Heart Rate Estimation from Neck Photoplethysmography using FFT-Based Scoring and a Shallow Neural Network.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence in Healthcare

Background:

• Continuous heart rate monitoring is crucial for vital sign assessment.
• Existing monitoring devices may have limitations in signal acquisition due to sensing modality or body location.
• The need for accurate and user-friendly heart rate estimation systems is significant.

Purpose of the Study:

• To develop and evaluate a novel method for estimating heart rate using neck photoplethysmography (PPG) signals.
• To assess the accuracy of the proposed method using FFT-Based scoring and a shallow neural network.
• To determine the clinical acceptability of the developed system for heart rate monitoring.

Main Methods:

• Utilized photoplethysmography (PPG) signals acquired from the neck.
• Implemented a novel FFT-Based scoring technique.
• Employed a shallow neural network for heart rate estimation.
• Evaluated performance using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and error Standard Deviation (STD).

Main Results:

• Achieved an average RMSE of 3.13 ± 4.66 and MAE of 1.96 ± 3.38 across all data.
• Demonstrated improved accuracy with an average RMSE of 1.55 ± 1.43 and MAE of 0.83 ± 0.86 after excluding outlier data.
• The novel FFT-Based scoring coupled with a shallow neural network showed competitive performance.

Conclusions:

• Neck PPG signals can be effectively used for heart rate estimation.
• The proposed FFT-Based scoring and shallow neural network method offers a promising approach for accurate heart rate monitoring.
• This technique has the potential for developing easy-to-use and clinically acceptable HR monitoring systems.