A Model-Free AI Framework for Robust PQRST Complex Detection in ECG Signals

Accurate delineation of PQRST complex is essential for clinical ECG analysis, but remains challenging due to waveform variability, low amplitude, and polarity inversions. This work presents PAD-PQRST, a model-free Polarity-Adaptive Delineation method for PQRST complex. This approach leverages search windows based on established cardiac physiology. The Q and S waves are identified by locating the corresponding extrema relative to the R-peak polarity. Meanwhile, the P and T waves are determined by finding the extrema with the maximal absolute deviation from local mean. Because this design inherently accommodates waves of any polarity, it eliminates the need for a predictive model. This research uses the Manikandan2012 algorithm for R-peak detection due to its robustness. Validation on QTDB and LUDB yields F1-scores from 98.5% to 99.8%for QRS components, 90.97% to 94.57% for P-waves, and 79%to 91.41% for T-waves, surpassing established wavelet-based and graph-based methods. The combination of physiologically informed search windows and polarity-adaptive detection delivers a simple yet powerful delineation framework suitable for diverse clinical datasets.

@article{Luangsodsai2026,
  title     = {A Model-Free AI Framework for Robust PQRST Complex Detection in ECG Signals},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {17},
  number    = {7},
  year      = {2026},
  publisher = {The Science and Information Organization},
  author    = {Arthorn Luangsodsai and Manh Duong Dang and Krung Sinapiromsaran},
  doi       = {10.14569/IJACSA.2026.0170771},
  url       = {https://doi.org/10.14569/IJACSA.2026.0170771}
}