Upload the case report as the only source in a new notebook, then run this before anything else.
Using only the uploaded case as your source, confirm: (1) patient age/sex/presenting complaint, (2) the stage-by-stage sequence of events as documented (presentation → history/exam → workup → intervention → outcome), (3) which data types are present (history, exam, labs, imaging, medications, progression notes), (4) any obvious gaps in the documentation, and (5) information a clinician would normally want at presentation that is not recorded in the source. Flag anything you infer versus anything explicitly stated.
Build the scaffolding. Unpack every jargon term and decision into the mechanism behind it — don't just accept that a step was taken, understand why it was valid.
From the uploaded case only: (1) Summarize the history of presenting illness in 3 sentences. (2) List all active problems as a problem list. (3) List all current medications with dose/route/frequency as documented. (4) Pull out the 5 most abnormal lab or vital sign values and state the normal range for each. (5) State what a "textbook" illness script for the leading diagnosis looks like, and note where this case matches or deviates from it.
Acting as if teaching a new intern: (1) List every clinical term, abbreviation, scoring system, or named threshold used in this case and define each in one or two sentences, including the specific numeric value or range relevant here. (2) For every test ordered or treatment given in the case, explain the underlying physiological or pharmacological mechanism that makes it the correct move at that point — not just what was done, but why it works.
Using only the information present at the initial presentation stage of this case — not the final diagnosis — (1) list the differential a clinician should have considered at that point. (2) Separately, list the dangerous "can't-miss" diagnoses that present similarly to this chief complaint in general, even ones this case does not turn out to be, so the pattern generalizes beyond this one patient.
Quiz me: ask me one question at a time about this case's key facts and the glossary terms we just covered. Wait for my answer before giving the next question. Tell me if I got it right before moving on.
The mechanism is assumed known. Focus on where reasonable clinicians could branch differently, why this path was chosen, and how this case links into the wider network of similar presentations.
Based only on the uploaded case: (1) Generate a ranked differential diagnosis (top 5) with one supporting and one refuting data point for each, drawn from the source. (2) Explain how the ranking of that differential shifted as each new finding came in — which single finding did the most work narrowing it down? (3) Identify the single most useful next investigation to discriminate between the top two, and explain why it beats the alternatives. (4) Flag any drug-drug or drug-disease interaction risk in the current medication list.
Identify 2-4 moments in this case where a reasonable clinician could have gone a different direction. For each: (1) state the path actually taken, (2) state a reasonable alternative, (3) explain why this path was chosen over the alternative. Then: where did management follow standard guideline/textbook approach, and where did it diverge — and was that divergence justified by an atypical feature of this patient, or is it arguably a pitfall worth flagging?
Play devil's advocate: challenge my leading diagnosis. Give me the strongest argument, drawn from this case's actual data, for why an alternative diagnosis on the list could be right instead. Make me defend my choice before agreeing or disagreeing.
What other conditions or presentations does this case overlap with clinically? For each overlapping condition, state the single discriminating feature that would separate it from this case's actual diagnosis, so this case connects into a broader network rather than staying an isolated example.
If this patient deteriorated overnight based on the trajectory in the notes, walk me through: (1) what bedside check I'd do first, (2) what immediate step I'd take, (3) at what point I'd escalate — reasoning only from what's documented, not hypotheticals outside the source.
Consolidate the session — see how the reasoning actually changed as you moved from Intern to Resident lens.
Compare how the answers changed across the Intern-lens and Resident-lens prompts I ran on this case. (1) What did the Resident lens add that the Intern lens didn't? (2) Where did Intern-level reasoning turn out to be incomplete or need revision at the Resident level? (3) What's the one thing worth remembering from each lens for this case?
Pull the takeaways out of this single case into a tagged format, so pearls accumulate across every case you run through this workflow instead of staying trapped in one notebook.
From everything discussed in this notebook, extract 2-3 Intern pearls and 2-3 Resident pearls. Intern pearls should be mechanism-anchored and generalizable, in the format "[Finding/pattern] leads to [action], because [mechanism]." Resident pearls should be judgment-anchored, in the format "[Situation] means deviating from the standard approach when [condition], because [reasoning]." Output each pearl as a single row in this format: Organ system | Topic | Pearl type (Intern/Resident) | Pearl text
Translate the case into standard clinical documentation format — a disciplined write-up exercise, distinct from the reasoning tiers above.
Using only the uploaded case, write a SOAP note: (1) Subjective — the presenting complaint and history as documented. (2) Objective — vitals, exam findings, and lab/imaging results as recorded. (3) Assessment — the working diagnosis or problem list as it stands in the source. (4) Plan — next steps, treatments, and follow-up as documented, distinguishing completed actions from pending ones.