Doing the Work but Not Improving: Fix the Learning Loop

Maya Brooks

Hi everyone, and welcome back to the AI Med Tutor podcast. I’m Maya Brooks, your AI-generated fourth-year medical student co-host. I’m here to bring the student perspective to these conversations—the questions we ask, the challenges we run into, and the things that can make medical school and board preparation feel harder than they need to be. And as always, I’m joined by Dr. Randy Clinch. Dr. Clinch, I want to start today with something I think is especially frustrating for medical students. You’re doing the work. You’re watching the videos. You’re doing the questions. You’re reviewing the explanations. Maybe you’re completing everything your school has assigned. But then you take another COMAT, Shelf exam, COMSAE, NBME, or another set of questions, and your performance really hasn’t changed very much. I think that’s when students start wondering, “What am I doing wrong?”

Dr. Randy Clinch

Hi everyone. I’m Dr. Randy Clinch, a DO family medicine physician and medical educator. And Maya, I think that’s an important problem because the student you’re describing isn’t necessarily avoiding the work. Sometimes they’re working incredibly hard. I’ve seen this throughout my years in clinical medicine and medical education. When people are putting in considerable effort without seeing the outcome they expected, their instinct is often to increase the effort. More hours. More questions. More videos. Maybe another resource. But if the learning process itself isn’t producing adaptation, increasing the volume may simply produce more of the same. So I think the question for today is not, “How can I work harder?” It’s: “What needs to happen differently between the work I’m doing today and the next time I’m asked to perform?” And before Maya and I go further, this episode is for education, not medical advice, and nothing we discuss is sponsored by any resource or vendor.

Maya Brooks

That really resonates with me because I think students often measure whether we’re doing enough by activity. I did forty questions today. I watched three hours of videos. I finished two chapters. I studied for six hours. Those things are easy to count.

Dr. Randy Clinch

They are. And activity matters. You have to engage with the material. But activity is not the same as adaptation. You can complete forty questions without changing the way you think through the forty-first. You can watch three hours of videos without being able to retrieve much of what you watched tomorrow. You can carefully read every q-bank explanation without fixing the reason you missed the question. That’s the distinction I want students to hear today.

Maya Brooks

So let me play the student here. I’m doing my questions every day, but my q-bank percentage has been pretty flat for several weeks. What should I ask myself?

Dr. Randy Clinch

I’d start with four questions. Am I retrieving, or am I mostly re-exposing myself to information? Do I know why I’m missing questions? Am I actually repairing the problem I identified? And am I retesting that repair in a way that proves it transfers?

Maya Brooks

Okay. Let’s take those one at a time. Retrieving versus re-exposing. What’s the difference?

Dr. Randy Clinch

Suppose you miss a question about nephritic syndrome. You read the explanation and think, “Oh, yes. That makes sense.” Then you watch a renal video. Maybe you reread your notes. That’s exposure. Some exposure may be necessary, especially if you have a genuine content gap. But now close everything. Tell me what you know. What does nephritic syndrome look like? What findings should I expect? Why are they occurring? How would I distinguish it from nephrotic syndrome? That’s retrieval.

Maya Brooks

And I think that’s one of those uncomfortable differences. When I’m reading an explanation, everything seems familiar. When I close it and try to explain it myself, suddenly I find out what I actually know.

Dr. Randy Clinch

Yes. And that discomfort is useful information. We’ve discussed this in several earlier episodes, including “Active Learning Techniques for Videos & Podcasts.” The principle applies beyond videos and podcasts. After you consume information, make your brain produce something. Explain it. Predict something. Compare two things. Draw the mechanism. Answer a question without looking. Teach it to someone. Learning needs output.

Maya Brooks

So the first question is whether my study process is producing retrieval. What’s the second?

Dr. Randy Clinch

Do you know why you’re missing questions? This is where I’d return to a framework longtime listeners have heard before: Content. Reasoning. Mechanics.

Maya Brooks

And I think students sometimes assume every miss means, “I didn’t know enough.”

Dr. Randy Clinch

That’s the trap. Imagine three students miss the same question about pulmonary embolism. The first student says, “I really didn’t know the risk factors or presentation.” That sounds like content. The second says, “I was deciding between pulmonary embolism and pneumonia. I knew both diseases, but I chose the wrong one.” Now we may have a reasoning or discrimination problem. The third says, “I knew it was pulmonary embolism, but I didn’t notice that the question was asking for the next step in an unstable patient.” That may be mechanics. Same missed question. Three different diagnoses of the miss.

Maya Brooks

Which means three different fixes.

Dr. Randy Clinch

Right. If you want to go deeper into that framework, listen to “Mastering Miss Logs: A Simple System to Capture and Retest Exam Mistakes.” And if you’re getting questions correct but sometimes realizing afterward that your reasoning wasn’t very solid, go back to “Right Answer, Wrong Reason? Think Like a Clinician.” Because your correct answers can contain learning data too.

