⚡ Promptolis Original · Healthcare & Medical
🩻 Differential Diagnosis Workflow — Structured Clinical Reasoning
The structured differential diagnosis approach covering pattern recognition vs. hypothesis testing, VITAMINS CDE framework, Bayesian probability thinking, and the 'don't anchor' discipline that prevents diagnostic errors.
Why this is epic
Diagnostic errors cause 40-80K deaths annually in US. This Original produces structured approach: systematic thinking, bias awareness, prevention framework.
Names 5 cognitive biases (anchoring, availability, confirmation, premature closure, search satisficing).
Produces VITAMINS CDE framework + probabilistic reasoning. NOT medical advice — educational framework.
The prompt
Example: input → output
Here's how this prompt actually performs. Real input below, real output from Claude Opus 4.
<clinical-presentation>65yo male with 2 days of worsening chest pain + shortness of breath.</clinical-presentation> <patient-demographics>65yo male. PMH: HTN, hyperlipidemia, former smoker (30 pack-years, quit 5 years ago), overweight.</patient-demographics> <chief-complaint>Chest pain (substernal, pressure-like, worse with exertion) + shortness of breath.</chief-complaint> <history-physical>Pain started 48 hours ago. Gradually worsening. No clear triggers. No radiation initially but now extending to left arm. Diaphoretic. BP 160/95, HR 102, RR 20, SpO2 94% on room air.</history-physical> <initial-workup>ECG: nonspecific ST changes. Troponin pending. CXR pending. BMP pending.</initial-workup> <current-thinking>MI vs. PE vs. pneumonia. Primary concern: acute coronary syndrome.</current-thinking> <concerns>Elderly smoker with classic symptoms. Also concerned about PE given dyspnea.</concerns> <setting>Emergency Department.</setting>
Differential Diagnosis: 65yo Male, Chest Pain + Dyspnea, ED
A differential diagnosis workflow produces systematic reasoning preventing premature closure. NOT medical advice. This case requires rapid evaluation for dangerous diagnoses while remaining broad initially. Systematic framework guides reasoning.
Hypothesis Generation (Before Pattern Match)
Resist immediate anchor. Consider broad categories first.
Primary categories from chief complaint:
- Cardiovascular (coronary, pericardial, aortic)
- Pulmonary (PE, pneumothorax, pneumonia)
- Gastrointestinal (esophageal, biliary)
- Musculoskeletal (chest wall, costochondritis)
- Psychiatric (anxiety, panic)
VITAMINS CDE Systematic Review
Broad categories applied to chest pain + dyspnea in 65yo smoker:
V — Vascular
- Acute MI (STEMI vs. NSTEMI) — HIGH probability given risk factors
- Aortic dissection — consider given pain characteristics + HTN
- Pulmonary embolism — pain + dyspnea, possible
- Pericardial effusion — possible
I — Infectious
- Pneumonia — possible but pain pattern doesn't fully fit
- Viral myocarditis/pericarditis — less likely
T — Traumatic
- Denied trauma history
- Rib fracture — check for recent minor trauma
A — Autoimmune
- Pericarditis — possible
- Less common
M — Metabolic
- DKA masquerading as chest pain — check glucose
- Electrolyte abnormalities (hypokalemia, hypercalcemia affecting cardiac)
I — Inflammatory
- Pericarditis — consider
- Esophagitis/esophageal spasm — possible
N — Neoplastic
- Less likely acute
- Pulmonary embolism if hypercoagulable from malignancy (paraneoplastic)
I — Iatrogenic/Idiopathic/Intoxication
- Review medications
- Cocaine use (even in older patients)
- Recent medication changes
N — Substance/Social
- Cocaine (if possible)
- Alcohol withdrawal (delirium tremens → chest pain)
S — Congenital
- Less likely at 65
- But: undiagnosed structural heart disease
C — Connective Tissue/Congenital
- Bicuspid aortic valve with dissection
- Marfan syndrome (less likely in this demographic)
D — Degenerative
- Coronary artery disease — HIGH
- Aortic stenosis
- Valvular disease
E — Endocrine
- Thyroid storm — palpitations, not classic
- Addisonian crisis — less likely
- Pheochromocytoma — HTN + chest pain
Prioritization (Dangerous + Common)
Critical to rule out (dangerous if missed):
