A robust sensor fusion method for heart rate estimation

Area of Science:

• Biomedical Engineering
• Signal Processing
• Clinical Monitoring

Background:

• Clinical physiologic data often suffers from corruption, leading to inaccurate readings, data loss, and false alarms.
• Existing methods lack robust strategies for combining data from multiple sensors to enhance data quality.

Purpose of the Study:

• To develop a sensor fusion method for combining heart rate measurements from multiple sensors.
• To achieve an artifact-free heart rate estimate with associated confidence values.
• To produce a more accurate heart rate estimate than individual sensors can provide.

Main Methods:

• A novel method discriminates between valid and erroneous sensor measurements, fusing only reliable data.
• Consensus algorithms, incorporating predicted heart rate values and physiological plausibility, identify good readings.
• A Kalman filter, utilizing 16 possible data state hypotheses, generates the fused heart rate estimate.

Main Results:

• The sensor fusion method demonstrated effective performance with clinical data.
• The approach successfully generated artifact-free heart rate estimates with confidence values.
• The fused estimates proved more accurate than those from individual sensors.

Conclusions:

• The presented sensor fusion technique offers a viable solution for improving clinical heart rate monitoring.
• Further research is needed to address limitations and explore advanced applications of the method.