The combination of deep learning and the Internet of Things (IoT) has brought many changes to the prediction of health problems and the provision of solutions for making a timely diagnosis to patients. This paper proposes a new system to connect the IoT data acquisition method with a deep learning image recognition paradigm to diagnose medical conditions at an early stage without a properly labeled dataset. Medical images are classified using convolutional neural networks (CNNs) where the classification performance is measured by precision and recall as well as F1-score which stand at 92%, 88%, and 90%, respectively. The system also plays a role in guaranteeing real-time monitoring using IoT devices, with data transfer indeed taking 1.2 s only as oppose to the traditional approach. Quantitative assessments support the effectiveness of the proposed approach, and the solution solves important problems of traditional health care, including diagnostic procrastination and the lack of scalability. Outcomes of classification and comparative latency are graphically represented to present instant and real-time performance of the system. This research lays down groundwork to apply the advanced technologies into the healthcare structures with an objective of early identification of the issues and efficient working of the patient care system.