Smartwatch-Derived Nocturnal Scratching Metrics Capture Disease Activity and Severity in Pediatric Atopic Dermatitis.

Background/

Objectives

The itch-scratch cycle is a key driver of exacerbation in atopic dermatitis (AD) and requires objective monitoring, yet patient-reported itch scores are often unreliable in children. This study aimed to evaluate smartwatch-derived nocturnal scratching metrics as digital biomarkers of disease activity and treatment response in pediatric AD.

Methods

In this prospective observational study, 50 children (median age 9 years) with physician-diagnosed AD wore an Apple Watch with the Itch Tracker application for 5-14 nights during initiation of topical therapy. Three scratch metrics-scratch count rate (SCR), scratch duration ratio (SDR), and scratch burden index (SBI, duration × intensity)-were analyzed. Associations with clinical outcomes [Eczema Area and Severity Index (EASI), Patient-Oriented Eczema Measure (POEM)], serum thymus and activation-regulated chemokine (TARC), and itch numerical rating scale (NRS) were examined. Logistic regression models were evaluated to examine whether these metrics could identify children who achieved clinically meaningful improvement, defined as EASI-50 plus ≥ 4-point POEM reduction.

Results

All scratch metrics correlated with baseline EASI (r = 0.60-0.64, p < 0.001) and serum TARC (r = 0.58-0.60, p < 0.001). Reductions in scratching paralleled clinical improvement (r = 0.67-0.71, p < 0.0001). Among models, the SBI-based logistic regression demonstrated the best discriminative performance (AUC = 0.78, 95% CI: 0.64-0.92).

Conclusions

Wearable-derived nocturnal scratching metrics showed moderate but consistent associations with disease severity and short-term improvement. Although predictive capability remains to be established, these metrics may serve as treatment-responsive digital measures. Given the cross-sectional nature of biomarker analyses and other study limitations, further prospective validation is required before clinical application in real-world pediatric AD monitoring.