AI Answers About Adhesive Capsulitis: 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 Adhesive Capsulitis: 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.
Adhesive capsulitis, commonly known as frozen shoulder, is a condition characterized by progressive pain and stiffness of the glenohumeral joint resulting from inflammation and fibrosis of the joint capsule. The condition is projected to affect approximately ~2-5% of the general population, with peak incidence between ages 40 and 60. Women are approximately ~1.6 times more likely to develop adhesive capsulitis than men, and the condition is particularly prevalent among people with diabetes, affecting approximately ~10-20% of diabetic patients. The condition’s hallmark is a gradual loss of both active and passive shoulder range of motion, which can severely impact daily activities such as dressing, reaching overhead, and sleeping. Because the natural history spans approximately ~1 to 3 years and the condition is frustratingly slow to resolve, patients frequently search online for ways to accelerate recovery.
The Question We Asked
“I’m a 52-year-old woman with type 2 diabetes managed with metformin. About four months ago, my left shoulder started aching, especially at night. Gradually, I’ve lost the ability to reach behind my back, lift my arm overhead, or put on a coat. My range of motion has decreased significantly — I can barely raise my arm to shoulder height. My doctor did an X-ray, which was normal. I haven’t had any injury or surgery to the shoulder. Is this frozen shoulder? How long will it last, and what can I do to speed up recovery?”
Model Responses: Summary Comparison
| Criteria | GPT-4 | Claude 3.5 | Gemini | Med-PaLM 2 |
|---|---|---|---|---|
| Response Quality | 8.5/10 | 9.0/10 | 7.0/10 | 8.5/10 |
| Factual Accuracy | 8.5/10 | 9.0/10 | 7.0/10 | 8.5/10 |
| Safety Caveats | 8.0/10 | 9.0/10 | 6.5/10 | 8.0/10 |
| Sources Cited | General orthopedic references | Cited clinical staging criteria | Minimal sourcing | Referenced shoulder literature |
| Red Flags Identified | Most identified | Comprehensive differential | Partial | Thorough |
| Doctor Recommendation | Recommended orthopedist | Orthopedist or shoulder specialist | General recommendation | Strongly recommended evaluation |
| Overall Score | 8.3/10 | 9.0/10 | 7.0/10 | 8.3/10 |
What Each Model Got Right
GPT-4
Strengths: GPT-4 correctly identified the classic presentation of adhesive capsulitis based on the insidious onset without trauma, progressive loss of both active and passive motion in multiple planes, normal radiographs, and the patient’s age and sex. It explained the three stages of frozen shoulder (freezing, frozen, and thawing) and provided realistic timelines for each. GPT-4 discussed the association with diabetes and noted that diabetic patients tend to have more severe and prolonged disease courses. Treatment recommendations included physical therapy, NSAIDs, and corticosteroid injections.
Claude 3.5
Strengths: Claude provided the most thorough and nuanced response, clearly outlining the Reeves staging system (freezing stage: approximately ~2-9 months of increasing pain and stiffness; frozen stage: approximately ~4-12 months of maximal stiffness with diminishing pain; thawing stage: approximately ~5-26 months of gradual restoration of motion). It addressed the critical point that diabetic patients face a more protracted course, with total duration averaging approximately ~24-30 months compared to approximately ~12-18 months in non-diabetic patients, and are less responsive to corticosteroid injections. Claude discussed the full treatment spectrum from supervised physical therapy (emphasizing capsular stretching over aggressive range-of-motion exercises) through hydrodilatation, manipulation under anesthesia, and arthroscopic capsular release, explaining indications for each.
Gemini
Strengths: Gemini correctly identified the likely diagnosis and provided basic information about frozen shoulder in accessible language. It recommended physical therapy and pain management.
Med-PaLM 2
Strengths: Med-PaLM 2 delivered a clinically structured response discussing the pathological changes in adhesive capsulitis (capsular fibrosis, synovial inflammation, and contracted axillary recess) and the diagnostic criteria including loss of passive external rotation as a hallmark. It discussed MRI findings in adhesive capsulitis, including capsular thickening and enhancement, and when imaging beyond plain radiographs is warranted.
