Clinical validation of digital mobility outcomes in patients with chronic obstructive pulmonary disease

Megaritis, Dimitrios (2024) Clinical validation of digital mobility outcomes in patients with chronic obstructive pulmonary disease. Doctoral thesis, Northumbria University.

Text (Doctoral thesis) megaritis.dimitrios_phd(19012873).pdf - Submitted Version Restricted to Repository staff only until 28 September 2024. Download (10MB) | Request a copy

Abstract

Recent technological advances have enabled us to assess Digital Mobility Outcomes (DMOs) referring to gait characteristics measured in real world conditions using lower back inertial measurement units (IMUs) in health and disease. In COPD, mobility impairment is evident compared to healthy individuals, and whilst the predictive capacity of gait and mobility parameters is high, their construct validity and sensitivity to detect change following different pharmacological and non-pharmacological interventions remain uncertain. This thesis contributed to the clinical validation of novel DMOs developed as part of the joint EU/IMI-funded Mobilise-D project in patients with COPD.

A systematic review and meta-analysis evaluated the effect of all available pharmacological and non-pharmacological interventions on physical activity outcomes in COPD patients. Interventions involving bronchodilator therapy and those focusing on behavioural modifications to enhance physical activity demonstrated notable success in significantly increasing daily step counts. In contrast, interventions centred around exercise training, conducted within outpatient or community-based pulmonary rehabilitation programmes or through telerehabilitation, exhibited inconsistent efficacy in improving physical activity outcomes, with observed variations in effects and directions across studies.

The construct validity of novel DMOs was assessed during laboratory based, free-living habitual and real-world settings. Irrespective of the degree of lung dysfunction, cadence, walking speed, stride duration and single support duration were shown to be sensitive in reflecting COPD symptom burden. The efficacy of machine learning models for predicting COPD clinical outcomes using real-world DMOs was established. Notably, neural network models outperformed logistic regression and linear regression in terms of accuracy rates, demonstrating robust agreement, precision, accuracy, and sensitivity across variables.

DMOs reflecting real-world mobility hold great promise for remote patient monitoring, assessing treatment effects and establishing future biomarkers for regulatory approval while providing a more comprehensive understanding of the holistic health status in patients with COPD.

Actions (login required)

View Item

Downloads