Abstract: When analyzing brain functional connectivity (FC), differences in the hemodynamic response function (HRF) are often overlooked. However, these differences can affect the calculation of FC and may confound the accuracy of statistical results. This study uses diabetic patients as an example to compare HRF parameters (response amplitude, response time, and full width at half maximum) between diabetic patients and healthy individuals based on their resting-state functional magnetic resonance imaging (rs-fMRI) data. Significant differences in HRF parameters were identified in specific brain regions, which were then used as seed points to construct whole-brain functional connectivity. Statistical analysis was performed to compare the differences in connectivity strength both before and after deconvolution, within and between groups. The results indicate that in some key brain regions (such as brain64), significant differences in HRF parameters exist between the diabetic patient group and the healthy population. T-test results show that FC connectivity strength exhibits different significant patterns before and after deconvolution, reflecting that variability in HRF affects the calculation of FC connectivity strength. This, in turn, may influence the understanding of changes in the brain network of individuals with diabetes and its underlying neural mechanisms. The findings suggest that when conducting FC analyses based on blood oxygen level-dependent (BOLD) time series in metabolic brain diseases, it is essential to fully consider the potential impact of HRF variability on the results and to perform deconvolution on the BOLD time series to correct for HRF differences. This will help to more accurately capture changes and characteristics of brain functional connectivity, provide deeper insights into disease mechanisms, and partially address reproducibility issues in fMRI research.