What non-MELD variable most meaningfully improves longer-horizon prognostication in alcohol- associated liver disease, and how would you combine it with MELD-based absolute risk to guide when to escalate care (e.g., transplant referral) beyond MELD alone?