Our understanding of epilepsy mechanisms has shifted its focus towards a dynamic, whole-brain network perspective. Several factors, such as circadian and multi-day cycles, are assumed to influence the likelihood of seizure occurrence. Moreover, this seizure susceptibility is critically dependent on anti-epileptic drug dosage. In this study, we aim to uncover changes in inter-regional communication that are associated with a systematic increase in seizure susceptibility. Uncovering this relationship is important to understand the mechanisms of seizure generation. Method: We investigated long-term, intracranial EEG recordings of 9 drug-refractory epileptic patients. In the context of their presurgical evaluation, anti-epileptic drugs were continuously decreased until seizures occurred (typically between fourth and sixth day). We quantified brain network changes using Directed Transfer Function in the interval between admission and first seizure. Furthermore, each measurement was characterized by various factors, such as time, drug dose, and interictal epileptic spike rate. Finally, to track connectivity changes in time, we projected them into low-dimensional space using Principal Component Analysis. Result: Connectivity changes were consistently linked to time (Pearson’s r range 0.22-0.82), drug dosage (0.24-0.79), interictal spikes and high-frequency oscillations (0.25-0.94), and circadian rhythm (0.21-0.74). These correlations were significant in 8/9 subjects (p<0.05, FDR corrected). When analyzing connectivity dynamics, we observed time points linked with seizure buildup. Their timing was patient-specific and varied between one and three days before seizure. Conclusion: Seizure susceptibility influences inter-regional communication. Interictal spikes and high-frequency oscillations are the most prominent factor. Connectivity investigations can unveil patient-specific critical time points during seizure buildup.