Inductive Modeling in Newborn Sleep Stage Recognition. Vaclav Gerla, Miroslav Bursa, Lenka Lhotska, Pavel Kordik, Karel Paul, Vaclav Krajca. IWIM, Prague, 2007.

 Article (in pdf)

Abstract. This paper addresses automated classification of newborn sleep electroencephalogram (EEG) using inductive classification methods. Newborn EEG plays an important role in determining the maturity level of neonatal brain. Polysomnography (PSG) recording can be classified into four important behavioral states: quiet sleep, active (non-quiet) sleep, wakefulness and movement artifact. Infant sleep significantly differs from adult sleep; we therefore apply methods designed for the problem of differentiation between the described states. The proportion of these states is a significant indicator of the maturity of the newborn brain in clinical practice. In this study we use data provided by the Institute for the Care of Mother and Child in Prague (12 newborn polysomnografic signal; similar postconceptional age; all data are scored by an experienced neurologist). Automated classification is performed by inductive models evolution through ant-colony approach (ACO-DTree algorithm) and the GAME (Group of Adaptive Models Evolution) inductive models. The results are compared with standard cross validation method. Using inductive modeling methods produced better results with improved generalization skills of the classifier. The purpose of this study is to facilitate the work of neurologist.

Keywords. Inductive modeling, electroencephalograph, neonatal, sleep stage classification, ant colony ptimization, ACODTree, GAME.

References.

1. M. S. Scher, “Automated EEG-sleep analyses and neonatal neurointensive care”, Sleep Medicine 5, 2004.
2. V. Gerla, L. Lhotská, V. Kraj a, K. Paul: Multichannel Analysis of the Newborn EEG Data. IEEE ITAB 2006.
3. A. Värri, K. Hirvonen, V. Kakkinen, J. Hasan, P. Loula, “Nonlinear Eye Movement Detection Method for Drowsiness Studies”, Int. J. of Biomed. Comp 43, 1996.
4. A. Abraham, C. Grosan, V. Ramos. Swarm Intelligence in Data Mining (Studies in Computational Intelligence), Springer 2006.
5. M. Dorigo, G. D. Caro, L. M. Gambardella: Ant algorithms for discrete optimization Artif. Life, MIT Press, 1999, 5, 137-172.
6. M. Bursa, L. Lhotska: Automated Classification Tree Evolution through Hybrid Metaheuristics. HAIS07 conference, submitted paper.
7. M. S. Scher, S. G. Dokianakis, M. Sun, “Computer classification of sleep in preterm and full-term neonates at similar postconceptional term ages, “ Sleep, Vol. 19, No. 1, Pages 18-25, 1996.
8. G. Hauksson, “Automated Analysis of Newborn EEG,“ Master of Science Thesis, Stockholm, Sweden, XR-EE-SB, 2006.
9. J. E. Heiss, C. M. Deld, P. A. Estevez, C. A. Perez, C. A. Holzmann, j. P. Perez, “Classification of sleep stages in infants: a neuro fuzzy approach, “ Universidad de Chile, Santiago, Chile.
10. L. Teofilo, Lee-Chiong, “SLEEP: a comprehensive hanbook,“ Johm Wiley & Sons, Inc., Hoboken, New Jersey, 2006.
11. P. Kordik. Fully Automated Knowledge Extraction using Group of Adaptive Models Evolution. PhD thesis, Dep. of Comp. Sci. and Computers, FEE, CTU Prague, Czech Republic, September 2006.
12. Ivakhnenko A. G.: Polynomial theory of complex systems. IEEE Transactions on Systems, Man, and Cybernetics, SMC-1(1):364378, 1971.