Related Experiment Video
Updated: Aug 29, 2025
06:49
Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment
Published on: December 11, 2015
9.0K
Wearable-Based Stair Climb Power Estimation and Activity Classification
Dimitrios J Psaltos1, Fahimeh Mamashli1, Tomasz Adamusiak1
1Pfizer Inc., 610 Main Street, Cambridge, MA 02139, USA.
Sensors (Basel, Switzerland)
|September 9, 2022
Summary
We developed an algorithm using a lower-back accelerometer to estimate stair climb power (SCP), a key measure of leg function. This wearable sensor approach shows strong agreement with clinical tests and enables remote monitoring.
Area of Science:
- Biomedical Engineering
- Wearable Technology
- Musculoskeletal Health
Background:
- Stair climb power (SCP) is a critical clinical assessment of leg muscular function.
- Current in-clinic Stair Climb Power Tests (SCPT) are prone to human error and lack continuous monitoring capabilities.
- Wearable sensors offer a potential solution for continuous, remote assessment of lower-limb function.
Purpose of the Study:
- To propose and validate an algorithm for classifying stair climbing and estimating SCP using a lower-back worn accelerometer.
- To compare the performance of accelerometer-only, gyroscope-only, and combined sensor modalities for SCP estimation.
- To assess the feasibility of an at-home, accelerometer-based SCP assessment.
Main Methods:
- Collected data from 65 healthy adults performing SCPT and walking assessments using instrumented (accelerometer + gyroscope) lower-back sensors.
- Developed and applied two ensemble machine learning algorithms to classify stair ascent periods.
- Extracted features from accelerometer and gyroscope signals to estimate SCP.
Main Results:
- The proposed algorithm demonstrated strong agreement with the clinical standard for SCP estimation (r = 0.92, ICC = 0.90).
- Stair climbing periods were identified with >89% accuracy using accelerometer-based machine learning models.
- Minimal performance degradation was observed using gyroscope alone compared to the accelerometer, and combined sensor use offered only a slight accuracy improvement.
Conclusions:
- An accelerometer-based algorithm can accurately estimate stair climb power (SCP) and classify stair climbing, correlating strongly with clinical measures.
- Wearable sensor technology, particularly accelerometer-based systems, holds significant potential for at-home, continuous monitoring of lower-limb muscular function.
- The findings support the development of accessible, remote assessments for leg strength and function, optimizing battery life for practical at-home use.
