I don't know what you mean by "valid" here. That said, there exist plenty of claims which can't get proven nor refuted which come as necessary for inference in logical or mathematical systems. This no doubt extends to reasoning in science, since mathematical and logic theories do often get used there in one way or another. Each axiom of a correct set of axioms for group theory, ring theory, lattice theory, semigroup theory, field theory, Boolean Algebra, etc. for instance can't get refuted or proven. Each axiom comes as contingent upon the domain of discourse in question, and so qualifies as contingent in the context of the entire predicate calculus. Even in propositional logic, many, many statement forms can't get proven nor refuted. The truth values of those statement forms comes as contingent upon on what truth values the atomic variables take on. As a very simple example, suppose you have a natural deductive system and no axioms. Suppose you want to show that CApqAqp, where "C" denotes the material conditional and "A" disjunction. Though there do exist many ways to prove this in this context, at least most of them, rely on working with a contingent statement form like "Apq".
So, if mean by "valid" you mean "meaningful", then since in logic and mathematics there exist meaningful or valid claims which are not provable or refutable, it stands to reason that there exist valid scientific claims which are not provable or refutable. That said, the scope of a claim needs to get kept in mind. No axiom of say lattice theory can get proven or refuted, and if you lose track of the scope of an axiom, then you might get lead astray and think it can get proven or refuted.