I will be involved in a video game expo in September where there’ll be thousands of attendees. I’m wondering if I could make some MMI application to expose awareness of MMI (and bring people to our booth to try things;)
I realize the reality of such a fantasy is too early but I thought I’d at least throw the topic out there in this forum.
I think it’s a great idea. I would be happy to collaborate with you on this, though it’s a little short time period for development. The advanced processing methods I just posted can make a significant improvement in effect size using existing hardware.
I’ve got some code that produces this type of output:
It uses Scott’s recent random walk bias algorithm combined with his recent advanced processing methods for increasing effect size: https://drive.google.com/file/d/1rP7Ee35K0kbQ3zXCcZvnj3d5WqmROTT_/view
Code on Github is here.
It also requires a locally running MySQL database, since it stores all data for future analysis. You may notice that it’s pulling and storing all types of solar data as well, including the Kp geomagnetic index and Solar DST index. My intention was (still is, though it’s lost priority for me at this moment) to pull in a bunch of secondary data sources during trials and then use machine learning to find patterns. That is to include solar events, galvanic skin response data, and a bunch of data read through an Emotiv Epoch 14 channel EEG, which calculates performance metrics (excitement, interest, engagement, stress, focus, and relaxation) and also tracks brain-computer interface (BCI) mental commands (e.g. “move upwards”). EEG code is not in there but solar index stuff is. I doubt you are interested in that but just FYI.
@ScottWilber Yeah you are right, a collab like that would take a lot longer than I suspect the both of us have time before the event. Either way we could start brainstorming some novel application that could be deployed in similar scenarios where there are big crowds.
@fluidfcs1 Nice work with the Python trial experiment. I got it up and running locally. I even created a pull request with a README:
I noticed I was getting this error on startup:
err.raise_mysql_exception(self._data)
File "/usr/local/lib/python3.11/site-packages/pymysql/err.py", line 143, in raise_mysql_exception
raise errorclass(errno, errval)
pymysql.err.DataError: (1366, "Incorrect integer value: '' for column 'SN' at row 1")
so for now I just commented out “db_queue.put(data)” for the solar data on line 104 in main.py
@fluidfcs1 This is almost word-for-word-same as intentions I had a while back but never got the time to really explore. I used to have an Emotiv EEG years ago, but that was before I heard about MMI. A couple of years ago I got a Brain Health Headband Sensor | SenzeBand and even started playing around with graphing its output with my MEDs but I got sidetracked and ended up just commenting out the Neeuro related code. I dabbled around in trying to use Google’s ML to analyze GCP data but didn’t get far.
Awesome! Very cool that you got it working and appreciate the pull request; I’ve approved and merged it. I actually fixed that error a couple days ago if you pull the latest code. The problem was that the URL to get DST solar data has the current month it, which I didn’t realize, so the new code dynamically updates the URL rather than hardcodes.
Also, it’s good to pull most recent code anyway because I’ve been working with Scott over the last few days to validate logic and make some changes to ensure everything aligned with his new paper. Lastly, I’m going to be making some more updates today also, because I want the bar chart to have targets both on the top and bottom if two-tailed (trying to create both more 1s and 0s within the same trial) is selected. At Scott’s advice I’ll also switch it to run continuously if two-tailed is selected rather than being time-bound. I hope to have that all done and fully merged within the next 8 hours.
From there my only future plans are to add in more data sources. Various Emotiv Epoch EEG data as mentioned (brainwave band data, performance metrics, BCI), galvanic skin response (just got this guy), temp/humidity. I’m sure that stuff is less relevant for your needs.
I’m very happy to modify or support based on what you’re trying to do. I’d love to see this program getting some use, which so far it really has not from my end.
One idea was to see if your code could be run in a Jupyter notebook or something, then perhaps more people could access it online without having to setup their local environment (and we could use the MED devices I’ve made a public API for).
Then they could have an online account and see their history and stuff.
On the topic of GSR I’d always wondered whether fitness watches could be used. My Garmin measures some skin properties (I can’t remember what) but from the little research I did that particular data wasn’t exposed through their APIs so I didn’t further than line of investigation.
Interesting idea. I don’t have much experience with that. Those notebooks can access local devices easily?
BTW, I pushed some big updates to the code yesterday but need some more time still. I’ve gotten some updated instructions from Scott about how to perform some of the calculations and require additional clarification, which I’m still awaiting.
@WanderingIshiki lots of changes pushed. I believe I’ve implemented everything per Scott’s specifications but have sent him a PM asking for validation.
Some notable changes to the program here. I modified the readme.md also to have two demo videos of how it looks now.
Possible logic will shift slightly based on Scott’s feedback but I think it’s in pretty good shape generally.
https://github.com/danosb/quantum_influence/tree/main/med100kx8
An example of a two-tailed test with a 30 second defined run-time looks like this:
