Multicomponent blood analysis by near-infrared Raman spectroscopy

Prior research has shown that Raman spectroscopy can detect molecular signatures in biological fluids. However, no prior work had resolved how to apply this technique to measure multiple analytes in complex blood samples. Existing methods often require sample preparation or separate assays for each compound. This gap motivated researchers to explore whether Raman signals could be decoded to predict several analytes simultaneously. It was already known that partial least squares is a useful calibration method for spectral data. But applying it to blood samples posed new challenges. The need for noninvasive blood analysis drives interest in this area. This paper's contribution is to show whether Raman spectroscopy can yield accurate predictions for multiple analytes in serum and whole blood.

The researchers aimed to test whether Raman spectroscopy could measure multiple blood analytes at physiological concentrations. They wanted to determine if this method could predict several analytes from a single spectrum. The specific problem addressed was whether Raman signals from complex biological samples could be reliably decoded. The motivation came from the potential for noninvasive diagnostics using optical methods. They focused on serum and whole blood samples from multiple patients. The goal was to develop calibrations that work across different individuals. They also wanted to assess how stable these calibrations are over time. The study sought to evaluate if this approach could be robust enough for medical applications.

The team used a near-infrared diode laser emitting at 830 nm to generate Raman spectra. They collected data from serum and whole blood samples in vitro. Raman signals from multiple analytes overlapped in the spectra. To separate these signals, they applied partial least squares regression. They validated their calibrations using cross-validation techniques. The dataset included 66 patient samples for serum analysis. For whole blood, they conducted a preliminary analysis of the same analytes. They tested how well their calibrations held up over seven weeks of system drift.

The strongest finding was accurate prediction of six analytes in serum using 60-second spectra. Glucose, cholesterol, and urea showed significant accuracy in their predictions. The calibration models achieved good performance across the 66-patient dataset. The method demonstrated robustness against system drift over seven weeks. In whole blood samples, they predicted some analytes and hematocrit accurately. The Raman signals from different compounds overlapped but were still separable. The team used partial least squares cross-validation to build their models. These results suggest the method could be useful for medical diagnostics.

The authors suggest that Raman spectroscopy may offer a viable method for multianalyte blood testing. Their findings indicate that this approach could predict several analytes from a single spectrum. They propose that the method is robust enough to handle system drift over time. The results suggest potential for noninvasive diagnostics using optical methods. The team notes that their calibrations worked across multiple patients. They suggest that this technique could be applied to whole blood as well as serum. The authors caution that further validation is needed before clinical use. They conclude that these results support future exploration of Raman spectroscopy in medical diagnostics.