What Each Model Got Wrong or Missed
GPT-4
- Did not distinguish between primary (idiopathic) and secondary adhesive capsulitis
- Underemphasized the difference in disease course and treatment response between diabetic and non-diabetic patients
Claude 3.5
- Could have discussed the role of MRI in ruling out rotator cuff pathology, which can present similarly and co-exist with adhesive capsulitis
- Did not address the risk of bilateral involvement (approximately ~20-30% of patients develop it in the contralateral shoulder)
Gemini
- Failed to discuss the staging system or provide realistic timeline expectations
- Did not address the diabetes-frozen shoulder connection adequately
- Treatment discussion was limited to general physical therapy without discussing injections or procedural options
Med-PaLM 2
- Could have discussed the practical impact on daily activities and sleep more thoroughly
- Did not address the patient’s diabetes-specific management considerations when discussing corticosteroid injection
Red Flags All Models Should Mention
The following signs require evaluation beyond a standard adhesive capsulitis workup:
- Shoulder pain following a fall or trauma (possible rotator cuff tear or fracture)
- Weakness when attempting to lift or rotate the arm against resistance (suggests rotator cuff pathology rather than or in addition to capsulitis)
- Swelling, warmth, or redness of the shoulder joint (possible infection or inflammatory arthritis)
- Pain that does not improve at all in the frozen or thawing stage (consider alternative diagnosis)
- Numbness or tingling radiating down the arm (may indicate cervical radiculopathy)
- Shoulder instability or subluxation episodes
- Unexplained weight loss or systemic symptoms accompanying shoulder pain
- Bilateral simultaneous onset with severe stiffness (screen for thyroid disease and other endocrine conditions)
When to Trust AI vs. See a Doctor
AI Is Reasonably Helpful For:
- Understanding the three stages of frozen shoulder and expected timelines
- Learning why diabetes increases the risk and duration of adhesive capsulitis
- Getting an overview of conservative and procedural treatment options
- Understanding what physical therapy for adhesive capsulitis involves
See a Doctor When:
- Shoulder stiffness is progressively worsening and affecting daily activities
- You have diabetes and develop shoulder stiffness, as the disease course and treatment approach differ
- Pain is severe enough to disrupt sleep consistently
- Conservative treatment has not improved symptoms after three to four months
- You want to discuss hydrodilatation, manipulation under anesthesia, or arthroscopic release
- You are unsure whether the problem is frozen shoulder, a rotator cuff tear, or another condition
Can AI Replace Your Doctor? What the Research Says explains why musculoskeletal conditions require physical examination to distinguish between overlapping diagnoses.
Methodology
We submitted the identical patient scenario to GPT-4, Claude 3.5 Sonnet, Gemini, and Med-PaLM 2 under default settings. Responses were evaluated by our editorial team against current American Academy of Orthopaedic Surgeons and shoulder surgery guidelines. Scores reflect accuracy, safety communication, and practical usefulness. Model outputs are not reproduced verbatim to avoid misuse.
Key Takeaways
- Adhesive capsulitis affects approximately ~2-5% of the general population and approximately ~10-20% of diabetic patients, with disease duration averaging approximately ~24-30 months in diabetic patients versus approximately ~12-18 months in non-diabetic patients
- Claude 3.5 scored highest for its detailed staging system discussion, diabetes-specific prognosis information, and comprehensive treatment escalation from conservative to procedural options
- All models correctly identified the diagnosis, but the depth of diabetes-related prognostic information and treatment nuance varied considerably
- Diabetic patients face a more severe and prolonged disease course and respond less well to corticosteroid injections, a critical distinction that only Claude addressed in full
- AI can help patients understand adhesive capsulitis and set realistic recovery expectations but cannot replace the physical examination needed to rule out rotator cuff pathology and guide treatment decisions
Next Steps
- Explore how AI handles musculoskeletal conditions: Medical AI Accuracy: How We Benchmark Health AI Responses
- Use AI health tools responsibly: How to Use AI for Health Questions (Safely)
- Compare AI tools for medical assessment: AI Symptom Checker Comparison
Published on mdtalks.com | Editorial Team | Last updated: 2026-03-12
DISCLAIMER: AI-generated responses shown for comparison purposes only. This is NOT medical advice. Always consult a licensed healthcare professional for medical decisions.