um.. those are already baked in.
The underpinnings of the MBTI was originally cooked up under a mental illness model.
It's all there to be found, don't be lazy.
I think you're confusing this with the MMPI, which was basically a huge amount of questions given to a control "sane" group, and a group with some sort of mental illness. No matter how stupid the question, if it gave a statistically relevant difference between the two groups, it was kept. The resulting questions were made into the MMPI and is currently the most widely used test to determine if someone has a mental illness, since its origins adhered the most to scientific procedure and it has the largest amount of studies behind it. (I believe this is accurate. Either way, very widely used.)
MBTI was originally created into order to create more efficiency in job distribution during war time- the idea was that if the job fit your personality, you'd be happier and work more efficiently and the work force overall would be a better place. The two women who created the inventory, just to give you an idea, both self-identified as INFJs.
This information can be found in "Gifts Differing" and "Cult of Personality"- the first is by Isabel Briggs-Meyers, while the second is a critical look at various personality typing systems, including MBTI and the MMPI.
Also, comparisons to the Big 5 typing systems show that while there are correlations between most of the traits and MBTI axis, neuroticism is the one trait that doesn't significantly correlate to an MBTI axis.
As for changes in MBTI, the complaint I've run into a lot at a INTP forum was the order of functions didn't accurately describe many people. The way the function order is determined is very elegant, and makes intuitive sense, but at the same time, elegance means little if it doesn't accurately describe reality. I'd really love more flexibility with function order and perhaps some studies to see what sort of function order is most likely.
Also, yes, better descriptions. I'm not sure if introverted/extroverted functions are a part of the original MBTI (please correct me if I'm wrong), but they are a major part in Socionics, and it's a terrific way to get a better sense of how the E/I axis can influence the expression of different functions. Also, better descriptions in general, since I always see people asking, "So what exactly
is sensing/intuition/practically everything?" Right now, it's easy to get a completely different view of an axis/function from someone else because most descriptions are so open. While the MBTI leans very heavily on self-identification, it's not very helpful when everything is very generalized and can very easily lead to a 'Forer effect' situation.