Assessment of instantaneous heartbeat dynamics in amnestic mild cognitive impairment

In this study, we employ a time-varying, probabilistic model of linear and nonlinear heartbeat dynamics to investigate the possibility of detecting subtle autonomic alterations in subjects suffering from amnestic mild cognitive impairment (aMCI) by exploiting heartbeat information alone. aMCI is a frequent form of cognitive dysfunction which increases the risk of culminating in Alzheimer's disease (AD)-related dementia, and previous studies have demonstrated that AD is accompanied by alterations in autonomic function, which in turn have been linked to cognitive performance in non-demented subjects. We compare 13 MCI patients without ouvert dysautonomia to 13 age- and gender-matched healthy controls by feeding an autonomic nervous system-related linear and nonlinear feature set into a classification framework. Our results show a satisfactory classification performance (73% balanced accuracy), which dropped to 65% when excluding cardiovascular nonlinear/complex features. This outcome confirms the presence of subtle autonomic dysfunction in aMCI (a possible prodromal condition to AD), which can only be detected through to the use of our comprehensive modeling strategy which comprises time-varying, nonlinear/complex estimates of heartbeat dynamics.