IoT as a concept has advanced at an extremely high rate, and with the improvements made within it, smart systems in healthcare systems have been made possible for real time monitoring and diagnosis through connected devices. The current study proposes a machine learning based image recognition for the IoT based smart healthcare devices to boost the diagnosing precision and patients’ safety consequently. The MSP approach integrates deep learning technologies, including CNNs, for image analysis of the images acquired from IoT technologies in remote health monitors, diagnostic equipment, and imaging equipment. These images are then analyzed in real time, giving the precise disease diagnosis and supporting the decision-making process. The proposed system was assessed based on a publicly accessible archive of medical images, yielding the following results: accuracy of 90% precision of 92%, recall of 89% and an F1 of-score 0.90. These findings show the ability of combining ML with IoT applications for digital health imaging in emergency care. Significant advantages of the suggested method over conventional image recognition methodologies such as SVM and shallow CNNs were exhibited in terms of all the performance indicators. This paper discusses how smart healthcare solutions using IoT can revolutionize diagnostic approaches by delivering improved accessibility and enhanced diagnostic accuracy in healthcare systems.