A study on the factor number determination methods in the partial least squares model for the urinalysis using Raman...

Area of Science:

• Analytical Chemistry
• Biomedical Engineering

Background:

• Non-intrusive measurement systems are crucial for remote patient care.
• Raman spectroscopy offers a potential method for real-time urinalysis.

Purpose of the Study:

• To develop and validate a Raman spectroscopy-based system for non-intrusive urinalysis.
• To compare two methods for determining the optimal number of factors in partial least squares (PLS) calibration models for accurate urine component prediction.

Main Methods:

• Raman spectroscopy was employed to analyze urine samples.
• Partial Least Squares (PLS) regression was used as the multivariate analysis method.
• The number of factors in PLS models was optimized by minimizing Prediction Residual Error Sum of Squares (PRESS) and maximizing the correlation coefficient.

Main Results:

• Both PRESS minimization and maximum correlation coefficient methods were evaluated for factor selection.
• The study compared the accuracy of urine component concentration predictions obtained using the two factor determination methods.
• Correlation coefficients were calculated between pre-examined values and predicted results for unknown samples.

Conclusions:

• The study identified the most suitable method for determining the optimal number of factors for accurate urine component prediction using Raman spectroscopy.
• The findings will inform the development of a non-intrusive urinalysis system for home-based patient monitoring.