Facial palsy (FP) considerably reduces patients’ quality of life, making accurate severity assessment essential for personalized treatment. EMG-based biofeedback has shown potential in enhancing recovery outcomes. This prospective study aimed to identify EMG time series features that can both classify FP severity and inform biofeedback. Surface EMG was recorded from FP patients and healthy controls during three facial movements. Repeated-measures ANOVAs (rmANOVA) examined the effects of MOTION (movement/rest), SIDE (healthy/affected), and House–Brackmann (HB) score across 20 EMG parameters. Correlations between HB and EMG asymmetry indices were calculated, and Fisher scores assessed the relevance of features in distinguishing HB levels. A total of 55 participants (51.2 ± 14.73 years; 35 female) were included. RmANOVAs revealed a highly significant effect of MOTION across nearly all movement types (p < 0.001). Combining rmANOVA, correlation, and Fisher score analyses, at least 5 of 20 EMG parameters emerged as robust indicators for evaluating paresis severity and guiding biofeedback. These findings demonstrate that sEMG can reliably quantify FP severity and inform biofeedback interventions, even in severe cases, supporting its integration into personalized rehabilitation strategies, though further research is needed to optimize recovery outcomes.
