Continuous Glucose Monitoring Data Compression Using Peak-Nadir Encoding in Diabetes: Method Development and...

Area of Science:

• Biomedical Engineering
• Data Science
• Endocrinology

Background:

• Continuous glucose monitoring (CGM) generates large volumes of high-frequency time-series data.
• Efficient storage, transmission, and analysis of CGM data present significant challenges.

Purpose of the Study:

• To develop and evaluate a novel CGM-specific data compression method.
• The method aims to achieve high compression ratios while maintaining signal fidelity and accuracy of clinically relevant glycemic metrics.

Main Methods:

• Introduced PN+, a content-based encoding approach using physiologically salient landmarks (peaks, nadirs) and support points.
• Reconstruction utilized piecewise cubic Hermite interpolation.
• Evaluated PN+ against other compression methods using synthetic and real-world CGM datasets, assessing compression ratio, mean absolute error, and R2.

Main Results:

• At a compression ratio of 13, PN+ demonstrated significantly lower reconstruction error (MAE=0.77) compared to other methods (2.75-3.45).
• PN+ achieved higher R2 values for glycemic metrics and reduced error by over 4-fold for excursion-sensitive measures.
• The method showed robust performance on real-world data with low computational overhead (<0.2 seconds per profile).

Conclusions:

• PN+ offers robust compression for CGM data by preserving physiologically meaningful glucose dynamics.
• The method surpasses generic compression techniques in reconstructing clinical metrics accurately.
• PN+ is suitable for large-scale CGM data storage, interoperability, and downstream analytics due to its efficiency and accuracy.