Over the past few decades, IoT has become indispensable in many industries. More people can now get healthcare and their general health can be improved thanks to recent developments in the healthcare sector. Predictive analytics in the medical field can help turn a reactive healthcare approach into a proactive one, thanks to advanced artificial intelligence and machine learning techniques that are penetrating the healthcare business. The main obstacles to utilising IoT for health monitoring are managing regulatory compliance while maintaining security, privacy, dependability, and data accuracy. Nonetheless, a solution has been suggested to get past these obstacles. Initially, the sensors will gather information from the patient and store it in a cloud-based data collection system. A deep learning-based SA-SGRU network receives data from the cloud to classify them as either abnormal or normal. The doctor receives patient information from SA-SGRU to determine if the particular patient is in an emergency condition or not. A notification will be sent to the patient in the event of an emergency. The patient will receive the diagnosis report if there is no emergency. The experiment result indicate that the suggested method outperforms both LSTM and FLSTM by achieving the accuracy of 95% in the detection of abnormality.