Sola-Soler J, Jane R, Fiz J A, Morera J
Dept. ESAII, Centre de Recerca en Enginyeria Biomèdica, Universitat Politècnica de Catalunya, Barcelona, Spain.
Conf Proc IEEE Eng Med Biol Soc. 2005;2005:2583-6. doi: 10.1109/IEMBS.2005.1616997.
Several studies have shown differences in spectral parameters between simple snorers and patients with Obstructive Sleep Apnea Syndrome (OSAS). Most of them analyzed a reduced number of snores and/or only post-apneic snores were selected in OSAS patients. Previous findings suggest that snore parameters have a greater variability as the severity of OSAS increases. In this work we propose to analyze the snoring variability through the magnitude of the first difference of snore parameters. The technique is applied to long time sound recordings from 9 simple snorers (15795 snores) and 15 OSAS patients (19263 snores). The snores are characterized by their sound intensity an by a set of spectral parameters The variability of snore parameters is well correlated to OSAS severity (r>0.7) and is significantly greater in OSAS patients than in simple snorers (p<0.005). The results are similar when post-apneic snores are excluded from the analysis. Snoring subjects are classified with a logistic regression model, which is validated with the live-one-out method. The model is adjusted to correctly classify 100% of OSAS patients for screening purposes previously to enroll for a whole polysomnographic study.
多项研究表明,单纯打鼾者与阻塞性睡眠呼吸暂停综合征(OSAS)患者的频谱参数存在差异。其中大多数研究分析的打鼾样本数量较少,和/或在OSAS患者中仅选择了呼吸暂停后的打鼾样本。先前的研究结果表明,随着OSAS严重程度的增加,打鼾参数的变异性更大。在这项研究中,我们建议通过打鼾参数一阶差分的幅度来分析打鼾的变异性。该技术应用于9名单纯打鼾者(15795次打鼾)和15名OSAS患者(19263次打鼾)的长时间声音记录。打鼾的特征在于其声音强度和一组频谱参数。打鼾参数的变异性与OSAS严重程度密切相关(r>0.7),且OSAS患者的变异性明显大于单纯打鼾者(p<0.005)。排除呼吸暂停后打鼾样本进行分析时,结果相似。使用逻辑回归模型对打鼾受试者进行分类,并采用留一法进行验证。该模型经过调整,在进行全面多导睡眠图研究之前,用于筛查目的时可正确分类100%的OSAS患者。