Aging is a biological process associated with many complex human diseases. Previously we have found that 3D human facial images can reliably predict physiological age, and define slow and fast agers (outliers) based on the difference between chronological and predicted age (AgeDiff). Here, we profiled the transcriptomes of peripheral blood mononuclear cells (PBMCs) using ribo-minus RNA-seq of 280 individuals from the same cohort. Outliers in AgeDiff identified either phenotypically or through their transcriptome significantly overlap. We found immune and inflammation related gene expression modules to be most significantly associated with AgeDiff, consistent with immune cell type changes. We identified transcriptomic changes associated with facial aging features, such as wrinkles, eye drooping and nose widening, and further inferred the molecular regulators mediating the impacts of different lifestyles on facial aging speed through a causal inference model. As an example, the model showed that smoking modulates facial AgeDiff through the changes of cytokines GRN and SEMA6B expression in the blood. Based on our data and analyses, we created the human blood gene expression-3D facial image (HuB-FI) association database to visualize the lifestyle-transcriptome-facial aging connections to illustrate how lifestyles may affect the human blood transcriptome and ultimately reflect as aging-related changes in morphological patterns of the face.