Background: Atopic dermatitis (AD) is a chronic, multifactorial skin condition with distinct subtypes, each exhibiting unique immune characteristics and treatment responses. Objective: Our study aimed to characterize biomarker profiles of AD subtypes and identify biomarkers that may predict the response to abrocitinib treatment. Methods: Serum and tape strips from lesional and nonlesional skin were collected from patients with AD (n = 50) pre- and post-4 weeks of abrocitinib treatment (n = 15) and 13 healthy controls. We compared serum and skin biomarker expression in 16 patients with extrinsic and 16 with intrinsic AD with matched disease severity. Results: Th1, Th2, Th17/22, general inflammation-related factors, and JAK-STAT signaling molecules were upregulated in the skin and serum of patients with AD. Compared with intrinsic AD, extrinsic AD showed higher FCER1A and chemokine (C-C motif) ligand 22 expression in the skin (P < 0.05), while intrinsic AD had higher Th17/Th22-related biomarker expression. Multivariate correlation models significantly improved the correlation between the biomarkers and the severity of AD, both before and after treatment. Conclusion: Extrinsic AD shares Th2-type inflammation with intrinsic AD, but intrinsic AD shows elevated Th17/Th22-type inflammation. Biomarker integration models can contribute to precise AD treatment.