Objective  To investigate the clinical value of Raman spectroscopy and routine detection in diabetic patients.

  Methods  A total of 1 000 healthy people, 300 diabetic patients and 19 newly diagnosed diabetic patients were selected. All samples were detected for glycated hemoglobin (HbA1c), glycated albumin (GA), free fatty acid (FFA), insulin (INS) and 1, 5-anhydroglucitol (1, 5-AG). Raman spectrum was used for analysis and the error rate was calculated.

  Results  There were significant differences in levels of HbA1c, GA, FFA, INS and 1, 5-AG between the healthy group and diabetic group, newly diagnosed diabetic group (P < 0.05). There were significant differences in levels of GA, FFA and 1, 5-AG between the newly diagnosed diabetic group and the diabetic group (P < 0.05), while the levels of HbA1c and INS showed no significant differences (P>0.05). The Raman peaks of the three types of samples were similar, and it was impossible to distinguish three types of samples from the peaks. The PCA-LDA analysis was able to distinguish three types of samples, and the overall error rate was only 0.3%.

  Conclusion  Application of PCA-LDA for further analysis of the measured Raman spectroscopy result can effectively distinguish three types of samples, and provide reference for clinical grading and treatment of diabetes.