Toward Contactless HRV: Validation of Radar-Derived Heart Inter-Beat Variability

Heart rate variability (HRV) quantifies autonomic modulation but typically requires contact ECG. Radar-derived heart-sound inter-beat intervals (HS-IBI) could enable contactless HRV assessment. The goal of this study is to evaluate agreement between HS-IBI from a 24-GHz radar heart-sound channel and RR intervals from a Biopac ECG (reference) at rest and immediately post-exercise, and to characterize condition-dependent changes in short-term HRV. Twelve adults completed two supine recordings per visit (Rest and Post-exercise). RR (ECG) and HS-IBI (radar) series were exported to Kubios HRV Scientific for standardized preprocessing and time domain metrics (HR, SDNN, RMSSD, pNN50; plus SD1, SD2). Agreement was summarized using Bland-Altman statistics and Pearson correlation. Data quality was high (median beats corrected 0%; segment lengths: Rest median 77.5 s, Exercise median 51.5 s). Group means (both modalities) showed higher HR during Exercise than Rest (79.1 ± 9.9 vs. 70.8 ± 9.4 bpm). Across conditions, HS-IBI closely matched ECG for key metrics: HR: bias ≈ 0.00 bpm, r = 1.000 (Rest and Exercise); SDNN: bias 0.90 ms (LoA -6.33 to 8.13), r = 0.985 at Rest; bias 0.62 ms (LoA -1.29 to 2.53), r = 1.000 at Exercise. RMSSD: bias 1.46 ms (LoA -11.69 to 14.61), r = 0.981 at Rest; bias 0.94 ms (LoA -2.38 to 4.26), r = 0.999 at Exercise. Withinsubject changes (Exercise - Rest) were similar for both modalities: HR increased by +8.3 bpm on average; SDNN and RMSSD showed small positive mean deltas with wide inter-individual variability; pNN50 increased modestly. Overall, Radar HS-IBI provided HRV metrics that were essentially interchangeable with reference ECG in short supine recordings at rest and early post-exercise recovery, with sub-millisecond to low-millisecond biases and very high correlations. These findings support contactless radar heart-sound sensing as a practical alternative for short-term HRV assessment; future work should examine longer recordings, standardized breathing, and frequency-domain metrics.