Classification of Diabetes Using Photoplethysmogram (PPG) Waveform Analysis: Logistic Regression Modeling.

Area of Science:

• Biomedical Engineering
• Medical Informatics
• Cardiovascular Physiology

Background:

• Diabetes mellitus is a global health concern requiring accurate early detection methods.
• Photoplethysmogram (PPG) signals offer a non-invasive approach to physiological monitoring.
• Existing diagnostic methods can be invasive or time-consuming, highlighting the need for efficient screening tools.

Purpose of the Study:

• To develop and validate a predictive model for diabetes classification using photoplethysmogram (PPG) waveform analysis.
• To assess the accuracy of a logistic regression model incorporating age, b/a, and SP indices for diabetes prediction.

Main Methods:

• A logistic regression model was developed using PPG waveform features (b/a and SP indices) and age.
• The study enrolled 587 subjects, with 459 used for training and 128 for testing the model.
• Model performance was evaluated using metrics including overall accuracy, sensitivity, and specificity.

Main Results:

• The developed classifier achieved an overall predictive accuracy of 92.3% for diabetes.
• The model correctly identified 63 diabetes cases and 70% of actual diabetes patients.
• It also correctly classified 479 non-diabetes subjects with 96.4% accuracy for this group.

Conclusions:

• The PPG waveform analysis-based logistic regression model demonstrates high accuracy for diabetes classification.
• This non-invasive predictive model shows potential as a reliable tool for diabetes screening in clinical settings.
• The findings suggest PPG analysis can serve as a valuable surrogate for diabetes prediction.