Automated Quality Control for Sensor Based Symptom Measurement Performed Outside the Lab

Area of Science:

• Biomedical Engineering
• Digital Health
• Wearable Technology

Background:

• Wearable sensing technology enables remote, non-invasive clinimetric testing.
• Accuracy is challenged by irrelevant sensor data and unexpected user behaviors outside clinical settings.
• These deviations compromise data analysis and scientific conclusions.

Purpose of the Study:

• To develop a unified algorithmic framework for automated sensor data quality control.
• To identify reliable sensor data segments for accurate clinimetric analysis.
• To enhance the reproducibility and replicability of remote patient monitoring studies.

Main Methods:

• Developed a unified algorithmic framework for automated sensor data quality control.
• Integrated parametric and nonparametric signal processing techniques.
• Applied machine learning for automated segmentation and data reliability assessment.

Main Results:

• Achieved an average segmentation accuracy of approximately 90% across 100 subjects and 300 clinimetric tests.
• Demonstrated the framework's effectiveness across three distinct behavioral clinimetric protocols.
• Successfully identified reliable sensor data, mitigating confounding environmental factors.

Conclusions:

• The proposed framework significantly improves the accuracy of wearable sensor data for clinimetric testing.
• Automated data quality control is crucial for reliable remote patient monitoring.
• This approach enhances the potential of digital health technologies in clinical practice.