Fs/F=2^(n+1) . where Fs is the sampling frequency, F is the source operating frequency, n is the No. of level of decomposition

Without knowing the details of the signals themselves and what they represent, I can give a partly answer based on experiences with HL7 v3 messaging and EHR and devices interactions.

I think you can handle two - to three composition levels at a minimum.

Level 1 would identify the patient about whom the EEG is about.

Next on the same level information about sender (the EEG machine) and receiver (the Patient Record?) would be required. If it is the same machine presenting the data, then it is sender and receiver in one.

Next is the clinical observations and conclusions.

You would have to identify the set of all frequency signals as a whole. Here it is possible to move that to a third level if that is easier to work with.

On the clinical observations it might also be necessary to document circumstances, patient observations outside the signals (e.g. a leg movement leading to a peak in an EEG signal, then the leg movement is clinical observation data).

And the EEG will lead to a conclusion of a knowledgeable clinician. That is another set of data, frequently in free text.

Finally, there might be suggestions for interventions or care (even if to wash the patients hair immediately to remove the contact substance).

So level 1 subject / device / observations / conclusions / treatment.

Level 2 observation category EEG signals

Level 3 each single EEG signal uniquely coded and the signal / data organised as a value for this category.

I have no clue how many signals there are exactly, but I have seen patients going for EEG and having a couple of dozen signal sensors attached to their skull.

It is possible to collapse level 2 and 3, but handling the signals themselves separately as category from other categories helps to distinghuish each.

If there are types of signals, e.g. frontal signals ABC, parietal signals left DE, parietal signals right FG etc. then a fourth decomposition level would apply.

For more information see ISO TS 13972, health informatics Detailed Clinical Models, in particular the section on single data elements, small molecular groupings and larger compositions.

My job is normally composing from the details.

Hope this helps

William

please refer to the following paper:

Selection of Mother Wavelet Function for Multi-Channel EEG Signals Analysis during a Working Memory Task