Non-Stationary Autonomic Collapse and Afferent Cortico-Cardiac Decoupling During Human Perimortem Transition

Abstract Purpose Characterizing the physiological breakdown of the brain–heart axis during terminal dying transitions has been severely hindered by the extreme non-stationarity of agonal rhythms, prolonged pauses, and volume-conducted cardiac field artifacts in electroencephalography (EEG). Methods From a screening of 29 post-cardiac arrest comatose patients undergoing secondary terminal events in the I-CARE multicenter database, an expert neurophysiological quality audit was conducted. Twenty-three recordings were excluded due to CPR compressions, electrode detachment, saturation artifacts, or non-continuous EEG-ECG channels, isolating a pristine cohort of N = 6 patients monitored continuously to electromechanical cardiac arrest (T0). To prevent mathematical distortion across agonal pauses, we manually verified the continuous ECG series beat-to-beat to construct curated normal-to-normal (NN) series. Empirical Mode Decomposition (EMD) and the Hilbert–Huang Transform (HHT) resolved instantaneous modal energy across intrinsic modes (IMF1–3). Ascending visceral-cortical signaling was evaluated via Heartbeat-Evoked Potentials (HER) strictly within the canonical 200–500 ms post-R window and reference-independent cortical Global Field Power (GFP). Results Perimortem cardiac dynamics exhibited a multi-logarithmic collapse in instantaneous Hilbert energy (IA² < 10²) and spectral entropy. Instantaneous sympathovagal balance (HHT LF/HF = IMF2/IMF1) diverged into two diametrically opposed terminal profiles: sympathovagal failure with severe escape bradycardia (0.21–0.29) versus agonal hyper-adrenergic surges (2.04). Concurrently, HER waveforms across all scalp derivations demonstrated complete suppression (-0.2 to + 0.1 µV), with cortical GFP flatlining at baseline noise levels (~ 1.0–1.2 µV) devoid of phase-locked synchronization. Conclusion Functional ascending baroreceptor-cortical communication is extinguished well before peripheral electromechanical cardiac arrest. These findings reveal an asynchronous, hierarchical uncoupling of the human brain–heart axis during the dying process, demonstrating that manual data curation coupled with data-driven HHT analytics effectively overcomes the pitfalls of automated spectral HRV in critical agonal states.