In this paper, we focus on the automatic detection of sleep apnea-hypopnea syndrome (SAHS) from single-channel airflow (AF) recordings. Spectral data from a very low frequency band of AF is used to feed classifiers based on linear discriminant analysis (LDA). These are iteratively obtained through the
(ABM1) algorithm, which combines their performance in order to reach a higher diagnostic ability. We built an ABM1-LDA model, using a training set, which showed generalization ability as well as high diagnostic statistics in an independent test set (94.1% sensitivity, 85.7% specificity, and 92.7% accuracy). These results outperform those from recent studies focused on scoring apneas and hipopneas. Hence, the utility of our approach to assist in SAHS diagnosis is showed.