Assessment of Contact Pressure in Wrist Photoplethysmography and a Multi-Step Mitigation Scheme for Cardiac Measurement Accuracy Enhancements

Abstract

Modern commercial wearables utilize photoplethysmography (PPG) sensors to track health metrics with a lightweight, compact design. However, PPG sensors are vulnerable to variations in contact pressure, which refers to the level of tightness of wearables on the wrist. This can cause deformation of PPG waveform morphology and result in inaccurate estimation of health metrics. Electrocardiography (ECG) sensors are the gold standard for heart rate measurement, but they contain certain flaws. We assessed how contact pressure affects wrist PPG heart rate accuracy. In this study, we conducted a secondary data analysis on the publicly available WF-PPG dataset. From the dataset, we computed heart rate from both PPG and ECG signals and compared them to calculate heart rate error. The values of heart rate error were distributed across categories based on PPG waveform morphology types, each associated with a range of contact pressure. Thus, the impact of skin composition on the effect of contact pressure was partially accounted for. Median heart rate error for each category was calculated. The results suggested that contact pressure has a profound impact on the accuracy of PPG sensors. Median wrist error was lowest for Type 2E + 2L (2.38 bpm), except for Type 3 (2.25 bpm), which has a low signal-to-noise ratio (SNR); finger error was lowest for Type 2L (1.36 bpm). Blood oxygen saturation (SpO2), measured by a fingertip pulse oximeter, was found to be only negligibly correlated with contact pressure. The results highlight potential inaccuracy in consumer-grade medical data that should be considered during medical use. A multi-step engineering scheme was suggested to alleviate this challenge of modern wearables.