To Adapt or not to Adapt, Rethinking the Value of Medical Knowledge-Aware Large Language Models | ResearchPod