Smart Ring in Clinical Medicine: A Systematic Review (Preprint)

<sec> <title>BACKGROUND</title> Smart rings enable continuous physiological monitoring through finger-worn sensors. Despite growing consumer adoption exceeding 2 million users globally, their clinical utility beyond sleep tracking remains unclear. </sec> <sec> <title>OBJECTIVE</title> To systematically review evidence for smart ring applications in clinical medicine, assess measurement accuracy, and evaluate clinical outcomes across medical domains. </sec> <sec> <title>METHODS</title> We searched PubMed/MEDLINE, Embase, and Cochrane Library through July 31, 2025, for studies using smart rings in clinical applications. Inclusion criteria: adults ≥18 years, commercially available smart rings, health-related outcomes, and English language. We excluded pure validation studies, device development papers, and pediatric populations. Two reviewers independently screened studies and extracted data. Risk of bias was assessed using ROBINS-I and RoB 2.0. </sec> <sec> <title>RESULTS</title> From 862 citations, 107 studies met inclusion criteria (85 non-randomized studies, 14 case reports/series, 8 randomized trials) including approximately 100,000 participants. Studies were equally distributed between sleep (n=51, 47.7%) and non-sleep applications (n=56, 52.3%). The Oura Ring dominated research (77 studies, 72.0%). Smart rings demonstrated high accuracy: heart rate r=0.996 versus electrocardiography, heart rate variability (HRV) r=0.980, sleep detection 93-96% sensitivity. Non-sleep applications have demonstrated predictive capabilities across various fields: COVID-19 detected 2.75 days before symptoms (82% sensitivity, 63% specificity, n=63,153); inflammatory bowel disease flares predicted 7 weeks early (72% accuracy); bipolar episodes detected 3-7 days early (79% sensitivity); postoperative pain predicted with 70% accuracy (AUC 0.762). Gastric cancer patients showed 86% HRV reduction versus controls and HRV correlated with tumor size, tumor infiltration, lymph node metastasis and distant metastasis. However, 65% of studies had moderate-to-high risk of bias. Critical limitations included small samples (median 47-89 excluding outliers), proprietary algorithms preventing reproducibility (89% of studies), poor diversity reporting (only 35% reported race/ethnicity), and declining adherence (80% at 3 months to 43% at 12 months). </sec> <sec> <title>CONCLUSIONS</title> Smart rings have evolved beyond sleep tracking, with equal sleep/non-sleep applications confirming transformation into clinical tools capable of early disease detection through autonomic monitoring. Evidence supports use in sleep disorders, mental health monitoring, and chronic disease management. However, single-manufacturer predominance, algorithmic opacity, population homogeneity, and adherence challenges require attention before widespread clinical implementation. </sec>