About 7.9-27.6% of oral dysplasia, a premalignant lesion, transit to the invasive oral cavity squamous cell carcinoma (OCSCC). Prognostic biomarkers are critically needed to determine patients with oral dysplastic lesions at risk for malignant transformation and to guide the targeted development of novel therapies. Our goal is to establish a single-cell atlas of premalignant immune microenvironment (PRIME) in oral dysplasia and to identify immune features that predict the malignant transformation to OCSCC. We propose an innovative approach that combines high-dimensional imaging mass cytometry (IMC) and machine learning predictive modeling (iEN) to analyze a total of ~200 Formalin-Fixed Paraffin-Embedded (FFPE) patient tongue biopsies from the Oral Pathology Archive at the University of the Pacific (UOP), Arthur A. Dugoni School of Dentistry. IMC is a new multiplex imaging technology which combines high- dimensional mass cytometry with microscopy. Immune Elastic Net (iEN) is a machine learning algorithm specifically developing for the analysis of high-dimensional mass cytometry data. We plan to identify immune features that differentiate oral dysplasia severity (Aim 1) and predict OCSCC malignant transformation (Aim 2). In addition, we will analyze the iEN-selected immune features at UOP Han’s lab by conducting multiplex immunofluorescence (mIF) staining on the whole sections (Aim 1&2) to validate and generalize the IMC findings. The proposed research will establish immune landscape in oral dysplasia and identify biomarkers to predict the malignant transformation. It also provides an opportunity for predental or dental students participating in translational research and collaborating with multidisciplinary team to identify biomarkers to improve oral pathology diagnosis.