Por favor, use este identificador para citar o enlazar este ítem:http://uvadoc.uva.es/handle/10324/21715
Título
Exploring the Spectral Information of Airflow Recordings to Help in Pediatric Obstructive Sleep Apnea-Hypopnea Syndrome Diagnosis
Autor
Año del Documento
2014
Editorial
IEEE Conference Publications
Descripción
Producción Científica
Documento Fuente
Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conference, 2014, v. 2014, p. 2298-301
Resumen
This work aims at studying the usefulness of
the spectral information contained in airflow (AF) recordings
in the context of Obstructive Sleep Apnea-Hypopnea
Syndrome (OSAHS) in children. To achieve this goal, we
defined two spectral bands of interest related to the
occurrence of apneas and hypopneas. We characterized these
bands by extracting six common spectral features from each
one. Two out of the 12 features reached higher diagnostic
ability than the 3% oxygen desaturation index (ODI3), a
clinical parameter commonly used as screener for OSAHS.
Additionally, the stepwise logistic regression (SLR) featureselection
algorithm showed that the information contained in
the two bands was complementary, both between them and
with ODI3. Finally, the logistic regression method involving
spectral features from the two bands, as well as ODI3,
achieved high diagnostic performance after a bootstrap
validation procedure (84.6±9.6 sensitivity, 87.2±9.1
specificity, 85.8±5.2 accuracy, and 0.969±0.03 area under
ROC curve). These results suggest that the spectral
information from AF is helpful to detect OSAHS in children
Materias (normalizadas)
Sleep Apnea-Hypopnea Syndrome Diagnosis
ISSN
1557-170X)
Revisión por pares
SI
Patrocinador
Ministerio de Economía y Competitividad (TEC2011-22987)
Junta de Castilla y León (VA059U13)
Junta de Castilla y León (VA059U13)
Version del Editor
Idioma
eng
Derechos
openAccess
Aparece en las colecciones
Ficheros en el ítem
La licencia del ítem se describe como Attribution-NonCommercial-NoDerivatives 4.0 International