[New Method of Measuring Arterial Oxygen Saturation]

Area of Science:

• Biomedical Engineering
• Optical Sensing
• Physiological Monitoring

Background:

• Traditional arterial oxygen saturation measurement relies on the R value, derived from logarithmic photoplethysmography.
• The R value method involves dimension reduction based on the Lambert-beer law, which can lose information and introduce errors due to light scattering.
• Existing methods face limitations in precision and are susceptible to individual differences and optical path variations.

Purpose of the Study:

• To develop a new method for measuring arterial oxygen saturation that minimizes errors caused by light scattering.
• To improve the accuracy and stability of oxygen saturation measurements using photoplethysmography.
• To establish a predictive model that outperforms the traditional R value method.

Main Methods:

• A new linear regression model was developed using peak and valley values of dual-wavelength logarithmic photoplethysmography waves as independent variables.
• In-vivo measurements were conducted on 23 healthy volunteers, collecting 133 photoplethysmography wave samples and reference oxygen saturation values.
• Comparative analysis involved randomly selecting modeling and prediction sets, with the process repeated 10 times to assess predictive performance.

Main Results:

• The new method demonstrated superior predictive performance compared to the R value method.
• Average correlation coefficients for the new method and R value method were 0.8906 and 0.8468, respectively.
• Average root mean square errors were 0.8896% for the new method and 1.0373% for the R value method, indicating higher accuracy.

Conclusions:

• The proposed method, utilizing dual-wavelength photoplethysmography parameters, significantly improves arterial oxygen saturation measurement accuracy and stability.
• The predictive model based on four parameters offers a more robust approach, overcoming limitations of scattering effects.
• This technique holds promise for enhancing the measurement of human physiological information from limited spectral data.