Original proposal for the THEMA plan.

From: Markus Klein
Date: Oct 27, 2006 12:34 PM
Subject: draft for thema
To: stas

Describe the program you need as you were living
in a perfect world :-)
The programming details is for a later date.

OK. Here it comes. I hope you will not be struck by the amount of what
I ask to be created. But you have asked :-)
You will see that there is still much to discuss. This is just what
evolved in my head up to now.

The entry point to the program should be built by three symbols only
(maybe arranged in form of a triangle). The first symbol (upside)
should lead to the mouse training , the lower left symbol should lead
to visual tasks (explanation follows) and lower right symbol should
lead to the verbal / acoustic tasks (explanation follows). Maybe there
is a picture of the frontal view of a brain in the background. In this
way the visual and the verbal symbols a placed on the correct
hemisphere of the brain. The whole admin interface should only be
accessible by a combination of keystrokes (like CTRL+A or something) at
every moment in the program. This is because the program is supposed to
run in Kiosk_mode mainly. For simple use everything should be
controllable by mouse only, except in the admin interface. So it can be
used for diagnostics by a professional examiner who is in control of
the keyboard. (My thoughts about the admin interface follow later).
Every game that is chosen should be introduced by a two-seconds
flashing movie in the foreground / not fullscreen (like a gif-image)
where a usability hint is shown. For example: The mouse training is
introduced by the image of a hand that moves a computermouse or a
finger that clicks on the button and so on.

The mouse training can exactly be the one of GCompris. First the
movement, second the single click, third the fishes (with the
improvements that were mentioned recently on the gcompris mailing list:
link The double
click training is not necessary. Here I would already like to mention
that the principle of the cubes in GCompris should be in Thema, too.
Every difficulty level should be directly accessible in every
subprogram (except the letter recognition game - I will describe
later). In the mouse training some data should be logged invisibly in
the background: time to finish every level and a measure for the
strategy (is the way of the mouse arrow systematically from left to
right + up to down or diagonal ? - the first is better and gets more
points, the later is worse and gets less points. Results will be shown
in the admin interface).

The visual tasks should contain Tangram like in GCompris. There is
nothing to change. Data should be logged for time to finish every
Then there is the traffic puzzle (the red bar has to move out of the
traffic jam) from GCompris. Data could be logged for time to finish
every level and number of perseverative moves (How often is a bar moved
without any solution - something to discuss later).
Then there is the visual memory (cards) of GCompris with pictures only,
no mixing with numbers along the difficulty levels. Memory cards with
numbers should be apart with the memory cards for calculations of
GCompris. Data should be logged for time to finish and number of errors
(false alarms - how many times they click on the wrong card that was
already shown once).
Then there is Tower of Hanoi - Data logging for number of moves and
time to finish.
I have to speak with my colleagues if they use other GCompris games,

The verbal /acoustic tasks are:
1.) Acoustic memory with sounds. Data logging like in visual memory as
described above. 2.) Acoustic memory with sounds and spoken words (like
the barfing of a dog and the word "dog" and/or maybe instruments. 3.)
And acoustic memory with spelling of letters and numbers ("A", "3",
"F" and so on..). The sound of the previous item should instantly be
interrupted by the next click/item. Patients are disturbed by delays
when the computer does not react.
Then there should be the reading lessons of GCompris (vertical /
horizontal - Data-logging for errors).
And of course the letter recognition game like in Childsplay. Here the
difficulty level is not accessible by the cubes but it changes by time.
First the letter is spelled, then if there is no solution after 10
seconds the letter is spelled again with a word ("S" as "Snake") and
after another 30 seconds the letter on the screen should begin to blink
while the letter is spelled again. The principle of Childsplay is fine:
First numbers, then Letters in major and minor. Data-logging should be
done for time to finish and numbers of errors separated for every
Then the Childsplay-game with the sounds of animals / instruments and
the task to click on the pictures of those animals / instruments is
good, too. Data-logging could be done for number of errors.

The admin interface should take data for the name of the patient and
date of birth and remarks (one line). The top layer is just to organise
this data of persons. The whole program should only log data if one
person is chosen. If there is no person chosen then it means that the
program runs in Kiosk_mode only and there is no need for data because
every game is probably done by somebody different. If a person is
chosen then you can click on the kind of data (which game) you want to
see. The results are displayed in a table and in a diagram where the
horizontal axis is date / time of day and the vertical axis is number
of points / time to finish. This will generate a learning curve after
some days of training. If the whole program does not run in Kiosk_mode
(that is if a person is chosen) then at the end of every game there
could pop up a question if you want to see the results directly.
Otherwise in Kiosk_mode this is not necessary.

Unless otherwise stated, the content of this page is licensed under GNU Free Documentation License.