Millions of people worldwide suffer from oral cancer, a common and deadly condition. In order to improve treatment results and save lives, early identification of oral cancer is essential. In this study, a new framework is presented for oral cancer diagnosis that uses multimodal data fusion to combine Convolutional Neural Networks (CNNs) with Long Short-Term Memory (LSTM) networks. The proposed framework begins by ingesting diverse data modalities including clinical images. A CNN component processes the image data, extracting spatial features indicative of potential malignancy. Simultaneously, medical image data are fed into LSTM networks to capture temporal dependencies and contextual information. Through multimodal data fusion, the framework combines spatial and temporal cues, enabling a holistic understanding of oral cancer progression. This integrated approach facilitates the detection of indirect abnormalities and early-stage lesions that may be missed by conventional methods. The trained model exhibits high sensitivity and specificity, making it a promising tool for assisting healthcare professionals in early cancer screening. Integrating such automated systems into clinical workflows can aid in the timely diagnosis and treatment of oral cancer, ultimately improving patient outcomes and dropping the burden of this disease. This research highlights the transformative impact of deep learning technology in the pasture of medical image investigation and its potential to enhance healthcare diagnostics and patient care. To estimate the framework's efficacy, extensive experiments are conducted on a large dataset comprising oral cancer cases with diverse characteristics. Presentation metrics such as sensitivity, specificity, and area under the curve (AUC) are utilized to assess the model's diagnostic accuracy and generalization capability.