The Impact of the Accelerometer Sampling Rate on the Performance of Machine and Deep Learning Models in Wearable Fall-Detection Systems

Manny Villa, Eduardo Casilari · 2025

Read The Impact of the Accelerometer Sampling Rate on the Performance of Machine and Deep Learning Models in Wearable Fall-Detection Systems by Manny Villa, Eduardo Casilari (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12788221.txt.

Description

Manny Villa and Eduardo Casilari tune the sampling clock of wearable fall detectors in a 2025 study. A fall sensor that samples fast sees more detail but drains batteries and floods memory; one that samples slowly risks missing the fall's shape. The sweet spot for machine learning accuracy had not been pinned down.

Five algorithms, from deep hybrids to classic neighbors and margins, trained on a public fall dataset at four sampling rates from ten to one hundred samples per second. Validation widened to real-fall archives and a week of continuous wear. Deep models beat classic ones throughout. A hybrid convolutional-recurrent model at twenty samples per second struck the best balance, near ninety-nine percent accuracy with strong sensitivity and specificity, stable across every check. Rates near ten to twenty kept the dynamics while shrinking data and energy.

Catalog record

CREATORManny Villa, Eduardo Casilari
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC12788221

Browse the full library · About the Tower of Babel