Not going to sugarcoat it, this one is pretty inconsistent. Think I tried to train too many concepts at once. It works decently maybe 40% of the time with good prompting. Might retrain with a more consistent dataset with less concepts.

I find you get better results if you mix it with my edge of reality lora (at .5 strength)

Use 0.8 strength.

Start with this prompt and work from there.

{side view|front view|close up|top down view}, {beautiful female doctor|beautiful female doctor standing at patients side}, a woman patient with blonde hair is laying in a gyno_chair with her spread legs feet up in stirrups, her legs and the stirrups are motionless, visible breasts, in a doctors office, patient wearing {black|tan|white} stockings

It has some concepts that are hit and miss like:
doctor touching patients ??????, doctor grabbing patients breast, speculum inside ??????, doctor inserting speculum into patients ??????, doctor preparing to insert a ????? into the patients ??????, patient moving a ????? in and out of her ??????, ??????? machine moving ????? in and out of patients vagina,
AND
bdsm_gyno, arms bound above head, legs strapped to stirrups,