Stochastic Modeling for Photoplethysmography Compression^.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Data Compression

Background:

• Photoplethysmography (PPG) is crucial for health monitoring in clinical and wearable settings.
• Efficient storage and transmission of PPG signals are vital for diagnosis and healthcare.
• Current compression methods face limitations in achieving high compression ratios while maintaining signal integrity.

Purpose of the Study:

• To develop a novel stochastic modeling approach for photoplethysmography (PPG) signal compression.
• To improve storage and power efficiency for PPG data in wearable devices.
• To achieve high compression ratios with clinically acceptable signal recovery.

Main Methods:

• Modeled single cardiac periods of PPG waveforms using Gaussian functions.
• Employed adaptive quantization based on higher-order statistics of inter-cardiac period parameters.
• Evaluated performance on a wearable PPG dataset with 30 subjects.

Main Results:

• Achieved a compression ratio of up to 41 for 18-bit data at 200 Samples/s.
• Maintained Percentage Root-Mean-Square Difference (PRD) below 9%, ensuring clinical acceptability.
• Outperformed conventional delta modulation-based compression methods.

Conclusions:

• Stochastic modeling shows high potential for PPG compression, particularly for wearable devices.
• The proposed method offers significant improvements in compression ratio and efficiency.
• Contributes to enhanced remote and telehealth monitoring capabilities through efficient PPG data handling.