Background: Neural respiratory drive (NRD) using diaphragm electromyography through an invasive transesophageal multi?electrode catheter can be used as a feasible clinical physiological parameter in patients with chronic obstructive pulmonary disease (COPD) to provide useful information on the treatment response. However, it remains unknown whether the surface diaphragm electromyogram (EMGdi) could be used to identify the deterioration of clinical symptoms and to predict the necessity of hospitalization in acute exacerbation of COPD (AECOPD) patients. Methods: COPD patients visiting the outpatient department due to acute exacerbation were enrolled in this study. All patients who were subjected to EMGdi and classical parameters such as spirometry parameters, arterial blood gas analysis, COPD assessment test (CAT) score, and the modified early wing score (MEWS) in outpatient department, would be treated effectively in the outpatient or inpatient settings according to the Global Initiative for Chronic Obstructive Lung Disease guideline. When the acute exacerbation of the patients was managed, all the examination above would be repeated. Results: We compared the relationships of admission?to?discharge changes (Δ) in the normalized value of the EMGdi, including the change of the percentage of maximal EMGdi (ΔEMGdi％max) and the change of the ratio of minute ventilation to the percentage of maximal EMGdi (ΔVE/EMGdi％max) with the changes of classical parameters. There was a significant positive association between ΔEMGdi％max and ΔCAT,ΔPaCO2, and ΔpH. The change (Δ) of EMGdi％max was negatively correlated with ΔPaO2/FiO2 in the course of the treatment of AECOPD. Compared with the classical parameters including forced expiratory volume in 1 s, MEWS, PaO2/FiO2, the EMGdi％max (odds ratio 1.143, 95％ confidence interval 1.004–1.300) has a higher sensitivity when detecting the early exacerbation and enables to predict the admission of hospital in the whole cohort. Conclusions: The changes of surface EMGdi parameters had a direct correlation with classical measures in the whole cohort of AECOPD. The measurement of NRD by surface EMGdi represents a practical physiological biomarker, which may be helpful in detecting patients who should be hospitalized timely.