Publications

Objective: Accurate detection of inspiratory onset and offset in the diaphragm electromyographic signal (EMGdi) is clinically relevant to assess patient-ventilator interaction in COPD patients undergoing non-invasive ventilation (NIV). Manual annotations are time-consuming and subject to inter-observer variability, highlighting the need for reliable automatic methods. Method: We developed a fully automatic algorithm to detect EMGdi activity cycles and their onset/offset timing in overnight NIV recordings. Four ECG suppression approaches were combined with root mean square (RMS) and fixed sample entropy (fSE) envelopes, and a novel bias correction strategy based on inspiratory-to-basal signal-to-noise ratio (I2BSNR) was introduced. Performance was compared with double-blind annotations from two independent experts. Results: In a cohort of 10 severe COPD patients (9212 annotated cycles), the best configuration (adaptive filtering with fSE exponential envelope) achieved F $1=0.96$ , with onset bias -28 ms (SD 270 ms) and offset bias + 120 ms (SD 292 ms). We show that fSE-based envelopes consistently outperform RMS in onset/offset detection, and that I2BSNR-based correction reduces systematic bias to within accepted clinical timing windows. Conclusions: The proposed method provides accurate and robust onset/offset detection of EMGdi during NIV in COPD patients. This enables reliable quantification of patient-ventilator asynchronies such as ineffective efforts and delayed cycling, offering direct clinical value for optimizing nightly ventilator settings in severe COPD. Clinical and Impact: Reliable detection of patient inspiratory activity offers a practical tool to guide real-time ventilator adjustments and reduce patient-ventilator asynchronies

JTD Keywords: Annotations, Asynchrony, Chronic obstructive pulmonary disease, Chronic obstructive pulmonary disease (copd), Electromyography, Emg, Filtering, Fixed sample entropy (fse)., Non-invasive ventilation (niv), Patient-ventilator asynchrony (pva), Recording, Reliability, Root mean square, Surface diaphragm electromyography (emgdi), Time, Timing, Ventilation, Ventilators

In this study, we propose a model-based tool for the detection of obstructive apnea episodes by using ECG features from a single lead channel. Several sequences of recurrent apnea were provoked in separate 15-min periods in anesthetized rats during an experimental model of obstructive sleep apnea (OSA). Morphology-based ECG markers and the beat-to-beat interval (RR) were assessed in each sequence. These markers were used to train dynamic Bayesian networks (DBN) with different orders and feature combinations to find a good tradeoff between network complexity and apnea-detection performance. By using a filtering approach, the resulting DBNs were used to infer the apnea probability signal for subsequent episodes in the same rat. These signals were then processed using by 15-s epochs to determine whether epochs were classified as apneic or nonapneic. Our results showed that fifth-order models provided suitable RMSE values, since higher order models become significantly more complex and present worse generalization. A global threshold of 0.2 gave the best overall performance for all combinations tested, with Acc = 81.3%, Se = 69.8% and Sp = 81.5%, using only two parameters including the RR and Ds (R-wave downslope) markers. We concluded that multivariate models using DBNs represent a powerful tool for detecting obstructive apnea episodes in short segments, which may also serve to estimate the number of total events in a given time period.

JTD Keywords: chronic respiratory diseases, obstructive sleep apnea, probabilistic models, Obstructive sleep apnea,probabilistic models,respiratory events,chronic respiratory disease, Respiratory events, Sleep-apnea syndrome,automated detection,oxygen-saturation,classification,recordings,signal

Assessment of the dynamic interactions between cardiovascular signals can provide valuable information that improves the understanding of cardiovascular control. Heart rate variability (HRV) analysis is known to provide information about the autonomic heart rate modulation mechanism. Using the HRV signal, we aimed to obtain parameters for classifying patients with and without chronic heart failure (CHF), and with periodic breathing (PB), non-periodic breathing (nPB), and Cheyne-Stokes respiration (CSR) patterns. An electrocardiogram (ECG) and a respiratory flow signal were recorded in 36 elderly patients: 18 patients with CHF and 18 patients without CHF. According to the clinical criteria, the patients were classified into the follow groups: 19 patients with nPB pattern, 7 with PB pattern, 4 with Cheyne-Stokes respiration (CSR), and 6 non-classified patients (problems with respiratory signal). From the HRV signal, parameters in the time and frequency domain were calculated. Frequency domain parameters were the most discriminant in comparisons of patients with and without CHF: P_Tot (p = 0.02), P_LF (p = 0.022) and fp_HF (p = 0.021). For the comparison of the nPB vs. CSR patients groups, the best parameters were RMSSD (p = 0.028) and SDSD (p = 0.028). Therefore, the parameters appear to be suitable for enhanced diagnosis of decompensated CHF patients and the possibility of developed periodic breathing and a CSR pattern.

JTD Keywords: cardiovascular system, diseases, electrocardiography, frequency-domain analysis, geriatrics, medical signal processing, patient diagnosis, pneumodynamics, signal classification, Cheyne-Stokes respiration patterns, ECG, autonomic heart rate modulation mechanism, cardiovascular control, cardiovascular signals, chronic heart failure, decompensated CHF patients, dynamic interaction assessment, elderly patients, electrocardiogram, enhanced diagnosis, frequency domain parameters, heart rate variability analysis, patient classification, periodic breathing, respiratory flow signal recording, Electrocardiography, Frequency modulation, Frequency-domain analysis, Heart rate variability, Senior citizens, Standards