Utility of photoplethysmography in detecting elevated blood glucose among non-diabetics

Abstract Introduction: This study aimed to evaluate a technique of using photoplethysmography (PPG) for detecting elevated blood glucose in individuals. Method: This is a prospective, cross-sectional study in which 500 healthy volunteers were recruited at a tertiary hospital in Singapore from October 2021 to February 2023. Capillary glucose was measured concurrently with PPG signals acquired using the wrist-worn Actxa Tracker (Spark + Series 2) and the In-Ear Prototype model SVT, which were worn for a duration of 8 min. Participants with a capillary blood test reading ≤11.1 mmol/dL had to consume a standard glucose tolerance drink and return 1 h later for a second capillary blood test. Two hundred and forty-four features were subsequently extracted from the PPG signals. Results: Of the 500 volunteers, 17 were excluded because of incomplete records. This led to a total of 483 participants’ records being included in the final analysis. For predicting elevated capillary blood glucose level, demographics alone achieved an area under the curve (AUC) of 0.75. When wearable features derived from PPG were combined with demographics, AUC improved significantly to 0.82 ( P = 0.0001). Conclusion: This study shows that a non-invasive method of assessing diabetes mellitus risk using PPG combined with demographics is a viable option to provide a cheaper and more accessible modality for population-wide diabetes mellitus risk assessment.