Introduction: Delivering synchronous assist during non-invasive ventilation (NIV) is challenging with flow- or pressure-controlled ventilators, especially in patients with chronic obstructive pulmonary disease (COPD). Neurally adjusted ventilatory assist (NAVA) uses diaphragm electrical activity (EAdi) to control the ventilator. We evaluated patient-ventilator interaction in patients with COPD during NIV with pressure support ventilation (PSV) and NAVA using a recently introduced automated analysis. Methods: Twelve COPD patients underwent three 30-minute trials: 1) PSV with dedicated NIV ventilator (NIV-PSVVision), 2) PSV with intensive care unit (ICU) ventilator (NIV-PSVServo-I), and 3) with NIV-NAVA. EAdi, flow, and airway pressure were recorded. Patient-ventilator interaction was evaluated by comparing airway pressure and EAdi waveforms with automated computer algorithms. The NeuroSync index was calculated as the percentage of timing errors between airway pressure and EAdi. Results: The NeuroSync index was higher (larger error) for NIV-PSVVision (24 (IQR 15 to 30) %) and NIV-PSVServo-I (21 (IQR 15 to 26) %) compared to NIV-NAVA (5 (IQR 4 to 7) %; P < 0.001). Wasted efforts, trigger delays and cycling-off errors were less with NAVA (P < 0.05 for all). The NeuroSync index and the number of wasted efforts were strongly correlated (r2 = 0.84), with a drastic increase in wasted efforts after timing errors reach 20%. Conclusions: In COPD patients, non-invasive NAVA improves patient-ventilator interaction compared to PSV, delivered either by a dedicated or ICU ventilator. The automated analysis of patient-ventilator interaction allowed for an objective detection of patient-ventilator interaction during NIV. In addition, we found that progressive mismatch between neural effort and pneumatic timing is associated with wasted efforts.