Send to

Choose Destination
Chaos. 2014 Jun;24(2):024404. doi: 10.1063/1.4869825.

Symbolic dynamics marker of heart rate variability combined with clinical variables enhance obstructive sleep apnea screening.

Author information

Institute for Technological Development and Innovation in Communications (IDeTIC), Universidad de Las Palmas de Gran Canaria, Las Palmas de G.C. 35017, Spain.
Department of Mathematics. Universidad de Las Palmas de Gran Canaria, Las Palmas de G.C. 35017, Spain.
Pulmonary Medicine Department. Hospital Universitario de Gran Canaria Dr. Negrín, Las Palmas de G.C. 35010, Spain.
Department of Physics, Humboldt-Universität zu Berlin, Berlin 10115, Germany.
Sleep Center, Charité Universitätsmedizin, Berlin 10117, Germany.


Many sleep centres try to perform a reduced portable test in order to decrease the number of overnight polysomnographies that are expensive, time-consuming, and disturbing. With some limitations, heart rate variability (HRV) has been useful in this task. The aim of this investigation was to evaluate if inclusion of symbolic dynamics variables to a logistic regression model integrating clinical and physical variables, can improve the detection of subjects for further polysomnographies. To our knowledge, this is the first contribution that innovates in that strategy. A group of 133 patients has been referred to the sleep center for suspected sleep apnea. Clinical assessment of the patients consisted of a sleep related questionnaire and a physical examination. The clinical variables related to apnea and selected in the statistical model were age (p < 10(-3)), neck circumference (p < 10(-3)), score on a questionnaire scale intended to quantify daytime sleepiness (p < 10(-3)), and intensity of snoring (p < 10(-3)). The validation of this model demonstrated an increase in classification performance when a variable based on non-linear dynamics of HRV (p < 0.01) was used additionally to the other variables. For diagnostic rule based only on clinical and physical variables, the corresponding area under the receiver operating characteristic (ROC) curve was 0.907 (95% confidence interval (CI) = 0.848, 0.967), (sensitivity 87.10% and specificity 80%). For the model including the average of a symbolic dynamic variable, the area under the ROC curve was increased to 0.941 (95% = 0.897, 0.985), (sensitivity 88.71% and specificity 82.86%). In conclusion, symbolic dynamics, coupled with significant clinical and physical variables can help to prioritize polysomnographies in patients with a high probability of apnea. In addition, the processing of the HRV is a well established low cost and robust technique.

[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for American Institute of Physics
Loading ...
Support Center