FREE-LIVING HIP ACCELEROMETRY DETECTS AND FORECASTS FRAILTY DECLINE IN A SAMPLE OF COMMUNITY-DWELLING OLDER ADULTS

Sep 28, 2026

Frailty is a geriatric syndrome marked by reduced strength, exhaustion, slow gait, weight loss, and low activity, which predicts adverse health outcomes. This retrospective study uses 7 days of free-living hip accelerometer data to detect the frailty phenotype concurrently and to forecast frailty worsening after 1 year among community-dwelling older adults. Researchers extracted 115 accelerometry features (activity, sleep, harmonic variables) and trained five machine learning models with demographic and health variables. Best models achieved cross-sectional F1=0.84 and AUROC=0.84 and longitudinal F1=0.87 and AUROC=0.84, supporting wearable real-time screening and clinical monitoring.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages1

Reading Time3 min

AccessFree to download

Abstract

Methods

Machine Learning Models and Results

Implications

FAQs

How does the study detect and forecast frailty?

It uses 7 days of free-living hip accelerometer data to detect current frailty phenotype and to forecast whether frailty worsens over the following year.

What data features were extracted from the hip accelerometer readings?

The study extracted 115 accelerometry features covering activity, sleep, and harmonic variables using open-source code.

What performance did the best models achieve?

The best cross-sectional model reached F1=0.84 and AUROC=0.84, and the best longitudinal model reached F1=0.87 and AUROC=0.84.

Sep 28, 2026

Uploaded by Well Done