Machine Learning Approach for Early Detection of Diabetes Using Raman Spectroscopy

The application of machine learning technology for invasive diabetes diagnosis has become a research trend in medical sectors in recent years. In this research, we utilize the Raman spectroscopy of glucose fluid sample to detect the glucose level. We create glucose-liquid samples with 14 mixed rates between glucose and pure water to simulate the 14 glucose levels of human blood. Then, the Raman spectroscopy of each sample is obtained. Jittering augmentation method is used for enriching the dataset, which is 20 times larger. Several machine learning models and a 1-D Convolution Neural Network are utilized to identify glucose levels in samples. The result is completely optimistic with high accuracy for predicting glucose level of sample.