Non-invasive estimation of random blood glucose from smartphone-based PPG. (7th July 2022)

• Record Type: Journal Article
• Authors: Roy, Uttam K.; Ganguly, Shivashis; Ukil, Arijit
• Is Part Of: International journal of biomedical engineering and technology. Volume 39:Number 3(2022)
• Page Start: 297
• Page End: 313
• Publication Date: 2022-07-07
• Subjects: healthcare -- blood glucose estimation -- non-invasive measurement -- smartphones -- photoplethysmogram -- PPG -- regression -- calibration; Biomedical engineering -- Periodicals; 610.28
• Languages: English
• ISSNs: 1752-6418

Abstract

Traditional blood glucose metres are invasive in nature, i.e., blood is collected by needle pricking, which is painful and damages tissues over repetition resulting high risk of infections. Although, a few non-invasive methods have been proposed, they require high-end, costly, non-portable and custom devices. This paper proposes a non-invasive method to estimate average blood glucose using only smartphone that enables 24 × 7 monitoring without any extra hardware. The estimation is based on reflectance mode photoplethysmogram (PPG) that records the relative change in light absorbance of glycated haemoglobin (HbA1c) due to change in absorption coefficient and path length. Smartphone-based PPG gets often corrupted due to ambient noise, motion artefacts, etc. We rigorously cleaned the noisy PPG signal and measured reflected intensity from PPG of 25 patients, applied nonlinear regression to estimate glucose and cross-validated against a laboratory invasive method. The RMS error comes out to be 2.1525 mg/dL which is superior to existing non-invasive techniques. To prove the correctness, three standard techniques viz. geometric regression, Bland-Altman analyses and surveillance error grid are used.