Here are some clues. In principle, you can directly type any character in the groff ms input file, and then tell groff what encoding you used; the character will then be converted into a unicode sequence \[u....].
Typically, most editors allow for a utf-8 encoding, so if you entered characters using it you can just run groff -Kutf-8 (or -Kutf8) to have them converted. Eg typing the euro symbol € on a keyboard might become \[u20AC].
See man groff_char.
The bigger problem is getting the unicode character converted into, presumably, PDF or PostScript. This needs an appropriate font to be installed, and groff to be told to use it. This is where my experience ends. I've seen people suggest you can target the LaTex dvi output (-Tdvi) and use its tools to convert:
groff -Kutf8 -Tdvi -mec -ms test.ms >test.dvi
dvipdfm -cz 9 test.dvi # produces test.pdf
The -mec option replaces the standard fonts by the EC fonts. See man grodvi.
I did write an answer on how I managed to use an alternative method to convert a font that included a particular unicode character, but the original poster did not have any success with it.
Note there is a second man page at man 7 groff with some details.
.AMmacro from the page you linked? Are some of the diacritics/accents you need missing from what.AMsupports?