Multi-Scale and Context Aware Optimized Glucose Prediction using Neural Networks

Diabetes is a chronic condition characterized by an increase in blood glucose levels in the human body as a result of either insufficient insulin production or body cells that do not respond to insulin. At most hospitals, numerous tests are performed to collect diabetes diagnostic information. The data are then used to advise treatment. Obtaining an accurate diabetes prediction is highly challenging due to the availabil-ity of limited amount of data and the occurrence of outliers. Deep learning has recently been utilized in healthcare and medical research to achieve good results in a variety of tasks such as predicting patient conditions and disease identification. This study has compared the ARIMA (Auto Regressive Integrated Moving Average) model and LSTM model to fore-cast Blood Glucose Level (BGL). The deep learning model consists of the LSTM NN model. The performance of the suggested LSTM NN is compared to the ARIMA model in terms of Root Mean Square Error (RMSE).