The impaired inhibition of negative information in patients with major depressive disorder based on event-related potentials and machine learning algorithms.

Major depressive disorder (MDD) may exhibit deficits in the inhibition of negative information. This study investigated whether there are inhibition deficits in MDD and evaluated the utility of event-related potentials (ERPs) as markers for identifying MDD patients. Sixty MDD patients and 60 healthy controls (HC) were recruited, with ERPs being recorded during an emotional Go/Nogo task. ERP signals exhibiting significant inter-group differences were filtered using least absolute shrinkage and selection operator regression. Subsequently, a support vector machine (SVM) classifier model was employed to differentiate between MDD patients and HC. The MDD group exhibited lower accuracy and longer reaction times than the HC group. The MDD group showed lower N2 and P3 amplitudes during negative Nogo trials than the HC group. Analysis of the difference wave indicated that the MDD group had significantly shorter Nd2 latency and lower Pd3 amplitude in the negative Nogo minus neutral Nogo contrast than the HC group. Similarly, in the negative Nogo minus negative Go contrast, the MDD group demonstrated shorter Nd2 latencies and lower amplitudes of both Nd2 and Pd3 components than the HC group. Additionally, the SVM classifier achieved an accuracy of 81.39% in distinguishing MDD patients from healthy controls. Impaired early conflict monitoring and late response inhibition might be the underlying mechanisms contributing to the dysfunction in inhibiting negative information in MDD patients. The ERP signals associated with impaired inhibition may serve as potential neurophysiological markers for young individuals with MDD.