This study examined whether adding an objective physiological measure like heart-rate variability (HRV), which reflects autonomic nervous system balance, could enhance standard psychometric scales and improve machine-learning predictions of panic disorder (PD) symptom severity. Researchers analyzed data from over 500 adults with PD from Gangnam Severance Hospital, comparing models trained on three input types: (1) both psychometric scales and HRV together (ExSH), (2) only the scales (ExS), and (3) only the HRV components across nine machine-learning models. Short five-minute ECG recordings were processed to extract HRV features alongside psychometric scale data commonly used to assess anxiety and depression. Across the machine-learning models, those integrating both HRV and psychometric data performed best, with the Random Forest model for ExSH showing the highest sensitivity and f1 (measure of overall accuracy) score compared to ExS, closely followed by CatBoost. While HRV alone underperformed compared to scales, combining physiological (HRV) and psychometric measures seemed to provide the most accurate predictions for PD symptom severity. The study suggests that multimodal approaches linking more objective physiological measures with psychometric scales can meaningfully improve the evaluation of anxiety disorders, though further research is needed to refine feature selection to reduce noise, account for comorbidities, and expand the classification and measurement categories to include more factors.