MindWare info/input about psyphr

[wendtke]

@iqis What do you think?

[wendtke]

Notes from today's meeting with MindWare reps (Eric and Jay):

Connections

• Greg Norman: psychophysiology expert; R user
• Mallory Feldman at Northeastern University recently contacted MindWare about a similar idea. Maybe she can contribute to psyphr. I think she is most interested in compiling HRV data.

Next steps

• MindWare is willing to serve in an advisory capacity throughout the process and will point MW users to psyphr when it is ready.
• When the repo is in good shape, I will share it with MW. They and Greg Norman can give feedback.
• Eventually, they would like us to do a demo video of psyphr for the MW website; however, we need to think about the rOpenSci process with this. Is it a concern to be sharing/advertising the package before the peer-review process is complete?

Answers to our questions

• MW will send templates for each data type. They are all structured in similar ways.
• File naming conventions. This is tricky. There is no MW mechanism or requirement. They thought the settings page within the Excel workbook might be useful in helping R identify the subject or task for a data subset. We discussed different approaches (e.g., identification by file/directory or user or indicators from settings page). I think our best bet is still building parameters within functions that calls for subject ID, task, etc.
• There is limited consistency or predictability in study design schema. They did mention they consistently see baseline, task, and recovery periods throughout study designs. The most popular analysis applications are HRV, EDA, and IMP. As such, participants could have 3 files per analysis type, or many more, if split into time periods (e.g., HRV baseline, HRV task...IMP recovery, etc.). This schema is decided in study design, and it is difficult for psyphr to get that information until the research team is ready to process data and are considering which tools are best.
• Quality control guidelines. MW likes the idea of incorporating filter options into psyphr and suggests that we make these filters suggested, not required; in some cases, they should be default=TRUE, but with the FALSE option. We also discussed the possibility of psyphr returning a summary of dropped segments and the reason (e.g., short; outside range) to the user when implementing the filter. There are no consistent QC guidelines for EDA and impedance, but we can build in suggestions about physiologically probable ranges (e.g., 2-35 microsiemens for skin conductance OR 0.12-0.40 Hz for respiratory band settings) for the user to be alerted about possible outliers. Inter-class correlations and multiple data editors (research assistants) help support quality control, especially for the data with limited empirical suggestions. They also mentioned that it would be VERY useful to build in an option to automatically drop the short segments (i.e., any < 30 seconds) that usually float at the end of task files (e.g., 305 seconds = 10 30-second segments and 1 5-second segment).
• Visualization ideas. Descriptive statistics and exploratory data analysis options would be beneficial. Greg Norman might have more suggestions on this point.

Extra information

• MindWare's study compiler from 2012 also incorporated demographic information.
• HRV, EDA, and IMP are the most popular MindWare analysis applications. We should focus on supporting these first.

Consider

• MW is interested in expanding their own tools in these similar ways (e.g., data visualization options). They want to stay connected with us, so we can give them end-user suggestions. I hope we can also support researchers in ways that MW cannot.
• Many researchers who use MW and BIOPAC also use SPSS (vomit). Let's think about how to make psyphr appealing to those who are resistant to learning "a whole, new" programming language.
• Balance between flexibility for the user and control/predictability for psyphr.
• How psyphr might connect to and support psychophysiological study design (in the distant future). If researchers are using psyphr from the beginning, the file naming conventions and study schema issues are reduced.

[wendtke]

I just emailed MW with question about accessing raw(er) data from the analysis applications in non-Excel format.

[wendtke]

@iqis do you know a way to copy this information to our Google Drive, so we can close this issue?

[iqis]

I think we can just close this issue. If we need to refer to it some time in the future, we can always search for it in the Issues tab.