LSTM Based Deep Learning Model for Blood Sugar Prediction

Abstract

Diabetes has become one of the most prominent health problems in the modern era. Neural networks aid in better medical diagnosis considering dynamic nature of learning model. LSTM model is a form of artificial recurrent neural network which is widely used in deep learning, specifically in sequence prediction data elements. Main benefit of opting for LSTMs model in this research is that it provided significant aid in sequence classification using raw time series data for data transformation and classification of blood sugar level. Results showed that blood sugar level can be predicted by using the LSTM model with an error margin of approximately ±39. Accuracy of the model can be improved by inclusion of additional parameters in the model to minimize the variation.