Non-invasive hemoglobin concentration measurements with multi-wavelength reflectance mode PPG sensor and CNN data...

Area of Science:

• Biomedical Engineering
• Optical Sensing
• Wearable Technology

Background:

• Accurate hemoglobin concentration monitoring is crucial for diagnosing and managing various medical conditions.
• Current invasive methods pose risks and discomfort to patients.
• Non-invasive techniques are highly desirable for continuous and accessible health assessment.

Purpose of the Study:

• To evaluate the feasibility of non-invasive hemoglobin concentration measurement using wearable photoplethysmography (PPG) devices.
• To develop and validate a method for estimating hemoglobin levels from PPG signal features.

Main Methods:

• Utilized PPG waveform analysis at multiple wavelengths in reflectance mode.
• Applied decomposition of temporal and spectral features of the PPG signal.
• Trained and validated a neural network model using a dataset of 840 PPG waveforms from 170 volunteers.

Main Results:

• Achieved a Mean Absolute Error (MAE) of approximately 13.6 g/l and a correlation coefficient (R) of 0.62.
• Performance was validated against invasive blood tests.
• Demonstrated the potential for accurate hemoglobin estimation.

Conclusions:

• Non-invasive, real-time hemoglobin concentration measurement is achievable with low-cost wearable sensors.
• The developed method shows accuracy comparable to existing non-invasive clinical instruments.
• This technology holds promise for improved patient monitoring and healthcare accessibility.