*3.3. Selected Algorithms for Comparison*

We choose some representative algorithms in each category to evaluate the context impacts considering the complexity and practicability. In heuristic methods, threshold (THR) and FSM [35] were selected because they are intuitive and simple to implement in real systems. In the signal processing category, STFT [50], DWT [50] and PTM [27] were selected. We chose STFT and DWT because they are popular and simple to implement. We chose PTM because it has a refined processing chain. In the machine learning category, we chose KNN [29] and SVM as the simple model and complex model, respectively.

These algorithms are chosen from simpler to more complex models considering the requirements and resources of a real system. The evaluation routine is shown in Figure 1.

**Figure 1.** The method of context impacts the algorithm and output. THR, threshold; PTM, Pan-Tompkins method.
