New study examines effect of time series length on measurement accuracy of dynamic properties of photoplethysmograms – ScienceDaily

Refinement of the potential applications of photoplethysmography, an optical technique used to measure heart rate, in cardiovascular and mental health monitoring requires analysis of complex nonlinear photoplethysmogram (PPG) data. Bypassing traditional analytical methods to resolve the complex and dynamic PPG signals, researchers at Tokyo University of Science have used nonlinear analysis to determine the accuracy of dynamic characteristics estimated using short PPG signals.

With the increasing burden of cardiovascular and mental disorders worldwide, the need for early detection and timely health monitoring is becoming increasingly urgent. Wearable devices serve as a practical, affordable and non-invasive approach for systematic and continuous health monitoring. “Photoplethysmography,” a simple optical technique based on photoelectric pulse wave signaling, has been used for decades to monitor parameters such as heart rate, oxygen velocity, and blood volume changes in clinical settings as well as in motion via wearable devices. These measurements rely on basic signal processing and analysis, such as noise filtering and motion reduction.

Information derived from dynamic photoplethysmograms (PPG), recordings of biological signals made in photoplethysmography, can be used to monitor physiological and mental health, but such advanced applications are hampered by high measurement noise and motion artifacts in PPG. especially those obtained using wearable devices.

So, how can the complex nonlinear dynamics of PPGs be analyzed to expand their clinical applications?

Diving deeper into analyzing the complex characteristics of PPG, a team of researchers from Japan evaluated the applicability of nonlinear analysis of short PPG signals in clinical measurements and the accuracy with which they can estimate the dynamic properties of PPG. A group of researchers led by Dr. Nina Sviridova, assistant professor at Tokyo University of Science, including Prof. Toru Ikeguchi of Tokyo University of Science, Dr. Tijun Zhao of Niigata University of Agri-Food and Prof. Akimasa Nakano of Chiba University published their findings in the “Mobile Health Data Analytics” special issue of the journal Sensors. The study was published in the journal’s Volume 22, Issue 14 on July 9, 2022.

“Filtered signals can be used for traditional photoplethysmography applications; however, they are not suitable for advanced analysis. Alternatively, only high-quality short segments of PPG signals can be used, but the applicability of nonlinear analysis to such short recordings has not been investigated in details,” explains Dr. Sviridova.

Advanced nonlinear analysis methods used to estimate PPG dynamics are often limited by the length of the applied data. Previous studies have shown that repetitive quantification analysis (RQA), a non-linear analytical approach, is unaffected by signal length. In this study, researchers used RQA to extract the dynamic properties of PPG, such as determinism, divergence, predictability, and complexity from short signals. PPG recordings were obtained from thirty healthy subjects by measuring near-infrared light transmission from skin surfaces. These records were further subsampled to generate sparse time series data. Additionally, the chaotic “Rössler model” (a model used to describe continuous chaos in dynamic nonlinear systems) was used to calculate the relative error while accounting for noise.

The results suggest that dynamical properties such as “determinism”, “predictability” and “entropy” can be estimated with good accuracy (less than 1% error) using short time series of signals. Comparisons with the noisy Rossler system suggest that in the absence of noise, the lower length of the time series is acceptable for measuring these properties with accuracy. However, for some properties such as “divergence”, short PPGs were not sufficient for accurate estimation with an acceptable error (lower than 1%).

These observations can help estimate the error associated with the dynamic properties in cases where only short-length PPG signals are available, and support future studies using other photodetectors and studies in different experimental and real-world settings. Understanding the complex characteristics of PPG can further enhance the clinical applications of wearable health monitoring technologies.

Highlighting the wider applications of their research, Dr Sviridova says: “Findings from this study will help improve the assessment of health parameters using wearable devices, ultimately accelerating the World Health Organization’s goal of early detection of cardiovascular and mental illnesses.’

Leave a Comment