A Pilot Study Using a Smartwatch to Search for Biomarkers of Heart Rate Variability in Active Ulcerative Colitis

Abstract

Introduction: Autonomic imbalance has been reported to correlate with clinical remission in patients with ulcerative colitis (UC). This study evaluated heart rate variability (HRV), a potential digital biomarker, in patients with active UC using a smartwatch that is easy to handle.

Methods: Patients with active UC were recruited for this prospective study. The patients' HRV was measured via the Fitbit Inspire2^™ linked via Bluetooth to their smartphone. HRV during nighttime sleep was obtained from continuous data. Patients were required to input the Simple Clinical Colitis Activity Index (SCCAI) score once daily by the application on their smartphones for 3 months.

Results: Nine patients with UC were included. In clinically active disease, SCCAI scores showed a weak inverse relationship with parasympathetic activity, differences of successive R-R pulse intervals (RMSSD) (r = -0.44, p < 0.0001), high frequency (HF) (r = -0.42, p < 0.0001), and total autonomic nervous activity, low frequency (LF) (r = -0.43, p < 0.0001). Receiver operating characteristic analysis indicated that the RMSSD, HF, and LF were significantly higher in patients with active UC. Meanwhile, LF showed the most correlation with severity for bowel urgency scores.

Conclusion: Longitudinal nighttime HRV recorded using a smartwatch is associated with disease activity in patients with active UC. In particular, RMSSD and HF, which are indices of the parasympathetic nervous system, have been suggested as potential digital biomarkers for UC.

Keywords: Autonomic nervous system; Heart rate variability; Inflammatory bowel disease; Ulcerative colitis; Wearable smart watch.

Fig. 1.

Correlation between heart rate variability and SCCAI in clinically active disease in UC patients. In clinically active disease (187 points), SCCAI scores showed a weak inverse relationship with RMSSD (r = −0.44, p < 0.0001), HF (r = −0.42, p < 0.0001), and LF (r = −0.43, p < 0.0001). There was no correlation between the LF/HF ratio and SCCAI (r = −0.04, p < 0.57). Statistical analyses were performed using Spearman’s rank correlation coefficients. SCCAI, Simple Clinical Colitis Activity Index; RMSSD, root mean square of successive differences of RR intervals; LF, low-frequency; HF, high frequency.

Fig. 2.

Correlation between heart rate variability and SCCAI in the clinical remission of UC patients. In the clinical remission phase (552 points), all HRV indices showed no correlation with the SCCAI (RMSSD [r = 0.14, p < 0.0005], HF [r = 0.07, p < 0.12], LF [r = 0.08, p < 0.05], and LF/HF ratio [r = −0.09, p < 0.04]). Statistical analyses were performed using Spearman’s rank correlation coefficients. SCCAI, Simple Clinical Colitis Activity Index; HRV, heart rate variability; RMSSD, root mean square of successive differences of RR intervals; LF, low frequency; HF, high frequency.

Fig. 3.

Comparison of heart rate variability between clinically active disease and remission of UC patients. a RMSSD, and LF/HF ratio were significantly lower in the active phase than in the remission phase; however, HF and LF were not different between these two phases. *p < 0.05, Mann-Whitney U test. b Results from the ROC analyses for each HRV index, with the corresponding AUCs. RMSSD, Root Mean Square of Successive Differences of RR intervals; LF, low frequency; HF, high frequency.

Fig. 4.

Comparison of heart rate variability between moderate-severe and mild disease in UC patients. a RMSSD, HF, and LF were significantly higher in moderate-severe than in mild disease. ****p < 0.0001; ns, not significant by Mann-Whitney U test. b Results from the ROC analyses for each HRV index, with the corresponding AUCs. RMSSD showed an AUC value of 0.73, cut-off value of 24.3, sensitivity of 63.4%, and specificity of 81.0%. HF also showed an AUC value of 0.72, cut-off value of 154.9, specificity of 66.2%, and sensitivity of 81.0%. LF also showed an AUC value of 0.75, cut-off value of 335.8, specificity of 66.2%, and sensitivity of 83.5%, which indicated moderate accuracy. HRV, heart rate variability; RMSSD, root mean square of successive differences of RR intervals; LF, low frequency; HF, high frequency.

Fig. 5.

Comparison of BU and heart rate variability. a LF were significantly higher in the case of BU positive (1–3) than in the BU negative (0). *p < 0.05; not significant by Mann-Whitney U test. b LF and HF were significantly higher in the case of moderate-to-higher BU (2–3) than in mild BU (0–1). **p < 0.01; ***p < 0.001; ns, not significant by Mann-Whitney U test. c The results from the ROC analyses for LF showed an AUC value of 0.75, cut-off value of 694.5, sensitivity of 83.3%, and specificity of 63.3% and HF showed an AUC value of 0.70, cut-off value of 551.0, sensitivity of 50.0%, and specificity of 87.7%. BU, bowel urgency; HRV, heart rate variability; RMSSD, root mean square of successive differences of RR intervals; LF, low frequency band; HF, high frequency band; ns, not significant.