I want to get the merged shape and contour of two intersecting ellipses, but now I can’t remove the redundant lines inside and only keep the contour,

Assume that there are two group classification methods for people,
group_1 = “teacher”, group_2 = “student”；
group_a =”male”, group_b =”female”,
The teacher node is represented by square, and the student node is represented by dot.
The male node is colored in blue, and the female node is colored in pink.
So how should I design the corresponding vis-network code to achieve this function of supporting multiple groups?