Objective To compare the performance of different detection methods for constructing a prediction model for diabetic peripheral neuropathy (DPN) in type 2 diabetes patients based on local thermal hyperemia, so as to provide reference for early clinical screening. MethodsA total of 174 patients with type 2 diabetes were included in the study. Local skin heating was applied to induce local thermal hyperaemia with a temperature of 44 ℃，and laser Doppler flowmetry was used to measure dermal blood flow change. Simultaneously, clinical examination (pressure perception assessed by 10 g Semmes Weinstein monofilament, vibration perception by 128 Hz tuning fork, discrimination by pin prick, thermal perception by Tip therm, and reflexes by Achilles tendon reflexes),vibration perception threshold measurement, and nerve conduction study were performed to explore the influencing factors of DPN using multivariate logistic regression analysis based on their respective diagnostic criteria. Using nerve conduction study as the "gold standard" for DPN diagnosis, the optimal model was selected to construct the nomogram prediction model. ResultsAll three detection methods indicated that the rate of dermal blood flow change, duration of diabetes, diabetic nephropathy, and glycosylated hemoglobin were the influencing factors for DPN (all P<0.05). The performance of the logit model resulting from clinical examination was the best, and the nomogram model constructed based on the results of clinical examination showed good accuracy (Hosmer Lemeshow test χ2=11.147, P>0.05) and discrimination (AUC=0.872). ConclusionThe rate of dermal blood flow change, duration of diabetes, diabetic nephropathy, and glycosylated hemoglobin are influencing factors for DPN. The nomogram model constructed based on the results of clinical examinationhas good diagnostic efficiency, which can be used for clinical screening and identification of DPN in patients with type 2 diabetes.