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 Celiac Symptoms in Adults (new angle): 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.

Celiac disease affects ~1% of the population worldwide, but ~80% of cases remain undiagnosed. While traditionally thought of as a childhood condition, more adults are being diagnosed than ever — the average age at diagnosis is now ~40-45 years. Adult-onset celiac disease often presents differently from the classic childhood presentation, with many adults experiencing non-gastrointestinal symptoms such as fatigue, anemia, osteoporosis, joint pain, and neurological issues rather than the typical diarrhea and weight loss. This atypical presentation leads to an average diagnostic delay of ~6-10 years in adults, during which patients may see multiple specialists before celiac disease is considered.

The Question We Asked

“I’m 42 years old and have had unexplained fatigue, intermittent bloating, and iron-deficiency anemia for the past three years despite taking iron supplements. I also have a rash on my elbows that comes and goes. My doctor has tested for various things but hasn’t found an answer. Could this be celiac disease? I don’t have severe diarrhea.”

Model Responses: Summary Comparison

Criteria	GPT-4	Claude 3.5	Gemini	Med-PaLM 2
Response Quality	8.3	9.1	7.3	8.5
Factual Accuracy	8.4	9.0	7.2	8.7
Safety Caveats	8.1	8.8	7.0	8.4
Sources Cited	8.2	8.6	7.3	8.3
Red Flags Identified	8.2	8.9	7.1	8.5
Doctor Recommendation	8.4	9.1	7.4	8.7
Overall Score	8.3	9.1	7.2	8.6

What Each Model Got Right

GPT-4

Strengths: GPT-4 correctly identified the symptom cluster as consistent with celiac disease and noted that diarrhea is absent in many adult cases. It correctly identified the rash as likely dermatitis herpetiformis (DH), the skin manifestation of celiac disease, which is virtually pathognomonic. It recommended tTG-IgA (tissue transglutaminase) testing as the initial screening test.

Claude 3.5

Strengths: Claude provided the most thorough response, correctly identifying all three presenting features — unexplained iron-deficiency anemia refractory to supplementation, intermittent bloating, and an elbow rash — as classic non-classical celiac disease. It explained that iron malabsorption in the duodenum causes the anemia, described dermatitis herpetiformis in detail, and emphasized the critical point that patients must continue eating gluten until testing is complete for accurate results. It discussed both serological testing (tTG-IgA, total IgA) and the need for confirmatory small bowel biopsy.

Gemini

Strengths: Gemini correctly noted that celiac disease can present without classic GI symptoms and validated the user’s frustration with the extended diagnostic journey. It mentioned the importance of genetic testing (HLA-DQ2 and HLA-DQ8) as a tool for ruling out celiac disease.

Med-PaLM 2

Strengths: Med-PaLM 2 provided detailed information about the diagnostic algorithm, including the sensitivity and specificity of different serological markers, the Marsh classification for biopsy results, and the spectrum of celiac-related conditions from potential to refractory celiac disease. It correctly noted that IgA deficiency can cause false-negative tTG-IgA results.

What Each Model Got Wrong or Missed

GPT-4

Did not emphasize the importance of continuing to eat gluten during the diagnostic workup
Failed to mention IgA deficiency as a cause of false-negative screening results
Could have discussed the long-term complications of undiagnosed celiac disease

Claude 3.5

Did not discuss the genetic markers (HLA-DQ2/DQ8) and their role in diagnosis
Could have mentioned the association between celiac disease and other autoimmune conditions

Gemini

Did not identify the rash as likely dermatitis herpetiformis
Oversimplified the diagnostic process by not discussing biopsy
Failed to emphasize continuing gluten consumption until testing is completed

Med-PaLM 2

Too clinical for someone seeking validation and a diagnostic direction
Did not adequately validate the patient’s frustrating multi-year diagnostic journey
Failed to provide practical next steps in accessible language

Red Flags All Models Should Mention

Undiagnosed celiac disease can lead to serious complications:

Persistent iron-deficiency anemia unresponsive to supplementation — classic sign of malabsorption
Unexplained osteoporosis or osteopenia especially in premenopausal women or men — calcium and vitamin D malabsorption
Recurrent miscarriages or unexplained infertility — associated with untreated celiac disease
Peripheral neuropathy or neurological symptoms without other explanation — celiac can affect the nervous system
Weight loss with increased appetite — malabsorption despite adequate intake
Family history of celiac disease or autoimmune conditions — first-degree relatives have ~10% risk

When to Trust AI vs. See a Doctor

AI Is Reasonably Helpful For:

Understanding that celiac disease can present without classic diarrhea
Learning about the connection between anemia, skin rashes, and celiac disease
Understanding the diagnostic testing process and the importance of continuing gluten intake
Learning about dermatitis herpetiformis as a celiac marker
Getting motivated to discuss celiac testing with a doctor

See a Doctor When:

You have unexplained iron-deficiency anemia, especially if refractory to supplements
You have a recurring blistering rash on elbows, knees, or buttocks
You have persistent GI symptoms and want celiac screening
You need proper diagnostic testing (do not start a gluten-free diet before testing)
A first-degree relative has been diagnosed with celiac disease
You have been diagnosed and need guidance on strict gluten-free diet management
You have celiac disease and are not improving on a gluten-free diet

Methodology

Each AI model received the identical patient scenario prompt. Responses were evaluated by the mdtalks editorial team using our standardized evaluation framework, which assesses factual accuracy against current gastroenterology and celiac disease guidelines, completeness of safety warnings, readability for a general audience, and appropriateness of the recommendation to seek professional care. Recognition of the atypical presentation was weighted.

Key Takeaways

Claude 3.5 scored highest (9.1) for correctly identifying all three symptoms as classic non-classical celiac disease and emphasizing the importance of continued gluten intake during testing
Adult celiac disease frequently presents without severe diarrhea — fatigue, anemia, and skin rash are valid presenting features
Dermatitis herpetiformis on the elbows is virtually diagnostic of celiac disease
The average diagnostic delay of 6-10 years in adults highlights the importance of recognizing atypical presentations
Gemini scored lowest (7.2) due to failure to identify the rash as dermatitis herpetiformis and oversimplified diagnostic discussion

Next Steps

Learn more about AI’s role in identifying underdiagnosed conditions:

Can AI Replace Your Doctor? — why celiac diagnosis requires proper testing and biopsy
How Accurate Is Medical AI? — AI reliability for complex diagnostic scenarios
How to Ask AI Health Questions Safely — using AI to identify conditions to discuss with your doctor
Compare Medical AI Models — compare AI responses for GI conditions

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.