Virtual Patient for Health Science Students

[brkurt]

Hello everyone...

Starting in January, I will be teaching a Psychology for the Health Sciences class.
I am looking at creating a Virtual Patient for my students to work with.
For example, I will do Facial Expressions / Lying Patient study.
I know that Makehuman was intended for such an audience, and help from the Crew would be appreciated.

[joepal]

(sitting at the Mid Sweden University department of psychology while writing this)

Interesting. However, a few more details would be helpful. For example, Is the patient going to be animated? Should there be interaction?

[brkurt]

Yes, the patient will be animated, using a stable version of Blender 2.64; that way I can share the model on the forum here.
I'll keep the interaction fairly simple, with an HTML 5 interface, and a MySQL backend, so that I can used triggers.

A very simple example: an illustration of the Eysenck Personality Questionnaire, but having various patients respond to a therapist's questions. I will integrate the Ekman Facial Action Coding system (as much as possible) to help the budding therapist spot lies, both conscious and unconscious.

Others of you can use the models I generate for other purposes; mostly I'm going to focus on the problems of patient self-report.

[joepal]

It will be very interesting to see how you're going to connect all those parts into a functioning whole. Where will you place the majority of the logic? As python scripting in blender?

[brkurt]

It's not as complicated as it sounds, actually.

Mostly a LAMP server -- I use XAMP -- does the heavy lifting. HTML 5 eliminates the need for tricky php coding.

Each video is stored as a record in a MySQL table; the user clicks on a video --one per page--and then clicks on radio checkbox for 'Admit' or 'Deny'.
(That is, the therapist is using a slowed-down animation to detect microexpressions that the Ekman system analyzes to be a lie.)
Here's where Makehuman and Blender really shine, since I can change the frame rate, and slo-mo the facial expressions.

Once the therapist has completed analyzing all the interview videos (ten would be enough, twenty would be a Voight-Kamp test!)
they would press 'Submit' and their analysis would be passed as a new record to another MySQL table.

For this demo, I would start with microexpressions around the eye muscles (the orbicularis oculi especially) and the mouth muscles.

I would create a second demo that included neck-twisting and shakes of the head, mostly likely with live hair.

[joepal]

Sounds viable. I thought you meant "live" animating via blender, where the doll would react upon input and set an expression that way.

Anyway, people up here sounded kind of curious when I mentioned the project.

[joepal]

People at the psych department won't leave me alone after I mentioned this.

It seems there was an idea a few years ago to implement Ekman in visual experiments with psychophysiological measurements. The idea was to study social phobia and use gradual morphs (say on scale from neutral to angry) and see if patients with diagnosed social phobia would get detectable EEG peaks earlier on the scale than "normal" people. This all failed on lack of stimulus material. An attempt was made with image morphing between neutral and angry, but they never got it to look acceptable.

Anyway, the EEG/experiments guy responsible for the psychology lab up here says maybe we could connect via skype or something during the spring and share thoughts about using Ekman and MH as stimulus material.

[joepal]

For reference, I've added a feature request for Ekman here: http://code.google.com/p/makehuman/issues/detail?id=341

[brkurt]

Yes, we certainly should connect on this one. The potential is enormous.

I'm going to work with my Psychology students to build a few working models, using Makehuman Alpha 7 and Blender 2.65 (that should be stable for mesh modelling now).

I will post the models on this forum throughout January and February.