Scholarlink

<p>Physiological traits such as skin colour, body fat, skin thickness, pH value, hydration level, and others play a crucial role in non-invasive glucose monitoring as they directly impact the performance of blood glucose level measurement (BGLM). The DC component of the photoplethysmogram (PPG) signal accounts for a significant portion of BGLM error caused by these physiological traits. This paper proposes a novel method to estimate the error correction factor (ECF) by correlating the DC component with BGM error. Linear Regression (LR) and Support Vector Machine (SVM) regression models are employed to predict blood glucose levels (BGL) and estimate the ECF. The proposed method uses PPG records from 15 healthy subjects for model training and 30 additional records for validation. Testing across diverse age groups demonstrated the value of Root Mean Squared Error (RMSE) and Coefficient of Determination (R<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(^{2}\)</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mn>2</mn> </mmultiscripts> </math></EquationSource> </InlineEquation>) as ±5.1575 mg/dL and 0.9581, respectively. Compared to existing work, the proposed method brings a reduction of 30.83% in RMSE and an improvement of 2.13% in R<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(^{2}\)</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mn>2</mn> </mmultiscripts> </math></EquationSource> </InlineEquation>. Clarke Error Grid Analysis confirmed that 99% of predictions fell within clinically accurate Zone A. The proposed method offers significant advantages, including reduced cost, simplicity, comfort, and suitability for wearable real-time BGM systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Non-invasive blood glucose level measurement method under the consideration of various physiological traits using machine learning

  • 期刊论文

  • 发表日期：

    2026年1月5日

查看全文
  • Dileep Kumar,
  • Niraj Pratap Singh,
  • Gaurav Verma

摘要

Physiological traits such as skin colour, body fat, skin thickness, pH value, hydration level, and others play a crucial role in non-invasive glucose monitoring as they directly impact the performance of blood glucose level measurement (BGLM). The DC component of the photoplethysmogram (PPG) signal accounts for a significant portion of BGLM error caused by these physiological traits. This paper proposes a novel method to estimate the error correction factor (ECF) by correlating the DC component with BGM error. Linear Regression (LR) and Support Vector Machine (SVM) regression models are employed to predict blood glucose levels (BGL) and estimate the ECF. The proposed method uses PPG records from 15 healthy subjects for model training and 30 additional records for validation. Testing across diverse age groups demonstrated the value of Root Mean Squared Error (RMSE) and Coefficient of Determination (R \(^{2}\) 2 ) as ±5.1575 mg/dL and 0.9581, respectively. Compared to existing work, the proposed method brings a reduction of 30.83% in RMSE and an improvement of 2.13% in R \(^{2}\) 2 . Clarke Error Grid Analysis confirmed that 99% of predictions fell within clinically accurate Zone A. The proposed method offers significant advantages, including reduced cost, simplicity, comfort, and suitability for wearable real-time BGM systems.