Increasing accuracy of pulse arrival time estimation in low frequency recordings

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Photoplethysmography (PPG) is used in wearable devices to measure vital signals.
• Lowering sampling frequencies in wearables reduces power and data requirements.
• Accurate Pulse Arrival Time (PAT) estimation is crucial for non-invasive blood pressure monitoring.

Purpose of the Study:

• To develop a novel strategy for estimating PAT at sampling frequencies as low as 25 Hz.
• To enhance the accuracy of PAT parameters derived from low-frequency wearable data.
• To support clinical decision-making using reliable vital sign measurements.

Main Methods:

• A template matching algorithm was developed.
• The method leverages the random nature of sampling time and expected PAT changes.
• Algorithm tested on a public dataset with healthy volunteers in sitting, walking, and running conditions.

Main Results:

• The algorithm significantly reduced the mean error by an average of 16.6% at lower sampling frequencies.
• The standard deviation of the error was reduced by an average of 20.2%.
• Error reduction was more pronounced in the sitting position (22.2% mean, 48.8% std dev).

Conclusions:

• The developed method offers a promising approach for accurate PAT estimation from low-frequency recordings.
• This advancement can improve the clinical utility of wearable vital sign monitoring.
• The technique addresses the trade-off between data efficiency and measurement accuracy in wearables.