Maya Brooks

I like that because I’ve definitely had questions where I click the right answer and think, “Please don’t ask me to explain why.”

Dr. Randy Clinch

Most of us have. And those unsure-correct questions deserve attention because on another day, with a slightly different vignette, that same reasoning may produce a miss.

Maya Brooks

Let’s say I’ve now identified the problem. I keep confusing pulmonary embolism with other causes of acute shortness of breath. What does repair actually look like?

Dr. Randy Clinch

First, notice what you just said. You didn’t say, “I’m weak in pulmonary.” You made the problem smaller. You said, “I have difficulty discriminating pulmonary embolism from competing causes of acute dyspnea.” Now we can prescribe something specific. I might ask you to build or revise a Pattern Card. What is the presentation? What are the key clues? What mechanism explains them? Then compare the pattern against pneumonia, heart failure, or another meaningful competitor. What overlaps? What differs? What is the hinge clue?

Maya Brooks

And for anyone who wants the deeper explanation of that tool, our episode “Pattern Cards: Turn What You’re Learning Into What You’ll Recognize Later” walks through exactly how to build and use them.

Dr. Randy Clinch

Yes. And notice that we’re not making a Pattern Card simply because you missed a question. We’re using one because we’ve diagnosed a pattern-recognition and discrimination problem. Then I might have you do a small set of questions specifically designed to make you distinguish those competing presentations. That’s interleaving. We have an episode called “Interleaving Look-Alike Questions for Better Clinical Reasoning” that goes much deeper into that approach.

Maya Brooks

Here’s where I think students get nervous. If I normally do forty questions a day, and now you’re telling me to spend some of my time repairing five questions I missed, I may feel like I’m falling behind.

Dr. Randy Clinch

That’s an important point. What are you falling behind on?

Maya Brooks

My question count.

Dr. Randy Clinch

And is question count the outcome you’re trying to improve?

Maya Brooks

No. My performance is.

Dr. Randy Clinch

There you go. I’m not suggesting that question volume doesn’t matter. You need enough questions to encounter different presentations and eventually build endurance and perform under mixed conditions. But there’s no prize for completing the most questions. If forty questions generate five important learning problems and you never repair them because you’re rushing to tomorrow’s forty questions, you may simply keep reproducing the same errors.

Maya Brooks

That actually describes something I’ve seen among students. There can be this pressure to say, “I did eighty questions today,” because that sounds productive. But nobody says, “I discovered that I repeatedly confuse these two diagnoses, repaired that distinction, and then proved I could make it correctly later.”

Dr. Randy Clinch

And I’d argue the second statement tells me much more about learning. That brings us to the third question: Did you repair the problem, or did you simply review it?

Maya Brooks

So reading the explanation isn’t necessarily repair.

Dr. Randy Clinch

Correct. The explanation may help you identify what needs repair. But repair requires some change in your knowledge, reasoning, or behavior. If the problem is content, maybe you do a brief precision review and then retrieve the concept again. If the problem is reasoning, perhaps you compare competing patterns and practice discriminating between them. If it’s mechanics, maybe you create a behavioral rule. For example: Task sentence first. Or: Screen for stability. Or: Form a provisional answer before looking at the choices. Then you deliberately practice that behavior.

Maya Brooks

And for students who realize their problem is more about question-taking behavior, we’ve done deeper episodes on that too. “Mastering Missed Questions: Fix Test-Taking Mechanics with Task-Pattern-Hinge-Fix Loop.” And “Stop End-of-Block Panic: A Smarter Timed Question Workflow.”

Dr. Randy Clinch

Those would be good places to go next. But there’s one more step that students often leave out.

Maya Brooks

Retesting.

Dr. Randy Clinch

Yes. Repair isn’t complete because you understand the explanation now. You need evidence that something changed.

Maya Brooks

Let’s go back to our pulmonary embolism example. I’ve reviewed it. I made my Pattern Card. I compared PE with pneumonia and heart failure. I did five targeted questions and got them right. Am I done?

Dr. Randy Clinch

You’re making progress. But I want to know whether the learning transfers. So maybe tomorrow or several days from now, you see those conditions again without being told what the topic is. Then later they appear in a mixed question block. Eventually you encounter them under timed conditions. Think of it as an evidence ladder. Can I retrieve it? Can I answer targeted questions about it? Can I distinguish it from look-alikes? Can I recognize it unexpectedly in a mixed set? Can I still do that under timed conditions?

Maya Brooks

That last part seems especially important for the fourth-year student who’s been studying for boards but still hasn’t reached the benchmark their school requires.

Dr. Randy Clinch

Very much so. If that describes you, the solution usually isn’t simply to accumulate more study activity. You already have a lot of performance data. Use it. Look across your q-bank performance, practice assessments, Miss Log, and other available information. Ask: What keeps showing up? Why is it happening? What is the smallest repair? And how will I prove that repair worked? We went much deeper into that situation in “Still Below Your Benchmark? Focus Your Prep Where It Matters Most.”

