Development of a Low-Cost Real-Time ECG Platform: Signal Processing and Multi-Subject Validation

This study investigates whether consumer-grade ECG hardware can produce physiologically meaningful cardiac metrics through optimized signal processing. A complete real-time ECG acquisition and analysis pipeline was designed, implemented, and evaluated using a SparkFun AD8232 analog front-end and Arduino Uno R4 Minima at a total hardware cost of $90.78. The pilot study included six participants (age range 38–74 years; 2 female, 4 male) and 30 resting trials. A fourth-order Butterworth bandpass filter (0.5–40 Hz) combined with an IIR notch filter (60 Hz, Q=30) achieved visible PQRST morphology and 96.3% mean R-peak detection accuracy (range: 89.4–100.0%, N=5 trials). Three structured experiments evaluated physiological response, electrode placement robustness, and filter algorithm performance. BPM measurements were validated against Apple Watch Series 8 across 25 paired measurements from five subjects, achieving a combined MAPE of 9.30%, Pearson r=0.762 (95% CI: 0.52–0.89, R²=0.581), and mean bias of +6.24 BPM. A pronounced heart-rate-dependent bias was observed in this implementation: in the higher-rate subgroup (Subjects 2–4), MAPE remained below 4.5%, while the lower-rate subgroup (Subjects 5–6) showed MAPE above 16%. This pattern, observed in the current dataset, should be confirmed with a larger validation study but directly motivates future adaptive threshold calibration.