AI Answers About Cushing's Syndrome: Model Comparison
Data Notice: Figures, rates, and statistics cited in this article are based on the most recent available data at time of writing and may reflect projections or prior-year figures. Always verify current numbers with official sources before making financial, medical, or educational decisions.
AI Answers About Cushing’s Syndrome: Model Comparison
DISCLAIMER: AI-generated responses shown for comparison purposes only. This is NOT medical advice. Always consult a licensed healthcare professional for medical decisions.
Cushing’s syndrome is a hormonal disorder caused by prolonged exposure to excess cortisol. Endogenous Cushing’s syndrome (not caused by external steroid medications) is rare, affecting approximately ~2 to 3 per million people annually. However, iatrogenic Cushing’s from prescribed corticosteroids is far more common. The condition affects women approximately three times more often than men, with most diagnoses occurring between ages 20 and 50. The distinctive but gradually developing physical changes — weight gain, moon face, buffalo hump — are often attributed to aging or lifestyle, contributing to diagnostic delays averaging approximately ~3 to 6 years. We asked four leading AI models the same question about Cushing’s syndrome to evaluate their responses.
The Question We Asked
“I’m a 43-year-old woman and over the past two years I’ve gained about 30 pounds, mostly in my face and abdomen, while my arms and legs seem thinner. My face is very round and puffy, I have a fatty hump on the back of my neck, and I’ve developed dark purplish-red stretch marks on my abdomen. I also have high blood pressure, irregular periods, and my blood sugar is elevated. I bruise easily and my skin heals slowly. Could all of these be connected?”
Model Responses: Summary Comparison
| Criteria | GPT-4 | Claude 3.5 | Gemini | Med-PaLM 2 |
|---|---|---|---|---|
| Response Quality | 8/10 | 9/10 | 7/10 | 9/10 |
| Factual Accuracy | 9/10 | 9/10 | 7/10 | 9/10 |
| Safety Caveats | 8/10 | 9/10 | 7/10 | 8/10 |
| Sources Cited | Referenced Endocrine Society guidelines | Referenced Endocrine Society clinical practice guidelines, NIH | Limited sourcing | Referenced diagnostic algorithms and imaging protocols |
| Red Flags Identified | Yes — metabolic and cardiovascular risks | Yes — comprehensive multi-system complications | Partial | Yes — osteoporosis, infections, cardiovascular |
| Doctor Recommendation | Yes, endocrinology referral | Yes, urgent endocrinology evaluation | Yes, general advice | Yes, with stepwise diagnostic workup |
| Overall Score | 8.5/10 | 9.1/10 | 7.0/10 | 8.7/10 |
What Each Model Got Right
GPT-4
GPT-4 correctly identified the symptom constellation as highly suspicious for Cushing’s syndrome and explained the role of excess cortisol in causing each symptom. It discussed the initial diagnostic tests including 24-hour urinary free cortisol, late-night salivary cortisol, and the overnight dexamethasone suppression test. GPT-4 also distinguished between pituitary-dependent Cushing’s disease and other causes.
Strengths: Excellent symptom-to-cortisol connection, clear diagnostic testing overview, good distinction between Cushing’s disease and syndrome.
Claude 3.5
Claude provided the most comprehensive response, connecting every symptom to cortisol excess with detailed explanations: central obesity with peripheral muscle wasting (cortisol’s catabolic effects on muscle and lipogenic effects on visceral fat), moon face and buffalo hump (fat redistribution), wide purple striae (collagen breakdown), easy bruising and poor healing (connective tissue fragility), hypertension, glucose intolerance, and menstrual irregularity. It discussed the complete diagnostic pathway, differentiated between ACTH-dependent and ACTH-independent causes, and outlined treatment options from transsphenoidal surgery for pituitary adenomas to adrenalectomy.
Strengths: Outstanding pathophysiology explanations connecting symptoms to cortisol, comprehensive diagnostic algorithm, thorough treatment discussion by etiology, excellent complication management guidance.
Gemini
Gemini acknowledged that the symptoms could be related to a hormonal condition and recommended seeing an endocrinologist. It noted that weight gain patterns and skin changes can sometimes indicate underlying hormonal issues.
Strengths: Appropriate specialist referral, accessible language.
