Fig. 5

Clinical datasets regarding the number of cases of AMR healthcare-associated infections (HAIs) were collected and normalized to the number of patients on the sampling day to represent the incidence rate at the hospital’s Inpatient Department (HAIs/patient). Their correlations were analyzed with the relative abundance of ARGs (a), and that of ARG-hosting bacterial genomes (b). Data collected in summer and winter were depicted with red squares and green dots, respectively; solid lines indicate significant correlations and are annotated with a P value and a coefficient of correlation (R²). The dashed curves generally describe the trend in the variance of the datasets by locally estimated scatterplot smoothing (span = 0.75). c Structural equation modeling (SEM) analysis of the impacts of environmental factors (Env) and reported HAIs in wards on the variations of the abundance of carbapenem (Cpm), cephalosporin (Cep), and other β-lactam-resistant genes carrying bacterial genomes. d Linear correlations between the estimated percentages of bacterial origins from the human-source associated microbiome and the total abundance of potential antibiotic-resistant bacteria (PARB). The linear relationship fitting a significant correlation (Pearson, P < 0.05) is depicted using a solid line with confidential intervals (gray shades), while the insignificant relationship is shown using a dashed line. The sizes of the dots are in proportion to the log₂-transformed HAIs that were recorded on the sampling day and the numbers of patients are indicated