Related Experiment Video
Updated: Apr 5, 2026
06:49
Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment
Published on: December 11, 2015
9.4K
Improving activity recognition using a wearable barometric pressure sensor in mobility-impaired stroke patients
Fabien Massé1, Roman R Gonzenbach2, Arash Arami3
1Laboratory of Movement Analysis and Measurement, Ecole Polytechnique Fédérale de Lausanne, Station 11, 1015, Lausanne, Switzerland. fabien.masse@epfl.ch.
Journal of Neuroengineering and Rehabilitation
|August 26, 2015
Summary
This study introduces a novel trunk-fixed sensor system that integrates barometric pressure and inertial sensors to accurately monitor daily activities and body elevation in stroke survivors. The new system significantly improves activity recognition, aiding in better mobility assessment for rehabilitation.
Area of Science:
- Biomedical Engineering
- Rehabilitation Science
- Wearable Technology
Background:
- Stroke survivors frequently experience mobility deficits, challenging objective daily-life assessment.
- Current clinical methods lack objective measures of patient mobility in daily living.
- Inertial sensor approaches struggle with activity transitions due to kinematic variability.
Purpose of the Study:
- To enhance daily activity and body elevation recognition in stroke survivors.
- To integrate barometric pressure (BP) and inertial sensor data for improved classification.
- To develop an event-driven activity classifier using fuzzy logic.
Main Methods:
- Utilized a trunk-worn inertial and BP sensor on 12 stroke patients.
- Devised a hierarchical Fuzzy Inference System (H-FIS) for event-driven classification.
- Estimated body elevation using a pattern-enhancing algorithm on BP data.
Main Results:
- Achieved 90.4% Correct Classification Rate (CCR) for posture/activity detection.
- Demonstrated 3.3% and 5.6% improvement over existing FIS-IMU and EPOCH classifiers.
- Obtained 98.2% CCR for body elevation estimation, with enhanced standing activity recognition.
Conclusions:
- Trunk-fixed sensors integrating BP and inertial data significantly improve activity monitoring in stroke patients.
- An event-based fuzzy-logic classifier enhances daily activity recognition and mobility assessment.
- This technology offers a more objective and accurate method for evaluating mobility deficits.
