Pulse-based photoplethysmography quality assessment improves wearable seizure detection performance

Area of Science:

• Biomedical Engineering
• Neurology
• Signal Processing

Background:

• Wearable photoplethysmography (PPG) estimates heart rate (HR) for seizure detection.
• PPG signals are susceptible to motion artifacts, compromising HR accuracy and seizure detection.
• Effective methods for artifact rejection in PPG-based seizure detection are needed.

Purpose of the Study:

• To introduce and evaluate a novel pulse-based quality assessment (PQA) method for PPG signals.
• To compare the performance of PQA against HR-based quality assessment (HRQA) and no quality assessment (noQA).
• To assess the impact of these quality assessment methods on seizure detection performance.

Main Methods:

• Utilized data from the PROMISE trial involving children with refractory epilepsy.
• Applied a novel PQA method to assess individual PPG pulse quality and exclude compromised pulses.
• Compared PQA with HRQA (outlier rejection) and noQA using varying HR increase thresholds.
• Employed permutation tests with bootstrapping for statistical comparisons.

Main Results:

• Both PQA and HRQA significantly improved seizure detection compared to noQA (p < 0.005).
• PQA demonstrated superior performance over HRQA, yielding greater increases in sensitivity, positive predictive value (PPV), and F1-score.
• At clinically relevant sensitivities, PQA resulted in 4%-15% higher F1-scores and 5%-19% higher PPVs compared to HRQA.

Conclusions:

• Implementing PPG quality assessment, particularly the pulse-based method, enhances wearable HR-based seizure detection.
• The PQA method offers a software-only upgrade to improve the accuracy of seizure detection devices.
• This approach increases the reliability of PPG-derived HR data for better seizure identification in epilepsy monitoring.