The early detection of hidden infections in at-risk groups could be crucial for prompt treatment and improved health outcomes. Recently, traditional diagnostic methods have struggled to identify infections at or before the onset of symptoms, causing delays in treatment. This study presents a Smart AI System that uses real-time data from wearable health devices, IoMT sensors, and electronic health records to help identify latent infections through sophisticated models of machine learning and deep learning. The system depends on data processing, feature selection, and predictive modeling to generate risk scores, enabling earlier interventions. Models tested here include LSTM, Random Forest, and Gradient Boosting to classify patients with an infection. The LSTM showed the highest accuracy. It is effective in monitoring potential vulnerability in nearly real-time. The results emphasize the system’s potential for early detection, which could lead to personalized healthcare interventions. There is still a need for further validation and improvement in clinical settings.