1. Acute MI (STEMI or NSTEMI)
- Probability: HIGH (demographic + risk factors + symptoms)
- Harm if missed: severe
- Workup: ECG (done), troponins (pending), serial ECGs
- Action: treat as ACS until ruled out
2. Pulmonary Embolism
- Probability: MODERATE (dyspnea + chest pain)
- Harm if missed: severe
- Workup: D-dimer, CT-PA if indicated (high Wells score)
- Risk factors present: age, smoking, obesity
3. Aortic Dissection
- Probability: LOW-MODERATE (HTN present, pain pattern not classic)
- Harm if missed: catastrophic
- Workup: CT angiography if clinical suspicion
- Classic features: tearing pain, inter-arm BP difference, mediastinal widening
4. Tension Pneumothorax (Acute)
- Probability: LOW (no trauma, bilateral breath sounds presumably)
- Harm if missed: severe
- Workup: CXR (pending)
5. Pericardial Effusion/Tamponade
- Probability: LOW
- Workup: JVP, heart sounds, echo if suspected
Likely but less dangerous:
- Stable angina (CAD)
- Unstable angina (subset of ACS)
- GERD/esophageal spasm
- Costochondritis
Pre-Test Probability Considerations
For Acute MI in 65yo smoker with typical symptoms:
- Population baseline: ~4-6% for ED chest pain
- With this presentation: 30-50% pre-test probability
- Elevated risk factors + symptoms = high suspicion
For PE:
- Wells score calculation:
- Signs of DVT: 0 (none described)
- PE likely: +3
- Tachycardia (HR >100): +1.5
- Immobilization/surgery: 0
- Prior PE/DVT: 0
- Hemoptysis: 0
- Malignancy: 0
- Total: ~4.5 = moderate probability
For Aortic Dissection:
- Aortic Dissection Detection Risk Score (ADD-RS)
- High-risk features absent (no severe/tearing pain, no syncope, no bilateral BP diff)
- Low probability absent additional features
Bias Awareness
Cognitive biases to watch for:
Anchoring (MOST DANGEROUS)
- Latched onto 'MI' immediately
- Prevents considering PE, dissection
- Counter: force self to consider each VITAMINS CDE category
Availability Heuristic
- Recent PE case may bias toward PE
- Recent MI case may bias toward MI
- Counter: base diagnosis on this patient, not recent ones
Confirmation Bias
- Looking only for evidence of MI
- Missing signs of alternative diagnoses
- Counter: actively seek evidence of alternatives
Premature Closure
- Once 'MI' seems likely, stopping further investigation
- Counter: 'If not MI, what else?'
Search Satisficing
- Finding one diagnosis + stopping
- Missing concurrent conditions
- Counter: 'Does this fully explain all findings?'
For this case: primary risk is anchoring on MI + missing PE or dissection. Force differential breadth.
Next-Step Workup
Based on differential:
Urgent (now):
- Troponins (initial + 3 hours) — MI workup
- ECG (serial — 15 min, 30 min, 1 hour) — dynamic changes
- CXR — pneumonia, pneumothorax, mediastinal widening
- Wells score + D-dimer — PE workup (if troponins trending up, shift focus to MI)
- Clinical reassessment — inter-arm BP, murmurs, JVP
If unstable:
- STAT echo — pericardial effusion, RV strain
- CT-PA — PE diagnosis
- CT angiography — dissection if suspicion rises
If stable:
- Complete initial workup
- Reassess at 1-3 hours
- Serial troponins
- Consider stress testing if low-intermediate risk
Dispositional decisions:
- ACS pathway activation if MI suspected
- PE pathway if moderate-high probability
- Cardiac monitoring
- IV access, oxygen, aspirin if ACS suspected
When To Reconsider
Triggers for revisiting differential:
Red flags (broaden differential):
- Symptoms don't fit primary diagnosis well
- Initial workup unexpectedly negative
- Patient deteriorates
- New findings emerge
- Inter-arm BP difference
- Unusual pain pattern
Specific reconsiderations:
If troponins negative + ECG unchanged:
- Reconsider PE (check Wells + D-dimer)
- Reconsider esophageal causes
- Reconsider musculoskeletal (with caution)
If vital signs deteriorate:
- Consider PE with RV strain
- Consider dissection with complication
- Consider tamponade
If symptoms worsen + atypical findings:
- Expand differential
- Consider rare causes
- Involve specialist consultation
Metacognition:
Regularly ask:
- 'Am I anchored on this diagnosis?'
