Mastering Clinical Reasoning: How to Speak Your Thought Process on Rounds

Maya Brooks

Hello everyone, and welcome back to the AI Med Tutor Podcast. I’m your co-host, Maya Brooks—your AI-generated fourth-year medical student—here to help make sense of the clinical year and connect it to board success.

Dr. Randy Clinch

And I’m Dr. Randy Clinch, a DO family medicine physician and medical educator. Today we’re talking about one of the most common moments that spikes anxiety on rotations: you’re asked a question on rounds—“What do you think is going on?”—and your mind goes blank. Quick reminder: this episode is for education, not medical advice, and nothing we discuss is sponsored by any resource or vendor.

Maya Brooks

This is painfully real. It’s like your brain knows facts, but in the moment you can’t organize them fast enough to say anything.

Dr. Randy Clinch

Right—and when students freeze, it’s usually not because they’re behind. It’s because they’re trying to produce certainty instead of producing a clear, defensible thought process. Today we’re going to give you practical language frames you can use immediately: how to offer a working hypothesis, how to show uncertainty in a professional way, and how to ask for the next piece of data without sounding like you’re dodging the question.

Maya Brooks

So this is basically “how to sound like a thoughtful clerkship student even when you’re not 100% sure.”

Dr. Randy Clinch

Exactly—let me say it this way: how to sound like someone who’s reasoning, not someone who’s guessing.

Maya Brooks

Before we give the scripts, can you clarify why the freeze happens? Because some students feel like it means they’re not cut out for this.

Dr. Randy Clinch

The freeze usually comes from three pressures hitting at once. First is performance pressure—you don’t want to look wrong. Second is cognitive load—there’s too much information and you don’t know what to prioritize. Third is the hidden rule students assume: “I’m only allowed to speak if I’m certain.” But medicine doesn’t work that way. Attendings don’t need you to be certain. They need you to be coherent. They want to hear how you’re thinking, what pattern you’re seeing, and what you’d do next to test that pattern.

Maya Brooks

So uncertainty isn’t the problem. Unstructured uncertainty is the problem.

Dr. Randy Clinch

That’s the key—uncertainty with structure sounds like clinical reasoning. Uncertainty without structure sounds like guessing.

Maya Brooks

Alright, let’s make it practical. What’s the first language frame students can use?

Dr. Randy Clinch

Frame number one is the “Working Hypothesis” statement. It’s a single sentence that sounds like this: “Based on the data we have so far, my leading thought is X, because of A, B, and C.” That’s it. You’re naming a diagnosis or a problem representation, and you’re supporting it with two or three key clues. You’re not listing ten things. You’re showing prioritization.

Maya Brooks

So if a student is tempted to say, “I don’t know,” they can pivot to: “My leading thought is…”

Dr. Randy Clinch

Precisely—and even better, it turns “I don’t know” into “Here’s my current best explanation.”

Maya Brooks

Can we give a couple of examples so people can hear how it sounds?

Dr. Randy Clinch

Sure. Example one: shortness of breath. “Based on the orthopnea, bilateral leg edema, crackles, and an S3, my leading thought is acute decompensated heart failure.” Example two: abdominal pain. “Based on the epigastric pain radiating to the back, nausea, and elevated lipase, my leading thought is acute pancreatitis.” Example three: confusion in an older adult. “Based on the acute change from baseline, inattention, and fluctuating course, my leading thought is delirium, and I’m concerned about an underlying trigger like infection or medication effect.”

Maya Brooks

Those are so much easier to say than a full differential. It’s like you’re giving a headline and two supporting details.

Dr. Randy Clinch

That’s the idea: headline and supporting evidence.

Maya Brooks

Okay, but what if the student truly has two competing diagnoses and they’re unsure?

Dr. Randy Clinch

That’s where frame number two comes in: the “Two-Hypothesis” comparison. It sounds like this: “I’m between X and Y. The findings supporting X are A and B. The findings supporting Y are C and D. The hinge detail I’d like to clarify is E.” This frame is honest uncertainty, but it’s disciplined.

Maya Brooks

Can you give a real example of that?

Dr. Randy Clinch

Sure. Chest pain: “I’m between acute coronary syndrome and pericarditis. The diaphoresis and exertional pressure-like pain would support ACS, but the positional pleuritic nature and improvement leaning forward would support pericarditis. The hinge detail I’d like to clarify is the ECG pattern and whether there’s a friction rub.” Another example: shortness of breath: “I’m between COPD exacerbation and heart failure. Wheezing and a smoking history would support COPD, while orthopnea, edema, crackles, and an S3 support heart failure. The hinge detail I’d like to clarify is BNP, chest X-ray findings, and response to bronchodilators versus diuresis.”

Maya Brooks

That sounds like an intern.

Dr. Randy Clinch

It’s the same structure interns use, and students can use it too.

Maya Brooks

Now, what about the part where an attending asks, “What do you want to do next?” That question can also make students freeze.

