Adding multiple noise sources with target SNR with ECG data using the MIT BIH Noise Stress Test Database

Question

I'm new to signal processing, and I believe to understand what additive noise is. However, while reading several ECG denoising papers, I've noticed that some combine multiple noise sources from the MIT BIH Noise Stress Test database. For example, in here and here.

I'm also considering that applying a single noise source to ECG data isn't straight forward. Luckily, Physionet provides a tool nst, which also describes how we can apply a signal noise source

The nst tool allows specifying a target SNR and takes as input a single noise source and an ECG record. My questions here is:

How can I apply multiple noise sources to reach a target SNR? In every paper I've read so far where authors applied multiple noise sources, they simply mention how they "combined the noise sources" and compared their methods using the same SNR as with a single noise source.

One thing I would like to rule out, is that the authors just applied the nst tool with the same SNR target for each noise source. As in, rather than checking whether the combined noise sources amount to the target SNR they applied the nst tool once for each noise source additively, where at each step they input the same target SNR.

For example, let record 100 be the noise free record and 10 is our target SNR. Then,

1. Apply nst with target 10dB with noise source EM, outputs 100_em
2. Apply nst with target 10dB with noise source BW, outputs 100_em_bw
3. Apply nst with target 10dB with noise source ma outputs, 100_em_bw_ma

I'd greatly appreciate if anyone could also point me to the appropriate theory, and it feels like I'm missing something really fundamental when authors collective do not believe it is important to mention how the noise sources are combined exactly.