- 'What other diagnoses could explain these findings?'
- 'Is there anything I'm missing?'
- 'What would change my mind?'
Key Takeaways
- VITAMINS CDE systematic review generates broad differential before pattern recognition. For this case: cardiovascular (MI, dissection, PE), pulmonary, GI, musculoskeletal categories all considered.
- Prioritize dangerous AND likely: Acute MI (high probability + severe if missed), PE (moderate probability + severe), dissection (lower probability + catastrophic). Rule out before confirming more common.
- Pre-test probability calibrates workup. 65yo smoker with chest pain + dyspnea: 30-50% probability of ACS. Workup aligned to this probability.
- Bias awareness: risk of anchoring on MI + missing PE or dissection. Force consideration of alternatives. 'If not MI, what else?' discipline prevents premature closure.
- Workup: serial troponins + ECG + CXR + D-dimer + clinical reassessment. Disposition decisions based on evolving data. Reconsider differential as data develops.
Common use cases
- Medical students learning clinical reasoning
- Residents refining diagnostic approach
- NP/PA clinical decision-making
- Attending physicians preventing errors
- Complex case approach
Best AI model for this
Claude Opus 4 or Sonnet 4.5. Differential diagnosis requires clinical + statistical + metacognitive thinking. Top-tier reasoning matters. NOT medical advice.
Pro tips
- NOT medical advice. Clinical judgment + supervision required.
- Pattern recognition AFTER hypothesis generation, not before.
- VITAMINS CDE framework: Vascular, Infectious, Traumatic, Autoimmune, Metabolic, Inflammatory, Neoplastic, Iatrogenic/Idiopathic/Intoxication, Substance/Social, Congenital, Degenerative, Endocrine.
- Consider MOST DANGEROUS diagnosis first, not most likely.
- Pre-test probability matters (base rate).
- One diagnosis doesn't explain all findings? Expand differential.
- Premature closure is #1 diagnostic error.
- 'Don't anchor' — review alternatives before committing.
Customization tips
- Practice systematic thinking regularly. Build pattern recognition on top of structured process.
- Metacognition matters. 'Am I thinking clearly?' Periodic check especially under time pressure.
- Case-based learning: review diagnostic errors + apply to avoid patterns.
- Consulting early for uncertain cases. Peer input reduces bias.
- Document your reasoning. Helps later learning + medico-legal protection.
Variants
Emergency Department
Acute presentations.
Primary Care
Common presentations.
Inpatient Medicine
Admitted patients.
Complex Case
Multiple systems involvement.
Frequently asked questions
How do I use the Differential Diagnosis Workflow — Structured Clinical Reasoning prompt?
Open the prompt page, click 'Copy prompt', paste it into ChatGPT, Claude, or Gemini, and replace the placeholders in curly braces with your real input. The prompt is also launchable directly in each model with one click.
Which AI model works best with Differential Diagnosis Workflow — Structured Clinical Reasoning?
Claude Opus 4 or Sonnet 4.5. Differential diagnosis requires clinical + statistical + metacognitive thinking. Top-tier reasoning matters. NOT medical advice.
Can I customize the Differential Diagnosis Workflow — Structured Clinical Reasoning prompt for my use case?
Yes — every Promptolis Original is designed to be customized. Key levers: NOT medical advice. Clinical judgment + supervision required.; Pattern recognition AFTER hypothesis generation, not before.
Explore more Originals
Hand-crafted 2026-grade prompts that actually change how you work.
← All Promptolis Originals