University of Twente Student Theses

Login

Redesigning a Gait Detection Algorithm Based on Linear Accelerations to Include Gyroscope Data

Groot, Sietse J. de (2023) Redesigning a Gait Detection Algorithm Based on Linear Accelerations to Include Gyroscope Data.

[img] PDF
433kB
Abstract:The algorithm presented in this assignment focuses on identifying walking bouts, which allows for the analysis of gait characteristics by other algorithms. A common technique used for data collection for gait detection, is to extract walking data from an inertial measurement unit (IMU), worn by a patient in an uncontrolled environment. An algorithm was previously created to differentiate between walking and non-walking data, using the linear accelerations of an IMU. The aim of this assignment was to improve the accuracy and reduce uncertainty by integrating gyroscope data into the gait detection algorithm. Multiple approaches were explored to include gyroscope data. A second goal of this assignment was to change the threshold used by the original algorithm to fit a different population, by personalizing the threshold. A personalized threshold, designed by using experimental data of walking and non-walking activities, replaced the generic threshold used by the original algorithm. Experiments were then performed to validate and compare the developed algorithms. Participants were recorded by video and IMU during 5 minutes of free movement. The video data were labeled as "walking" and "non-walking" and used for ground-truth validation. Out of all tested algorithms, the combination algorithm showed the highest accuracy, precision and F1-score.
Item Type:Essay (Bachelor)
Faculty:ET: Engineering Technology
Subject:44 medicine
Programme:Biomedical Technology BSc (56226)
Link to this item:https://purl.utwente.nl/essays/97160
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page