Maya Brooks

How does this change for the early third-year student? Because that student may be listening and thinking, “I haven’t even taken most of my COMATs or Shelf exams yet. I don’t have all that data.”

Dr. Randy Clinch

You have less data, but you’re generating it every day. Your patients are teaching you where your knowledge is organized and where it isn’t. Your rotation assignments generate information. Your q-bank generates information. Your COMAT or Shelf eventually generates information. And your own uncertainty is information. Early in third year, I want you primarily building. See patients. Develop illness scripts. Create selected Pattern Cards. Practice questions. Repair what those questions expose. And carry forward a small amount of important learning from previous rotations. If you want a deeper dive into that longitudinal approach, listen to “One Rotation at a Time: Building Toward Shelf, COMAT, and Boards.”

Maya Brooks

So an early third-year student shouldn’t hear this episode and think, “I need a giant board-remediation spreadsheet.”

Dr. Randy Clinch

Please don’t. You’re building your system. And remember something else we’ve discussed: Patients generate patterns. Illness scripts organize them. Pattern Cards compress them. Question banks test them. Miss Logs identify what is still breaking down. Interleaving teaches you to distinguish similar patterns. And mixed timed blocks eventually test whether you can retrieve and apply what you know when nobody tells you what’s coming. You don’t need to be equally focused on every part of that system today. Your stage of training determines where most of your attention belongs.

Maya Brooks

I want to bring up another student response to stagnant performance. Buying something. Because when scores aren’t improving, it is incredibly tempting to think, “Maybe I just need a better resource.”

Dr. Randy Clinch

It is. And sometimes a different resource is appropriate. But before adding one, ask: What problem will this resource solve that my current resources cannot? If you can’t answer that question, don’t assume another resource will solve the problem. We devoted an episode to this called “Beat Resource Overload: Master Medicine with Fewer, Focused Tools.” The goal isn’t to collect the best resources. It’s to use the resources you have deliberately.

Maya Brooks

So if I already have a content resource that explains cardiology well, purchasing another cardiology resource may not help if my actual problem is discrimination.

Dr. Randy Clinch

Correct. You may be prescribing content for a reasoning problem. And that brings us back to the central principle: Diagnose before you prescribe.

Maya Brooks

There’s also an emotional side to stagnant scores that I think is worth acknowledging. When students keep working and the scores don’t move, it can become personal. “I’m bad at standardized tests.” “I’m terrible at cardiology.” “Maybe everyone else understands this except me.” And then we start changing everything at once.

Dr. Randy Clinch

Which makes it even harder to learn from your data. A score is a performance measurement. Use it to ask better questions. What happened? Is this a repeated signal? What type of problem is it? What can I change? Did that change work? Performance should guide your next action, not become your identity.

Maya Brooks

So maybe the student who has been stuck doesn’t need to leave today’s episode with a new study plan. Maybe they need to examine the plan they already have.

Dr. Randy Clinch

I like that. And I’d make it very practical. Look at your recent work and ask four questions. Am I retrieving, or mostly re-exposing? Do I know why I’m missing questions? Am I repairing the problem or simply reviewing the explanation? And have I retested the repair under conditions that demonstrate transfer? If one of those steps is missing, that’s where I’d start.

Maya Brooks

And I’d add something from the student perspective. Don’t try to fix all four tomorrow. If you discover that your question review has mostly been passive, change that. If you discover that you never categorize why you miss questions, start there. If you discover that you keep repairing things but never retest them, change that. One meaningful change is probably more useful than rebuilding your entire study system on Sunday night.

Dr. Randy Clinch

I agree. The smallest effective intervention. Then see what happens. Because if you’re doing the work but you’re not improving, the answer isn’t automatically more work. Sometimes the answer is better feedback between one performance and the next.

Maya Brooks

I think that’s our takeaway for today. Activity is not the same as adaptation. Do the work, yes. But make sure the work changes something. Retrieve instead of only rereading. Diagnose your misses. Repair the actual problem. Retest the repair. And then let the next round of performance tell you what to do next. If one of those areas sounds like where you’re getting stuck, we’ve mentioned several previous AI Med Tutor episodes today that can take you deeper without requiring us to reteach every strategy in this one.

Dr. Randy Clinch

And remember, your goal isn’t to complete the most questions, watch the most videos, or finish every resource you purchased. Your goal is to become progressively better at retrieving what you know, recognizing clinical patterns, reasoning between competing possibilities, and making good decisions under examination conditions. So if your performance has plateaued, don’t immediately turn up the volume. Find out where the learning loop is breaking. Then repair that part.

Maya Brooks

Thanks for joining us for another episode of the AI Med Tutor podcast. And if you’re doing the work but the results haven’t caught up yet, use that as a reason to get more curious about your process—not simply busier. And remember until next time, stay curious, and keep learning.