Discriminant Analysis of Urine Vitamin C by Visible Spectrum Based on Ultra‐Dimensionality Reduction and Information Enhancement Technology

Urinary vitamin C (UVC) is a routine urine test. Rapid and accurate detection is of great significance in clinical and health management. This study first uses the successive projections algorithm (SPA) to select four wavelengths highly correlated with UVC. Secondly, the m-SPA-SI information enhancement (IE) framework is constructed by gradually fusing the spectral index value (SIV). Finally, a random forest (RF) discriminant model is established. The results show that the m-SPA-SI6-RF strategy achieves the best model performance, with test accuracy and sensitivity of 93.75% and 0.8182%, respectively, which are far better than those of the full-spectrum RF and SPA-RF models. The number of wavelengths decreased from 457 to 4, a 99.12% reduction. This study not only provides a new method for the first time to achieve efficient spectral discrimination of UVC, but also expands the application potential of spectral analysis technology for simultaneous detection of multiple components in urine.