You can take as a reference these notes (check corollary 5.3.5.1). The corollary states the following:
If $M$ is a differentiable surface of $\mathbb{R}^3$ connected and
compact with Gauss curvature $K \ge 0$ and not identically zero then
$M$ is homeomorphic to a sphere.
My understanding is that one requires differentiability to be able to use Gauss curvature.
Now, the ingredients to arrive to this corollary are the following:
a classification of topological surfaces which you cite and which is stated at theorem 5.3.5.1 of the document.
the Gauss-Bonnet theorem which has been mentionned in the comments and appears in the document at theorem 5.3.4.1.
Realize that homeomorphic implies homotopy equivalent.
In fact, the document cites a stronger result by Hadamard in theorem 5.3.6.2:
If $M$ is an ovalid then the Gauss map $\stackrel{\to}{N}: M \to
\mathbb{S}^2$ associated with any unitary normal $N$ is a
dipheomorphism. In particular, $M$ is dipheomorphic to a sphere.
So as you can see you need quite a bit of machinery to prove the result. However you get to a nice result for ovaloids.