Periodontitis is a chronic inflammatory disease affecting approximately 45% of adults worldwide, representing a leading cause of tooth loss and a significant public health burden. Early detection is critical for effective intervention, yet conventional diagnostic methods periodontal probing and radiographic imaging are operator dependent, time consuming, and inaccessible in many primary care settings. This paper presents a comprehensive framework for IoT enabled periodontitis detection using edge/cloud artificial intelligence (AI). The system integrates three core innovations: (1) a network of low cost, non invasive IoT sensors (optical, electrochemical, and mechanical) that capture salivary biomarkers, gingival tissue reflectance, and pocket depth estimates within 60 seconds; (2) an edge AI module (quantized neural network, 0.9 MB footprint) that performs initial screening and anomaly detection with <100 ms latency, enabling real time feedback in clinical or home settings; and (3) a cloud AI module that aggregates anonymized data for multi modal fusion, longitudinal tracking, and continuous model improvement via federated learning. We validate the framework using a simulated dataset (1,200 patients, 80% periodontitis prevalence) and a physical testbed (20 sensor nodes, 5 edge gateways). The edge classifier achieves 87.3% sensitivity and 83.1% specificity for distinguishing healthy from diseased sites. The cloud fusion model improves sensitivity to 93.8% and specificity to 91.7% using combined optical and electrochemical biomarkers (IL 1β, MMP 8, and hemoglobin). System level latency is 85 ms for edge inference and 340 ms for full cloud processing, well within clinical tolerances. This work establishes that IoT edge cloud architectures can democratize periodontal screening, enabling early detection in primary care, teledentistry, and at home monitoring while maintaining privacy and scalability.