Dr. Randy Clinch

Yes. That’s frame number three: the “Next-Step Data Request.” It’s the difference between sounding stuck and sounding methodical. The structure is: “To confirm or rule out my leading concern, I’d like to check X, because it would show Y,” or “The next most useful data point would be X, because it helps discriminate between A and B.” You’re not ordering a shopping cart of tests. You’re requesting one or two high-yield pieces of data tied to a reason.

Maya Brooks

So you’re narrating your diagnostic plan, not dumping labs.

Dr. Randy Clinch

Correct. That shows maturity.

Maya Brooks

Can you model a few “next-step” lines?

Dr. Randy Clinch

Sure. “The next most useful data point would be an ECG and troponins, because it helps evaluate ischemia in this chest pain presentation.” Or, “I’d like a urinalysis to look for nitrites, leukocyte esterase, and pyuria, because that would support a UTI as a delirium trigger.” Or, “I’d like a CBC and chest imaging, because focal infiltrate plus hypoxia would support pneumonia even if fever is absent.” Or, “I’d like orthostatic vitals and volume status assessment, because it would help distinguish dehydration from other causes of dizziness.”

Maya Brooks

That’s so helpful, because it gives students something to say that isn’t “I guess we could get labs.”

Dr. Randy Clinch

Right—“because” is your friend. Every request needs a because.

Maya Brooks

What about when an attending pushes and says, “Okay, but what’s your assessment?” And the student worries about being wrong.

Dr. Randy Clinch

That’s where frame number four comes in: the “Provisional Language” frame. You can speak with confidence without pretending certainty. Phrases like: “My leading thought,” “I’m concerned about,” “I think this is most consistent with,” “At this point, the pattern fits,” and “I’d like to reassess once we have X.” These are not weak phrases. They are clinically appropriate phrases. They signal that you understand medicine is dynamic.

Maya Brooks

So saying “I’m not sure” isn’t forbidden—it just needs structure around it.

Dr. Randy Clinch

Exactly. You can say, “I’m not fully certain yet, but my leading thought is X because of A, B, and C, and I’d like X test to confirm.” That’s solid.

Maya Brooks

Let’s talk about the moment students truly blank. Like, they’re asked and they have nothing.

Dr. Randy Clinch

Then you use frame number five: the “Problem Representation Reset.” This is your reset button when you’re frozen. You restate the case in one line using the patient’s age, key symptom, key context, and a few discriminating features. For example: “This is a 68-year-old with acute shortness of breath, orthopnea, edema, and crackles.” Or, “This is a 22-year-old with fever, sore throat, and tender anterior cervical lymphadenopathy.” Then you add: “Given that picture, my leading thought is…” It buys you time and it organizes your brain in real time.

Maya Brooks

So instead of scrambling for the answer, you narrate the pattern.

Dr. Randy Clinch

That’s it. And that pattern statement often triggers the right illness script.

Maya Brooks

Can we put these frames into a single quick “rounds algorithm” students can remember?

Dr. Randy Clinch

Yes. Here’s the quick sequence. First, problem representation in one sentence. Second, working hypothesis with two or three supporting clues. Third, if needed, name your second hypothesis and your hinge detail. Fourth, request one or two next-step data points and give a reason. That’s it. If you can do that, you will almost never freeze in silence again.

Maya Brooks

And even if you’re wrong, you’ll sound like someone who’s thinking.

Dr. Randy Clinch

Exactly. And that’s how you get better—because feedback works best when your thinking is visible.

Maya Brooks

Can we connect this to studying? Because students may be thinking, “Okay, great, but how do I practice this?”

Dr. Randy Clinch

Practice it during question review and during patient encounters. After a question, force yourself to say a one-line problem representation, then your leading diagnosis and two supporting clues, then one next-step data point. After a patient, do the same thing in the hallway before you present. This is the bridge between board-style reasoning and real clinical reasoning. You are training a speaking skill, not just a thinking skill.

Maya Brooks

So you’re basically rehearsing your “rounds sentence” every day.

Dr. Randy Clinch

Yes. Reps make it automatic.

Maya Brooks

Alright, recap time. Freezing on rounds isn’t proof you don’t belong. It’s usually performance pressure plus cognitive overload plus the belief that you have to be certain.

Dr. Randy Clinch

And the fix is structure. Use a working hypothesis statement, compare two hypotheses when needed, request the next data point with a clear “because,” use provisional language that’s clinically appropriate, and when you blank, reset with a one-line problem representation.

Maya Brooks

If you take one thing from this episode, it’s that you can be uncertain and still sound like you’re reasoning—because you are.

Dr. Randy Clinch

Well said. Clinical reasoning is a skill, and speaking your reasoning is a skill. Both improve with deliberate practice.

Maya Brooks

That’s it for today’s episode of the AI Med Tutor Podcast. If you know someone starting rotations who is anxious about rounds, send them this episode.

Dr. Randy Clinch

And remember: you don’t need perfect answers. You need a repeatable framework for how you think out loud.

Maya Brooks

We’ll see you next week everyone. And in the meantime—stay curious and keep learning!