Florida Atlantic University Researchers Develop Novel, Accurate Balance Assessment Using Wearable Sensors and Machine Learning
Boca raton, FL – Researchers at Florida Atlantic University (FAU) have developed a groundbreaking new approach to balance assessment utilizing wearable sensors and advanced machine learning algorithms. This innovation addresses critical limitations of current balance evaluation methods, offering a more objective, comprehensive, and accessible solution with the potential to transform balance disorder management, particularly in remote and home-based care settings.
The Problem with Current Balance Assessments:
Customary balance assessments frequently enough rely on expensive, specialized equipment and are heavily dependent on clinician expertise, leading to potential variability in results.A need exists for more objective and widely accessible tools for accurate balance evaluation.
FAU’s Innovative Solution:
The FAU team, from the College of Engineering and Computer Science, tackled this challenge by leveraging the power of wearable technology and machine learning.Their approach utilizes wearable sensors placed on the ankle, lumbar (lower back), sternum, wrist, and arm to capture detailed motion data during balance tests.
Study Methodology:
The study focused on the Modified Clinical Test of sensory Interaction on Balance (m-CTSIB), a widely used clinical assessment. Participants were evaluated under four sensory conditions – eyes open/closed on stable and foam surfaces - for approximately 11 seconds each. Data collected from inertial measurement unit (IMU) sensors was compared to “ground truth” scores obtained using falltrak II, a leading fall prevention tool, to train and validate machine learning models. Researchers employed Multiple Linear Regression, Support Vector Regression, and XGBOOST algorithms to estimate m-CTSIB scores from the wearable sensor data. the study also investigated optimal sensor placement for maximizing accuracy.
Key Findings:
* High Accuracy: The new method demonstrated a high degree of accuracy and strong correlation with established balance scores, proving its effectiveness and reliability.
* strategic Sensor Placement: Data from sensors placed on the lumbar region and the dominant ankle proved most effective in estimating balance scores, highlighting the importance of targeted sensor placement.
* XGBOOST Model Performance: The XGBOOST model, utilizing lumbar sensor data, achieved outstanding results in both cross-validation and demonstrated a high correlation with low error rates.
* Objective, Quantifiable Metrics: The combination of wearable sensors and machine learning transforms movement data into objective, quantifiable balance metrics.
Expert Perspectives:
“Wearable sensors offer a practical and cost-effective solution for capturing detailed movement data, which is essential for balance analysis,” said Behnaz Ghoraani, Ph.D., senior author and associate professor at FAU. “These sensors provide insights into 3D movement dynamics, essential for applications such as fall risk assessment.”
Stella Batalama,Ph.D., dean of FAU’s College of Engineering and Computer Science, added, “This approach is more accessible, cost-effective and capable of remote administration, which could have significant implications for health care, rehabilitation, sports science or other fields where balance assessment is important.”
Implications & Future Directions:
This research signifies a significant step forward in balance assessment, offering a potentially revolutionary method for:
* Remote Monitoring: Enabling healthcare professionals to evaluate patients’ balance remotely.
* Cost-effectiveness: Providing a more affordable alternative to traditional assessment methods.
* Accessibility: Expanding access to accurate balance assessments in diverse settings.
* Improved Understanding: Providing deeper insights into the nuanced effects of sensory inputs on balance.
The study’s findings were published in the journal Frontiers in Digital Health.