The global prevalence of type 2 diabetes is estimated to reach 4.4% by 2030, placing a significant burden on our healthcare system. Therefore, the ability to identify patients in early stages of the disease is essential for both prevention and effective management, and diagnostic methods based on traditional Chinese medicine (TCM) may be suitable for identifying patients with early-stage type 2 diabetes. Here, a panel of three physicians trained in TCM classified 44 pre-diabetic subjects into three syndrome subtypes using TCM-based diagnostics. In addition, ultra-weak photon emission (UPE) was measured at four anatomical sites in each subject. Ten properties encompassing 40 parameters were then extracted from the UPE time series. Statistical analyses, including multinomial logistic regression, were performed using the results of each parameter measured at the four sites. Sixteen UPE parameters were then selected and used to discriminate between the three subtypes of pre-diabetic subjects. Next, Spearman's correlation coefficient was used to quantify the correlation between the 16 UPE parameters and the TCM-based diagnoses. The resulting correlation networks accurately reflected the differences between the three syndrome subtypes. These results suggest that UPE is a suitable tool for detecting subtypes in early-stage type 2 diabetes. In addition, our results provide evidence that TCM may represent an important step toward personalized medicine.