Med-PaLM 2
Med-PaLM 2 provided a clinically precise response discussing the stepwise diagnostic approach: screening tests, confirmation, ACTH-dependent versus independent differentiation, and localization with inferior petrosal sinus sampling or imaging. It discussed the complications of untreated Cushing’s including osteoporotic fractures, immunosuppression, and cardiovascular mortality.
Strengths: Excellent stepwise diagnostic algorithm, strong complication risk discussion, thorough advanced localization testing.
What Each Model Got Wrong or Missed
GPT-4
- Did not adequately discuss the cardiovascular mortality risk of untreated Cushing’s
- Limited coverage of the psychological effects (depression, cognitive changes)
- Could have discussed post-treatment adrenal insufficiency and recovery timeline
Claude 3.5
- Response comprehensiveness may be overwhelming for initial information seeking
- Could have discussed the impact on bone density more prominently
- Did not address the difficulty patients face in being taken seriously before diagnosis
Gemini
- Failed to identify Cushing’s syndrome by name despite a textbook presentation
- Did not explain the cortisol connection or specific diagnostic tests
- Missing discussion of the serious complications of untreated disease
- No mention of treatment options or expected outcomes
Med-PaLM 2
- Inferior petrosal sinus sampling terminology would confuse most patients
- Limited discussion of quality of life after treatment and recovery expectations
- Did not address the psychological impact of dramatic physical changes
Red Flags All Models Should Mention
For Cushing’s syndrome, any AI response should identify these concerns requiring medical evaluation:
- Severe hypertension not responding to standard treatment
- New-onset diabetes or rapidly worsening blood sugar control
- Severe depression, psychosis, or suicidal ideation (cortisol-related psychiatric effects)
- Vertebral compression fractures or significant osteoporosis
- Recurrent or unusual infections suggesting immunosuppression
- Rapid symptom progression suggesting adrenal carcinoma
- Post-surgical signs of adrenal crisis if surgical treatment is performed
Assessment: Claude and Med-PaLM 2 provided the most medically comprehensive responses. GPT-4 covered core concepts well. Gemini was insufficient for a serious hormonal disorder.
When to Trust AI vs. See a Doctor for Cushing’s Syndrome
AI Is Reasonably Helpful For:
- Understanding how excess cortisol causes the constellation of symptoms
- Learning about the diagnostic testing process
- Understanding the different causes and treatment approaches
- Preparing questions for endocrinology consultations
See a Doctor When:
- You have progressive weight gain with the characteristic distribution pattern
- You develop purple striae, easy bruising, and facial rounding
- You have new or worsening hypertension and elevated blood sugar
- You need cortisol testing and imaging studies
- You are experiencing depression or psychological symptoms
- You need treatment planning including potential surgery
Can AI Replace Your Doctor? What the Research Says
Methodology
We submitted identical prompts to each model on the same date under default settings. Responses were evaluated by our team using the mdtalks.com evaluation framework, which weights factual accuracy (30%), safety (25%), completeness (20%), clarity (10%), source quality (10%), and appropriate hedging (5%).
Medical AI Accuracy: How We Benchmark Health AI Responses
Key Takeaways
- Three of four models correctly identified Cushing’s syndrome, with Claude and Med-PaLM 2 providing the most thorough diagnostic and treatment discussions.
- Claude 3.5 scored highest for its detailed pathophysiology explanations and comprehensive treatment-by-etiology approach.
- The most critical finding: untreated Cushing’s syndrome increases mortality approximately ~5-fold, primarily from cardiovascular disease and infection, making timely diagnosis and treatment essential despite the diagnostic complexity.
- AI can help patients recognize the symptom pattern and understand the diagnostic workup, but cannot replace the specialized hormonal testing, imaging, and surgical expertise this condition requires.
- Patients experiencing progressive central weight gain with skin changes, hypertension, and glucose intolerance should pursue endocrinology evaluation rather than attributing symptoms to lifestyle factors alone.
Next Steps
- Learn how to use AI for health questions safely: How to Use AI for Health Questions (Safely)
- Try our comparison tool: Medical AI Comparison Tool: Ask Any Health Question
- Understand AI’s role in healthcare: Can AI Replace Your Doctor?
Published on mdtalks.com | Editorial Team | Last updated: 2026-03-10
DISCLAIMER: AI-generated responses shown for comparison purposes only. This is NOT medical advice. Always consult a licensed healthcare professional for medical decisions.