Predictive Internet of Things Based Detection Model of Comatose Patient using Deep Learning

Abstract

The needs and demands of the healthcare sector are increasing exponentially. Also, there has been a rapid development in diverse technologies in totality. Hence varied advancements in different technologies like Internet of Things (IoT) and Deep Learning are being utilised and play a vital role in healthcare sector. In health care domain, specifically, there is also increasing need to find the possibility of patient going into coma. This is because if it is found that the patient is going into coma, preventive steps could be initiated helping patient and this could possibly save the life of the patient. The proposed work in this paper is in this direction whereby the advancement in technology is utilised to build a predictive model towards forecasting the chances of a patient going into coma state. The proposed system initially consists of different medical devices like sensors which take inputs from the patient and helps aid to monitor the condition of the patient. The proposed system consists of varied sensing devices which will help to record patient’s details such as blood pressure (B.P.), pulse rate, heart rate, brain signal and continuous monitoring the motion of coma patient. The various vital parameters from the patient are taken in continuously and displayed across a graphical display unit. Further as and when even if one vital parameter exceeds certain thresholds, the probability that patient will go into coma increases. Immediately an alert is given in. Further, all such records where there are chances that patient goes into coma state are stored in cloud. Subsequently, based on the data retrieved from the cloud a predictive model using Convolutional Neural Network (CNN) is built to forecast the status of the coma patient as an output for any set of health-related parameters of the patient. The effectiveness of the built predictive model is evaluated in terms of performance metrics such as accuracy, precision and recall. The built forecasting model displays high accuracy up to 98%. Such a system will greatly benefit health sector and coma patients and enable build futuristic and superior predictive and preventive model helping in reducing cases of patient going into coma state.