A method for obtaining dynamic spectrum based on the proportion of multi-wavelength PPG waveform and applying it to noninvasive detection of human platelet content.

Dynamic spectroscopy (DS) theoretically eliminates individual differences and the impact of measurement conditions on accuracy and can achieve high-precision noninvasive blood component analysis. To further improve the extraction quality of DS, this paper proposes a photoplethysmography (PPG) signal waveform proportion extraction method, which realizes the extraction of DS by calculating the proportion coefficient between PPG waveforms. To verify the effectiveness of our method, the transmission spectra of 146 volunteers' fingers and the true value of platelet content (wavelength 600–1100 nm, platelet content range 32 × 10⁹/L–512 × 10⁹/L) were collected. Then the DS was extracted by the single-trail method and this method, and the prediction model is established by PLS. The experimental data showed, compared with the single-trail method, that the modeling effect of the PPG waveform proportion method is significantly improved, the Rc increased 9.61%, the RMSEc decreased 31.56%, the MAPc decreased from 12 to 6%, and the Rp increased 42.92%, the RMSEP reduced 24.39%, and MAPp decreased from 13 to 11%. The results show that the PPG waveform proportion method proposed in this paper has a higher model correlation coefficient and lower prediction error, significantly improves the extraction quality of DS, further improves the accuracy of noninvasive blood component quantitative analysis, and helps to promote the application process of noninvasive detection of blood components. [ABSTRACT FROM AUTHOR]