eng 25(12):2270–2284
Khan SH, Hayat M, Bennamoun M, Sohel FA, Togneri R (2017) Cost-sensitive learning of deep feature representations from imbalanced data. IEEE Trans Neural Netw Learn Syst 29(8):3573–3587
Khattar A, Quadri S (2022) Camm: cross-attention multimodal classification of disaster-related tweets. IEEE Access 10:92889–92902
Kim J, Jeong J, Shin J (2020) M2m: Imbalanced classification via major-to-minor translation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13896–13905
Kirik S, Çatak Z (2021) Vitamin b12 deficiency observed in children with first afebrile seizures. Cureus. https://doi.org/10.7759/cureus.13745
Klem GH, Lüders HO, Jasper H, Elger C et al (1999) The ten-twenty electrode system of the international federation. Electroencephalogr Clin Neurophysiol 52(3):3–6
Li J, Fong S, Mohammed S, Fiaidhi J, Chen Q, Tan Z (2016) Solving the under-fitting problem for decision tree algorithms by incremental swarm optimization in rare-event healthcare classification. J Med Imag Health Inform 6(4):1102–1110
Li J, Liu L-S, Fong S, Wong RK, Mohammed S, Fiaidhi J, Sung Y, Wong KK (2017) Adaptive swarm balancing algorithms for rare-event prediction in imbalanced healthcare data. PLoS ONE 12(7):0180830
Li G, Lee CH, Jung JJ, Youn YC, Camacho D (2020) Deep learning for EEG data analytics: a survey. Concurr Comput: Pract Exper 32(18):5199
Li J, Fong S, Hu S, Chu VW, Wong RK, Mohammed S, Dey N (2017) Rare event prediction using similarity majority under-sampling technique. In: International Conference on Soft Computing in Data Science. Springer, pp. 23–39
Liu S, Poccia SR, Candan KS, Sapino ML, Wang X (2018) Robust multi-variate temporal features of multi-variate time series. ACM Trans Multim Comp, Commun Appl (TOMM) 14(1):7
Liu L, Chen W, Cao G (2016) Prediction of neonatal amplitude-integrated eeg based on lstm method. In: 2016 IEEE International conference on bioinformatics and biomedicine (BIBM). IEEE, pp. 497–500
Li Y, Yu R, Shahabi C, Liu Y (2017) Diffusion convolutional recurrent neural network: data-driven traffic forecasting. arXiv preprint arXiv:1707.01926
Lun X, Yu Z, Chen T, Wang F, Hou Y (2020) A simplified CNN classification method for MI-EEG via the electrode pairs signals. Front Hum Neurosci 14:338
Luong M-T, Pham H, Manning CD (2015) Effective approaches to attention-based neural machine translation. arXiv preprint arXiv:1508.04025
Natu M, Bachute M, Gite S, Kotecha K, Vidyarthi A (2022) Review on epileptic seizure prediction: machine learning and deep learning approaches. Comput Math Method Med. https://doi.org/10.1155/2022/7751263
Ozcan AR, Erturk S (2019) Seizure prediction in scalp EEG using 3d convolutional neural networks with an image-based approach. IEEE Trans Neural Syst Rehabil Eng 27(11):2284–2293
Palatucci M, Pomerleau D, Hinton GE, Mitchell TM (2009) Zero-shot learning with semantic output codes. In: Advances in neural information processing systems, pp. 1410–1418
Park Y, Luo L, Parhi KK, Netoff T (2011) Seizure prediction with spectral power of EEG using cost-sensitive support vector machines. Epilepsia 52(10):1761–1770
Patel KS, Zhao M, Ma H, Schwartz TH (2013) Imaging preictal hemodynamic changes in neocortical epilepsy. Neurosurg Focus 34(4):10
Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V et al (2011) Scikit-learn: machine learning in python. J Mach Learn Res 12:2825–2830
Peng M, Zhang Q, Xing X, Gui T, Huang X, Jiang Y-G, Ding K, Chen Z (2019) Trainable undersampling for class-imbalance learning. In: Proceedings of the AAAI conference on artificial intelligence, vol. 33, pp. 4707–4714
Pinto M, Coelho T, Leal A, Lopes F, Dourado A, Martins P, Teixeira C (2022) Interpretable EEG seizure prediction using a multiobjective evolutionary algorithm. Sci Rep 12(1):1–15
Population, population change, and estimated components of population change: April 1, 2010 to july 1, 2019 (nst-est2019-alldata). https://www.census.gov/data/tables/time-series/demo/popest/2010s-national-total.html
Ramgopal S, Thome-Souza S, Jackson M, Kadish NE, Fernández IS, Klehm J, Bosl W, Reinsberger C, Schachter S, Loddenkemper T (2014) Seizure detection, seizure prediction, and closed-loop warning systems in epilepsy. Epilep Behav 37:291–307
Ranjan C, Reddy M, Mustonen M, Paynabar K, Pourak K (2018) Dataset: rare event classification in multivariate time series. arXiv preprint arXiv:1809.10717
