Estimation of Blood Glucose Level of Type-2 Diabetes Patients using Smartphone Video

TL;DR

This paper introduces a non-invasive method using smartphone videos to estimate blood glucose levels in type-2 diabetes patients by converting finger videos into PPG signals and applying machine learning for prediction.

Contribution

It presents a novel non-invasive approach combining smartphone video analysis, signal processing, and PCR modeling for blood glucose estimation.

Findings

Achieved SEP of approximately 18.31 mg/dL in glucose prediction.

Processed PPG signals effectively to extract relevant features.

Demonstrated feasibility of smartphone-based glucose monitoring.

Abstract

This work proposes a smartphone video-based approach for the estimation of blood glucose in a non-invasive way. Videos using smartphone camera are collected from the tip of the subjects finger and the frames are subsequently converted into Photoplethysmography (PPG) waveform. Gaussian filter along with Asymmetric Least Square methods have been applied on the PPG signals to remove the high-frequency noise, optical and motion interferences. Different signal features such as Systolic and Diastolic Peaks, the time difference between consecutive peaks (DelT), First Derivative peaks, and Second derivative peaks etc have been extracted from the processed signal. Finally, Principal Component Regression (PCR) has been applied for the prediction of glucose level from the extracted features. The proposed model, while applied to an unbiased dataset, could predict the glucose level with a Standard Error of Prediction (SEP) of around 18.31 mg/dL.