# Assessment of Frailty in Community-Dwelling Older Adults Using Smartphone-Based Digital Lifelogging: A Multi-Center, Prospective Observational Study

## TL;DR

Smartphone data like gait speed and step count can better predict frailty in older adults than traditional methods, offering a practical way to monitor health in daily life.

## Contribution

Smartphone-based lifelogs significantly enhance frailty assessment by capturing real-world mobility and health metrics beyond clinical indicators.

## Key findings

- Digital lifelogs explained 18.3% more variance in frailty than traditional clinical indicators.
- Gait speed and daily step counts were key predictors of frailty in community-dwelling older adults.
- A model using only digital lifelogs explained 28.8% of the variance in frailty.

## Abstract

What are the main findings?

Smartphone-based digital lifelogs, specifically usual gait speed, daily step count, and subjective health, are significantly associated with frailty and explain substantially more variance than traditional clinical indicators alone.

Continuous, real-world mobility metrics collected via embedded smartphone sensors provide meaningful insights into functional decline that are not captured by conventional clinic-based frailty assessments.

What are the implications of the main findings?

Smartphone-based monitoring offers a scalable, low-burden approach to community-based frailty assessment and monitoring, with potential to support future longitudinal risk prediction after validation.

Integrating digital lifelog data into geriatric care pathways can enable proactive intervention, personalized management, and more accurate frailty risk stratification in everyday living environments.

Frailty in older adults is a multidimensional syndrome characterized by reduced physiological resilience and heightened vulnerability to adverse outcomes, yet conventional assessments remain largely clinic-based. This study evaluated the feasibility and explanatory utility of smartphone-based digital lifelogs for assessing frailty in community-dwelling older adults. In a prospective observational study, 300 participants (mean age 73.30, SD 5.37 years) from three sites in Seoul, South Korea, used a custom mobile application for two weeks that passively collected sensor-derived gait speed, 30 s sit-to-stand counts, and daily and hourly step counts, alongside self-reported ratings of perceived exertion and subjective health. Frailty Index (FI) scores were computed, and Pearson correlations, hierarchical linear regression, and independent linear regression were applied to examine associations and model explanatory performance. Significant correlations were observed between FI and gait speed, sit-to-stand performance, daily step counts, perceived exertion, and subjective health. Incorporating digital lifelogs significantly improved explained variance in frailty beyond clinical indicators (ΔR2 = 0.183), with gait speed and daily step counts emerging as key predictors. A model including only digital lifelogs also significantly associated with frailty (R2 = 0.288). These findings suggest that smartphone-based lifelogging offers a feasible, practical, and informative method for two-week monitoring and cross-sectional assessment in community settings.