Pulse wave analysis of the radial artery waveform: a proof-of-concept for noninvasive anemia detection

Background Anemia is a global health concern, particularly in the elderly, and is often underdiagnosed early. Current non-invasive methods like pulse oximetry cannot assess hemoglobin levels. The radial artery blood pressure waveform (BPW), which can be captured via wrist-worn devices, offers a promising but underexplored source of information for this purpose. Objective This study aimed to develop and validate a frequency-domain analysis method, Pulse Frequency-Distribution Analysis (PFDA), using only the wrist BPW for non-invasive anemia detection. Methods In a hospital-based study, 103 older adults (≥65 years) were grouped into Robust (Hb-NA, n = 53) and Anemic (Hb-A, n = 50) based on clinical hemoglobin standards. From the BPW signal, we extracted 40 harmonic-related parameters. Results The BPW-based PFDA model demonstrated acceptable discrimination, with an overall AUC of 0.72. Gender-specific models performed similarly or better than the overall model, with performance notably enhanced in males (AUC = 0.80). PFDA scores correlated more strongly with hemoglobin levels (|R| > 0.38) than BPW-derived heart rate (|R| < 0.14) and showed relative independence from physiological parameters like blood pressure and age (all R 2 < 0.05). Conclusion This proof-of-concept study validates that the radial BPW alone contains sufficient physiological information for anemia screening. The proposed PFDA model is an accurate, stable, and computationally efficient tool, indicating considerable potential as a data-efficient solution for integration into wrist-worn wearable devices to facilitate early, non-invasive detection of anemia.