In the United States, approximately two-thirds of oral cancer patients are diagnosed after regional and distant metastasis, leading to poor survival rates. These outcomes result in part from difficulties in identifying oral precancers and cancers in the clinical setting and determining the likelihood of potentially precancerous oral lesions undergoing eventual malignant transformation. The research objective was to develop approaches to improve the identification and risk assessment of oral precancers and cancers, leveraging changes in optically detectable cellular and tissue features and altered molecular biomarker expression. First, an animal model of oral carcinogenesis that exhibited clinically relevant changes in histopathology, optical markers, and molecular biomarker expression was developed. Second, multimodal widefield autofluorescence and high-resolution microendoscopy images and cancer biomarker expression were analyzed to identify relationships between optically detectable changes and genetic alterations. Third, a process was developed to perform intraoperative oral surgical specimen assessment employing fluorescent vital staining, imaging with a novel artificial intelligence-enabled microscope with an extended depth-of-field, and the generation of virtual histology images for pathology evaluation. Finally, a policy analysis was conducted addressing barriers to oral cancer detection in rural and border regions of Texas. In all, this work is a step towards allowing clinicians to incorporate optical marker and molecular biomarker alterations into the clinical evaluation of potentially high-risk oral sites, better informing the risk assessment and subsequent management of oral precancers and early cancers.