Universal non-invasive blood glucose estimation method based on time series analysis

A universal non-invasive blood glucose estimation method based on time series analysis, characterized in that the method comprises the following steps:

• a) data input and preprocessing:
• acquiring relevant physiological parameters of human body continuously with a non-invasive method to obtain a feature value sequence xi(t), i=1, . . . , T, t=1, . . . , Z, where, i is the serial number of the feature, t is the serial number of sampling point, T is the total number of feature value points obtained by calculating the acquired physiological parameters, and Z is the length of the series;
• at the same time, using an invasive method to get the reference blood glucose level series Glu(t), and then normalizing the feature value sequence and the reference blood glucose level sequence;
• b) features screening:
• features are screened according to the similarity between the feature value sequence and the reference blood glucose level sequence, a subset of features that are highly related with the blood glucose is selected, and recorded in related features information;
• c) establishing single-feature model based on time series analysis:
• expressing the relation between each of the related feature value sequence and the reference blood glucose level sequence with a time series analysis model, to obtain single-feature model and single-feature model-based glucose profiles;
• d) multi-feature fusion:
• single-feature model-based glucose profiles are integrated using the weighted average method, to obtain multi-feature fusion model;
• e) carrying out non-invasive blood glucose estimation with the information of the related features, the single-feature model, and the multi-feature fusion model.