Dynamic Spectrum Extraction Method Based on Absolute Difference Summation and Statistical Theory

Dynamic spectrum theory has great importance in the field of noninvasive measurement of blood components. To enhance the computational efficiency and data utilization of the existing extraction methods, this paper proposes a new one based on the absolute difference summation (ADS) and statistical theory. The ADS is used to obtain the eigenvalue from a photoplethysmography (PPG) signal. The statistical method is used to obtain the final dynamic spectrum. The experimental data of PPG signal from 133 volunteers were extracted by the new method and the single-trial (ST) extraction method, and the partial least squares model was used to build the calibration models. Compared with the ST extraction, the new method showed better prediction ability. The correlation coefficient of the prediction set increased from 0.85 to 0.92, and the root mean square error of the prediction decreased from 13.49 to 9.86 g/L, which proved that this method can significantly improve the quality of the dynamic spectrum.