Fig 1.
Patient recruitment flowchart.
PSG: polysomnography; TST: total sleep time.
Table 1.
Demographic and clinical characteristics of the patient groups under study.
Table 2.
Polysomnographic variables of the groups under study: Sleep staging, respiratory related events, and oximetric indexes.
Table 3.
Common pulmonary functional measures of COPD patients derived from post-bronchodilator spirometry.
Fig 2.
Optimization (model selection) of the MLP ANN in the training set.
(A) In-hospital supervised monitoring. (B) At-home unattended monitoring. ICC: intra-class correlation coefficient; NH: number of neurons in the hidden layer; ν: regularization parameter.
Table 4.
Agreement between estimated AHI using the proposed ANN and conventional ODIs with actual AHI from PSG for the groups and settings under study.
Fig 3.
Bland-Altman plots showing agreement between estimated AHI from nocturnal oximetry and actual AHI from PSG.
(A) Supervised oximetry in the laboratory for non-COPD subjects. (B) Supervised oximetry in the laboratory for COPD patients. (C) Unattended oximetry at home for non-COPD subjects. (D) Unattended oximetry at home for COPD patients. AHIOX-LAB: apnea-hypopnea index from in-hospital oximetry; PSG: polysomnography; in-LAB: supervised setting in the hospital; non-COPD: patients without chronic obstructive pulmonary disease; COPD: patients with chronic obstructive pulmonary disease; AHIOX-HOME: apnea-hypopnea index from at-home oximetry; at-HOME: supervised setting at home.
Fig 4.
Mountain plots showing differences between the reference AHI from PSG and the estimated AHI of non-COPD and COPD groups.
(A) Supervised portable oximetry in the hospital simultaneous to PSG. (B) Unattended portable oximetry at home in a different night. AHIOX-LAB: apnea-hypopnea index from in-hospital oximetry; PSG: polysomnography; non-COPD: patients without chronic obstructive pulmonary disease; COPD: patients with chronic obstructive pulmonary disease; AHIOX-HOME: apnea-hypopnea index from at-home oximetry.
Fig 5.
Receiver operating characteristics curves of the estimated AHI.
(A) Supervised portable oximetry in the hospital using a cutoff of AHI ≥15 events/h. (B) Supervised portable oximetry in the hospital using a cutoff of AHI ≥30 events/h. (C) Unattended portable oximetry at home using a cutoff of AHI ≥15 events/h. (D) Unattended portable oximetry at home using a cutoff of AHI ≥30 events/h. AHI: apnea-hypopnea index from standard PSG; non-COPD: patients without chronic obstructive pulmonary disease; COPD: patients with chronic obstructive pulmonary disease; AUC: area under the ROC curve.
Table 5.
Diagnostic performance of the proposed MLPOX-LAB and conventional ODIs from in-laboratory portable oximetry (simultaneously to the PSG) in the non-COPD test group.
Table 6.
Diagnostic performance of the proposed MLPOX-LAB and conventional ODIs from in-laboratory portable oximetry (simultaneously to the PSG) in the COPD test group.
Table 7.
Diagnostic performance of the proposed MLPOX-HOME and conventional ODIs from portable oximetry at home (unattended in a different preceding/consecutive night to PSG) in the non-COPD test group.
Table 8.
Diagnostic performance of the proposed MLPOX-HOME and conventional ODIs from portable oximetry at home (unattended in a different preceding/consecutive night to PSG) in the COPD test group.
Table 9.
Evidences on the effectiveness of portable oximetry monitoring for OSAS detection in patients with COPD in the state-on-the-art and in the present study.