| dc.description.abstract | This is the first step in a series of studies to test the feasibility of using Raman Spectroscopy (RS) to non-invasively detect physiologically relevant blood alcohol concentrations. Blood tests, urine tests, and the breathalyzer are currently the most commonly used techniques to measure blood alcohol content. Blood tests are invasive and require wet labs. Although urine tests are non-invasive, they also require wet labs. The breathalyzer is non-invasive and does not require a lab, but its accuracy has come into question. This method measures alcohol content in the alveolar air, which is not always a reliable predictor of alcohol content in the blood. A Raman-based technique could alleviate these problems and eventually replace or complement the breathalyzer. To explore this option, samples of ethanol-in-water from 0-1.0% (wt/vol) were prepared, and a time study was performed to test both the ability of RS to detect alcohol in relatively short amounts of time and to find an optimum scanning time. The spectra were used to create a partial least squares (PLS) model, which uses differences in the spectra to construct a model to describe the relationship between certain peaks and alcohol concentration. The model was used to predict the concentration of several known samples. The accuracy of the model’s prediction was compared for each stage of the time study. This study then focused on measuring ethanol concentrations in plasma, a much more complicated media than water. Again, a model was constructed, used to predict alcohol concentrations, and tested for accuracy. This study showed that Raman spectroscopy has the ability to detect alcohol in the physiological range in rather complex environments and in limited time windows, but repeatability is the main question. Spectra have a tendency to be inconsistent at such low alcohol concentrations, which results in a weak model and less accurate predictions. | en |
