Navigation

Kevin Rätsch

  • Job title: Bachelor's Thesis
  • Working group: Clustering of physical activity patterns in COPD patients

Advisors
Martin Ullrich, M. Sc.Nils Roth, M. Sc., Dr. Wolfgang Geidl, Prof. Dr.-med. Jochen Klucken, Prof. Dr. Björn Eskofier

Duration
11/2018 – 03/2019

Abstract

A physically active lifestyle is a central goal of pulmonary rehabilitation. Therefore, it is crucial to substantially enhance long-term physical activity. Multiple studies measured physical activity in patients with chronic obstructive pulmonary disease (COPD) [1] [2]. Especially the review of Byrom and Rowe [2] shows, that over 50 % of the studies are only reporting the total activity per day and per hour. Although it is good practice to use the mean over a certain time interval to describe the physical activity of a person, this method does not provide a way to compare the patterns of the activity. A subject with a few highly intense bouts per day can have the same total activity over the day like a subject, with many low intensity activities per day. A closer look at the patient’s activity pattern will contribute to a better understanding about the question whether a therapy is successful or not.

In the STAR-Study (Stay Active After Rehabilitation [3]) of the FAU Department for Sport Science and Sport, physical activity of around 400 people with chronic obstructive pulmonary disease (COPD) is recorded using an inertial measurement unit (Actigraph wGT3X-BT [4]) worn at the hip. Data is acquired over seven days during three measurement periods, resulting in around 8400 sets of one-day measurements.

The goal of this thesis is to analyze the data with the focus on identifying typical activity patterns. In this context different methods for activity representation (e.g. activity barcoding [5]), dimensionality reduction (e.g. PCA, manifold learning) and unsupervised learning (e.g. clustering, ANN) will be investigated.

References:

  1. Mesquita, R. et al.: Physical activity patterns and clusters in 1001 patients with COPD. Chronic Respiratory Disease, 14 (3), 2017.
  2. Byrom, B and Rowe, D. A.: Measuring free-living physical activity in COPD patients: Deriving methodology standards for clinical trials through a review of research studies. Contemporary Clinical Trials, 47, 2016.
  3. Geidl, W. et al.: Effects of a brief, pedometer-based behavioral intervention for individuals with COPD during inpatient pulmonary rehabilitation on 6-week and 6-month objectively measured physical activity. Study protocol for a randomized controlled trial. Trials, 18 (1), 2017
  4. www.actigraphcorp.com
  5. Paraschiv-Ionescu, A. et al.: Barcoding human physical activity to assess chronic pain conditions. PLoS One, 7 (2), 2012.