Taeyun Ha
Eastlake High School
Abstract
When acquiring electrocardiogram (ECG) and photoplethysmogram (PPG) signals, the acquisition configuration—sampling rate, amplifier gain, band-pass filtering, notch filtering, and ADC resolution—inherently trades off diagnostic fidelity against resource cost. Selecting a single “best” configuration by a weighted sum suffers from arbitrary weighting and the inability to recover solutions on non-convex regions of the Pareto frontier. This work introduces the skyline query from databases to derive, without any weights, the set of Pareto-optimal configurations over a seven-dimensional attribute space: three signal-quality dimensions (SNR, ST/low-frequency fidelity LFF, QRS-detection F1) and four resource/noise dimensions (motion-artifact ratio MAR, power, data volume, latency). We further model three clinical scenarios—resting diagnostic ECG, ambulatory monitoring, and ICU real-time monitoring—as subspace skylines, showing that different priorities surface different optimal sets. Using an illustrative application to an Arduino-based educational kit (GT-BME), we discuss the value of jointly teaching database querying and biosignal processing.