Ravindranath M, Candan KS, Sapino ML (2020) M2nn: rare event inference through multi-variate multi-scale attention. In: 2020 IEEE International conference on smart data services (SMDS). IEEE, pp. 53–62
Shahsavari S, McKelvey T, Ritzen CE, Rydenhag B (2011) Cerebrovascular mechanical properties and slow waves of intracranial pressure in TBI patients. IEEE Trans Biomed Eng 58(7):2072–2082
Singh K, Malhotra J (2022) Two-layer LSTM network-based prediction of epileptic seizures using EEG spectral features. Compl Intell Syst. https://doi.org/10.1007/s40747-021-00627-z
Socher R, Ganjoo M, Manning CD, Ng A (2013) Zero-shot learning through cross-modal transfer. In: Advances in neural information processing systems, pp. 935–943
Stacey WC (2018) Seizure prediction is possible-now let’s make it practical. EBioMedicine 27:3–4
Storti SF, Del Felice A, Formaggio E, Boscolo Galazzo I, Bongiovanni LG, Cerini R, Fiaschi A, Manganotti P (2015) Spatial and temporal EEG-FMRI changes during preictal and postictal phases in a patient with posttraumatic epilepsy. Clin EEG Neurosci 46(3):247–252
The new york times coronavirus (covid-19) cases and deaths in the united states. https://data.humdata.org/dataset/nyt-covid-19-data
Tibshirani R (1996) Regression shrinkage and selection via the lasso. J Roy Stat Soc: Ser B (Methodol) 58(1):267–288
Tong Y (2014) Seizures caused by pyridoxine (vitamin b6) deficiency in adults: a case report and literature review. Intract Rae Dis Res 3(2):52–56
Tsiouris KM, Pezoulas VC, Zervakis M, Konitsiotis S, Koutsouris DD, Fotiadis DI (2018) A long short-term memory deep learning network for the prediction of epileptic seizures using EEG signals. Comput Biol Med 99:24–37
Usman SM, Usman M, Fong S (2017) Epileptic seizures prediction using machine learning methods. Comput Math Meth Med. https://doi.org/10.1155/2017/9074759
Usman SM, Khalid S, Aslam MH (2020) Epileptic seizures prediction using deep learning techniques. IEEE Access 8:39998–40007
Valderrama M, Nikolopoulos S, Adam C, Navarro V, Le Van Quyen M (2010) Patient-specific seizure prediction using a multi-feature and multi-modal EEG-ECG classification. In: XII Mediterranean conference on medical and biological engineering and computing 2010. Springer, pp. 77–80
Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser Ł, Polosukhin I (2017) Attention is all you need. In: Advances in neural information processing systems, pp. 5998–6008
Wang H-S, Kuo M (2007) Vitamin b\(\tilde{}\) 6 related epilepsy during childhood. Chang Gung Med J 30(5):396
Xiu Z, Tao C, Gao M, Davis C, Goldstein BA, Henao R (2021) Variational disentanglement for rare event modeling. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, pp. 10469–10477
Xu D, Zhang Z, Shi J (2022) Training data selection by categorical variables for better rare event prediction in multiple products production line. Electronics 11(7):1056
Yu B, Yin H, Zhu Z (2017) Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting. arXiv preprint arXiv:1709.04875
Zhu J, Xia Y, Wu L, He D, Qin T, Zhou W, Li H, Liu T-Y (2020) Incorporating bert into neural machine translation. arXiv preprint arXiv:2002.06823
Acknowledgements
Thanks to Dr Stephen Foldes and Austin Jacobson of Phoenix Children’s, Phoenix, AZ for their support with the work. Results presented in this paper were obtained using the Chameleon testbed supported by the National Science Foundation.
Funding
The work is primarily supported by Department Of Defense Grant No. W81XWH-19-1-0514. Some aspects of the research is also supported by NSF Grant Nos. #1909555, #2026860, #1827757.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Ethical approval
For data collection and analysis, the research is approved by the Phoenix Children’s Institutional Review Board (IRB #19-284).
Informed consent
For retrospective analysis as ours, informed consent was not required by the the Phoenix Children’s Institutional Review Board.
Additional information
Responsible editor: Panagiotis Papapetrou.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Ravindranath, M., Candan, K.S., Sapino, M.L. et al. MMA: metadata supported multi-variate attention for onset detection and prediction. Data Min Knowl Disc (2024). https://doi.org/10.1007/s10618-024-01008-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10618-024-01008-z