Non-invasive haemoglobin measurement among pregnant women using photoplethysmography and machine learning

Abstract

Estimation of haemoglobin(Hb) using non-invasive methods has caught wide interest among researchers all over the globe. Among the various devices and methods adopted for non-invasive haemoglobin (SpHb) measurement, prediction of SpHb using PPG can provide earlier and faster diagnosis. The primary objective and motivation of this study is to assess the efficiency and effectiveness of SpHb monitoring among pregnant women using PPG and generalized linear regression technique. PPG signal was acquired from pregnant women with prior consent and Hb was predicted from the time-domain attributes of PPG. Hb value as calculated by the invasive lab methods was compared with the predicted SpHb value. The absolute bias between the SpHb predicted and Hbref was 0.73 g/dL (SD 0.62). To analyse and evaluate the performance of the proposed word, correlation coefficient between SpHb and Hblab, was calculated using IBM SPSS software.