Near infrared spectrum noninvasive blood glucose detecting method and detecting network model training method thereof

Abstract

The invention provides a near infrared spectrum noninvasive blood glucose detecting method and a detecting network model training method thereof. The method comprises the following steps: training by using near infrared spectrum data and invasive blood glucose concentration data corresponding to the near infrared spectrum data to obtain a plurality of artificial neural networks; selecting two preferable artificial neural networks as basic structures; on that basis, optimizing weight coefficients of the two artificial neural networks by using a particle swarm algorithm to obtain a detecting network model; and adjusting the contribution ratio of the two artificial neural networks in the detecting network model by using the weight coefficients to overcome difference of the physiological law of a single individual every day and individual difference. When the obtained detecting network model is used for implementing near infrared spectrum noninvasive blood glucose detection, only near infrared spectrum data of blood glucose of an individual object to be detected need to be acquired, and then a near infrared spectrum noninvasive blood glucose detecting result is obtained. The detecting network model which is used for implementing the near infrared spectrum noninvasive blood glucose detection has high detecting precision, and can meet requirements of clinical application well.

Description

technical field [0001] The invention relates to the fields of physiological signal acquisition technology and digital signal analysis technology, in particular to a near-infrared spectrum non-invasive blood sugar detection method and a detection network model training method thereof. Background technique [0002] Diabetes is a serious threat to people's life and health, but there is no cure for diabetes in clinical practice. Frequent blood glucose concentration detection and drug control are often used in clinic to maintain the blood glucose concentration of diabetic patients at a normal level. Scholars at home and abroad have invested a lot of energy and resources in blood glucose concentration detection technology, and have achieved certain research results. The current blood glucose detection methods include three categories: non-invasive, minimally invasive and invasive. Among them, invasive and minimally invasive detection have higher accuracy and can meet the clinical ...

AI Technical Summary

Problems solved by technology

[0006] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a non-invasive blood glucose detection method by near-infrared spectroscopy, which does not need to rely entirely on the prior assumption of Beer-Lambert's law, and can have high detection accuracy and meet the clinical requirements. Application requirements, to solve the problems of large errors in the existing near-infrared spectrum non-invasive blood glucose detection schemes, difficulty in meeting clinical application requirements, and unfavorable home promotion

Images