Prediction of Falls in Community-Dwelling Adults Using Wearable Devices: A Systematic Review

Background

Unintentional falls in older adults are a major public health concern. Accurate fall-risk identification is essential for prevention and for preserving quality of life. This systematic review examined the use of wearable devices for fall-risk prediction in community-dwelling adults aged 65 years and older.

Methods

We conducted a PRISMA-compliant systematic review, searching PubMed, Scopus, and Web of Science for studies published between 2015 and 2026 on fall-risk prediction in community-dwelling older adults using wearable devices. Findings were narratively synthesized. Risk of bias was assessed using PROBAST. PROSPERO: CRD420251275796.

Results

Of 1,559 records, 12 observational studies conducted across 10 countries met the inclusion criteria. Studies used a range of wearable technologies, including smartphones, wearable IMUs, instrumented insoles, and other sensor-based systems, placed at different body sites. Half of the studies used data from daily-life activities, while the remainder were conducted in clinical settings. Fall-risk parameters included postural balance, gait dynamics, functional transitions, turning, and environmental mobility patterns derived from sensor or GPS data. Most studies were at high risk of bias, mainly because of limited external validation. However, some multidimensional models reported promising predictive performance, with values above 80% for sensitivity, specificity, or F1-score.

Conclusions

Wearable-device-based multidimensional models show promise for fall-risk screening in community-dwelling older adults. However, their clinical applicability remains limited by insufficient external validation and heterogeneity across studies. Further validation is needed before these approaches can be implemented as accessible screening tools in community and primary care settings.

Key messages

• Comprehensive analysis of gait, balance, and sensorimotor function within a wearable device modeling framework improves prediction accuracy.

• Predicting fall risk with wearable devices could be a more convenient and practical screening tool as an alternative to traditional methods.