Atrial cardiomyopathy phenotyping requires integration of electrical and mechanical substrate information, yet standard left atrial (LA) strain parameters show weak associations with invasively measured electrical properties. We hypothesized that dynamic features extracted from LA strain waveforms via hidden Markov Model (HMM) state decomposition and autoregressive (AR) spectral analysis capture substrate information that conventional metrics miss. In a pilot study of 13 patients with paroxysmal atrial fibrillation, we evaluated pre-procedural speckle tracking echocardiography (STE) against intra-procedural high-density electroanatomical mapping (EAM) and surface ECG. HMM-derived diastasis-to-contractile transition probability correlated with normalized total LA activation time (ρ=−0.68, p=0.010) and P-wave width (ρ=−0.72, p=0.005). AR spectral entropy correlated with both normalized LA activation time (ρ=0.63, p=0.022) and P-wave width (ρ=0.68, p=0.01). For activation timing, conventional strain amplitudes showed no significant associations (|ρ|≤0.42). Global LA strain waveforms contain recoverable electrical substrate information, motivating the development of software tools for non-invasive atrial substrate characterization.