The use of transfer learning to take estimation of running ground reaction forces via wearables from the treadmill to overground

Joshua Carter; Xi Chen; Dario Cazzola; Grant Trewartha; Ezio Preatoni

The use of transfer learning to take estimation of running ground reaction forces via wearables from the treadmill to overground

Joshua Carter, Xi Chen, Dario Cazzola, Grant Trewartha, Ezio Preatoni

Research output: Contribution to conference › Abstract › peer-review

25 Downloads (Pure)

Abstract

This study found that a transfer learning approach using ground reaction force (GRF) data from treadmill running helped improve GRF estimation accuracy of a deep learning model during overground running. Leveraging a treadmill dataset 70 times larger than the overground dataset enabled the model to learn generalisable features that could be adapted to the target use case, resulting in more accurate overground
GRF predictions than training on the smaller dataset alone.

Original language	English
Number of pages	1
Publication status	E-pub ahead of print - 2 Apr 2025
Event	Congress of the International Society of Biomechanics (ISB) - Stockholm, Sweden Duration: 27 Jul 2025 → 31 Jul 2025 Conference number: 30 https://isb2025.com/

Conference

Conference	Congress of the International Society of Biomechanics (ISB)
Country/Territory	Sweden
City	Stockholm
Period	27/07/25 → 31/07/25
Internet address	https://isb2025.com/

Access to Document

Carter(2025)_ISB2025Abstract_publishedFinal published version, 130 KBLicence: CC BY

Cite this

@conference{d5ce785a34414ac1afe8a635136147a3,

title = "The use of transfer learning to take estimation of running ground reaction forces via wearables from the treadmill to overground",

abstract = "This study found that a transfer learning approach using ground reaction force (GRF) data from treadmill running helped improve GRF estimation accuracy of a deep learning model during overground running. Leveraging a treadmill dataset 70 times larger than the overground dataset enabled the model to learn generalisable features that could be adapted to the target use case, resulting in more accurate overground GRF predictions than training on the smaller dataset alone.",

author = "Joshua Carter and Xi Chen and Dario Cazzola and Grant Trewartha and Ezio Preatoni",

year = "2025",

month = apr,

day = "2",

language = "English",

note = "Congress of the International Society of Biomechanics (ISB) ; Conference date: 27-07-2025 Through 31-07-2025",

url = "https://isb2025.com/",

}

TY - CONF

T1 - The use of transfer learning to take estimation of running ground reaction forces via wearables from the treadmill to overground

AU - Carter, Joshua

AU - Chen, Xi

AU - Cazzola, Dario

AU - Trewartha, Grant

AU - Preatoni, Ezio

N1 - Conference code: 30

PY - 2025/4/2

Y1 - 2025/4/2

N2 - This study found that a transfer learning approach using ground reaction force (GRF) data from treadmill running helped improve GRF estimation accuracy of a deep learning model during overground running. Leveraging a treadmill dataset 70 times larger than the overground dataset enabled the model to learn generalisable features that could be adapted to the target use case, resulting in more accurate overground GRF predictions than training on the smaller dataset alone.

AB - This study found that a transfer learning approach using ground reaction force (GRF) data from treadmill running helped improve GRF estimation accuracy of a deep learning model during overground running. Leveraging a treadmill dataset 70 times larger than the overground dataset enabled the model to learn generalisable features that could be adapted to the target use case, resulting in more accurate overground GRF predictions than training on the smaller dataset alone.

M3 - Abstract

T2 - Congress of the International Society of Biomechanics (ISB)

Y2 - 27 July 2025 through 31 July 2025

ER -

The use of transfer learning to take estimation of running ground reaction forces via wearables from the treadmill to overground

Abstract

Conference

Access to Document

Fingerprint

Cite this