Multidimensional feature-based assessment of microcirculatory hemodynamic dysfunction in patients with Wagner Grade 0 diabetic foot

OBJECTIVE This study characterizes the hemodynamic abnormalities of the microcirculation in the feet of patients with diabetic foot (DF) based on multidimensional features, including time-domain and frequency-domain metrics, rhythmicity, and symmetry. It further elucidates the relationship between these abnormalities and plantar pressure hotspots, specifically the forefoot first metatarsal head (MT1) and fifth metatarsal head (MT5), and constructs a cross-sectional discriminative model. METHODS A total of 286 consecutive participants were included in the study (non-DF: 157, DF: 129). Microcirculation signals from MT1 and MT5 were collected, with features extracted across various domains: time (mean, standard deviation, range, kurtosis), frequency domain (relative power endothelial, neurogenic, myogenic, respiratory, and cardiac bands), rhythmicity (number of peaks, peak intervals, and their dispersion), and symmetry (absolute value of the mean difference between left and right, left-right correlation coefficient). Inter-group comparisons were conducted using the Mann-Whitney U test, and effect sizes were calculated with the Hodges-Lehmann median difference (Δ) and Cliff's δ. Correlations were assessed using Spearman's method with Benjamini-Hochberg false discovery rate (BH-FDR) correction. Variables selected by LASSO were entered into a multivariable logistic regression model. Model performance was evaluated using the area under the receiver operating characteristic curve (ROC-AUC), and the optimal classification threshold was determined using the Youden index. RESULTS Mean perfusion at MT1 and MT5 was significantly lower in the DF group (both p < 0.001), representing the largest between-group differences among the assessed features. Variability metrics differed by measurement site. Notably, the relative power in the neurogenic and myogenic bands at MT1 was significantly decreased, suggesting a weakening of low-frequency autonomic regulation. Furthermore, MT1 exhibited fewer peaks, prolonged inter-peak intervals, and increased dispersion, indicating slower and less stable rhythmicity. Left-right correlation coefficients at bothsites were decreased (p < 0.001), whereas the absolute left-right mean differences did not increase, suggesting reduced synchrony rather than increased amplitude asymmetry. Spearman correlation and multifactor models consistently aligned in direction. Regarding the discriminative models, the area under the curve (AUC) for the MT1 model was 0.845, for the MT5 model was 0.822, and for the combined model (MT1 + MT5) was 0.906, which outperformed the single-site models. CONCLUSION Patients with DF demonstrate a composite pattern of microcirculatory dysfunction characterized by insufficient perfusion, attenuated autonomic regulation, altered rhythmicity, and impaired bilateral coordination. Multidimensional plantar microcirculatory features improve cross-sectional discrimination between DF and non-DF participants, providing a quantitative basis for risk stratification and phenotypic characterization of DF-related microcirculatory abnormalities.