Dynamic modeling of photoacoustic sensor data to classify human blood samples

• DOI: 10.1007/s11517-023-02939-3
• Publication Date: 2023-10-25T23:02:57Z
• Authors: Argelia Pérez-Pac...; Roberto G. Ramíre...; Rosa M. Quispe-Si...; Marco P. Colín-Ga...

Abstract The photoacoustic effect is an attractive tool for diagnosis in several biomedical applications. Analyzing signals, however, challenging to provide qualitative results automated way. In this work, we introduce a dynamic modeling scheme of sensor data classify blood samples according their physiological status. Thirty-five whole human were studied with state-space model estimated by subspace method. Furthermore, the are classified using parameters and linear discriminant analysis algorithm. classification performance compared time- frequency-domain features autoregressive-moving-average model. As result, proposed can predict five classes: healthy women men, microcytic macrocytic anemia, leukemia. Our findings indicate that method outperforms conventional signal processing techniques analyze medical diagnosis. Hence, promising point-of-care devices detect hematological diseases clinical scenarios. Graphical abstract