Predict the questions
Most panel questions fall into a few groups: why this problem, why this technology, how you tested it, what its limits are, and what you'd do next. Write five likely questions for each group, using your own chapters and system as the source.
Have a groupmate ask them in random order while you answer standing up. The goal isn't memorized answers. It's knowing where in your paper or your system each answer lives.
Answer in three beats
Give the short answer first, then the reason, then the evidence. “Yes, it works offline. We cache the last sync on the device because the clinic loses signal. In testing, offline records synced within a minute of reconnecting.”
If a question has two parts, say so and answer them in order. It shows you heard the whole question, and it buys you a few seconds to think.
When you don't know
You're allowed to pause. Repeat the question in your own words to confirm it, then breathe. A calm “Let me check that in our results” beats a fast guess.
If you really don't know, say what you do know and how you'd find out: “We didn't measure that. We could test it by…” Panels tend to respect an honest limit more than a confident wrong answer. Never make up numbers.