I want to get concentration and mellow on the android application created in Unity. In the old version of SDK under the android was able to receive concentration and mellow. The new version removed this feature. How to implement it now and preferably in Unity.
I believe the original algorithms were taken from this: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0130129
After studying the “My Virtual Dream” paper, I spent a few weeks creating my own version of the mellow and concentration algorithms. However, I’m not 100% happy with the outcome and I think I understand why Interaxon withdrew their versions. If you would like to try them out, they are hidden in Muse Monitor. To enable them, go to Muse Monitor’s settings screen and at the bottom, tap on the version number 10 times.
This is frustrating, is there a way to get the old SDK? I’m not sure what we’re supposed to develop for the Muse without the concentration score. I too need this in unity,
You could try decompiling the old SDK files for the algorithm, but other than that I don’t think you’ll have much luck.
I’m working on the same problem now. After a brief BCI/neurofeedback literature review, I found three possible algorithms with increasing levels of complexity/precision.
Concentration as beta performance (basically the one described in ‘my virtual dream’ article, cited above) which however I don’t find useful outside of prototyping as isolated band values are often very difficult to make sense of and certainly not a stable indicator of a cognitive state.
Concentration as theta by beta ratio is a better index (http://waset.org/publications/9999102/determination-of-the-concentrated-state-using-multiple-eeg-channels) but I still didn’t find it completely accurate’.
Higuchi/Katz algorithms yield the best accuracy according to this study: http://waset.org/publications/9999102/determination-of-the-concentrated-state-using-multiple-eeg-channels but are also fairly complex and I still don’t know how to implement them.
Anyone else has any other ideas?
#3 is basically the beta/theta ratio. Would that work for your case?
@slmille4 oh, ok. I didn’t realize that. I still haven’t wrapped my head around it. I think if I manage to implement it and test it I could see if it works better.
Do you have any more literature/case studies on this?
The biofeedback field isn’t exactly know for scientific rigor or peer review. Honestly, I think your best bet would be to try to record your target ideal state, and then analyze either the raw signal or the processed bands to figure out how to describe it with a protocol.
I have the same problem.
Maybe someone can help to just select concentration as value as in MuseIO, but from SDK library