Should We Consider Using AI to Help Treat Refractive MC?

By Wayne Persky

RSS Feed

A dramatic shift is underway in healthcare, and microscopic colitis (MC) patients are part of it. Increasingly, people with complex, chronic conditions are turning to artificial intelligence tools — not out of curiosity, but out of necessity. For many, AI has become a parallel information system used alongside, (or sometimes in place of) traditional medical care, when conventional approaches have failed to resolve their symptoms.

This article addresses a reality many MC patients face — after years of unsuccessful symptom management, dietary trials that don't work, medications that provide only partial relief, and physicians who seem unable to explain why some of us are still suffering, AI tools can seem like a promising alternative source of answers. This trend was recently highlighted in a New York Times article describing patients who, after exhausting traditional medical options, began using AI chatbots to identify overlooked diagnoses and treatment strategies (Astor, 2026, April 2).1

But before you turn to AI in frustration, you need to understand both why it might help and why it can be dangerous.

Why MC patients struggle with the current system:

Modern medicine is built around specialization. You're referred from gastroenterologist to allergist to rheumatologist to endocrinologist, with each specialist focusing narrowly on their domain. But MC — particularly when it doesn't respond well to standard treatment — often doesn't respect these boundaries.

Many MC patients experience symptoms that affect multiple systems, including chronic diarrhea and abdominal pain (gastroenterology), food sensitivities and possible mast cell involvement (immunology/allergy), fatigue and brain fog (potentially endocrine or neurological), electrolyte imbalances affecting heart rhythm (cardiology), bone loss from chronic inflammation and malabsorption (endocrinology), and anxiety or depression from chronic illness (psychiatry).

​Each specialist addresses a single symptom, yet none assembles the full picture. This fragmentation isn't incidental, it's structural. Medical training, reimbursement systems, and clinical workflows all reinforce a pattern where symptoms are assigned to categories, diagnoses are isolated and separated, and cross-system interactions are often overlooked.

For MC patients whose symptoms involve immune dysfunction, gut-brain axis disruption, autonomic nervous system effects, or complex food and medication reactions, this model frequently breaks down entirely. You leave appointments feeling unheard, with your complex reality reduced to "just IBS" or "stress-related symptoms" despite a confirmed MC diagnosis.

What AI offers that the healthcare system often doesn't:

AI tools provide something the fragmented specialist system cannot — the ability to synthesize large amounts of information across many disciplines, without boundaries. You can input your complete symptom history, upload lab results, ask about connections between seemingly unrelated symptoms, explore differential diagnoses, and investigate whether your medication side effects might be worsening your condition.

Unlike a specialist trained in gastroenterology who may have limited knowledge of autonomic dysfunction or mast cell disorders, AI doesn't operate within a single discipline. This makes it particularly appealing for MC cases that involve multiple organ systems, present with fluctuating or atypical symptoms, haven't responded to standard treatment, or seem connected to other conditions your doctors haven't explored.

Real success stories, and their limitations.

The New York Times article describes several cases where AI contributed to meaningful diagnostic breakthroughs. A long COVID patient connected her symptoms to dysautonomia after multiple specialists missed it. Another patient identified mast cell activation syndrome through AI suggestions and later obtained formal diagnosis and effective treatment. A third patient, medically trained herself, used AI to generate differential diagnoses that led to identifying and surgically correcting pelvic congestion syndrome.

These cases share a critical common theme: AI didn't replace physicians—it helped patients ask better questions and pursue overlooked possibilities that they then confirmed with appropriate medical testing and treatment.

However, success is not the norm. The same article makes clear that chatbots can be dangerously wrong. In one study, users reached correct diagnoses less than half the time when relying on AI tools. AI can hallucinate sources and citations, misinterpret lab values in dangerous ways, overemphasize irrelevant findings while missing critical ones, provide false reassurance about serious conditions, and suggest treatments that are contraindicated or unsafe.

Patients interviewed were generally aware of these risks but felt they had little choice. As one patient stated: "Is it a good thing to be depending on AI for medical advice? I don't think so. But it's the option that's available."

Who uses AI successfully?

One of the most important findings is that the most successful AI users tend to have strong analytical or medical skills. In the documented examples, users included a physical therapist and someone with a research background in illness and disability. These individuals could challenge incorrect AI suggestions, filter plausible from implausible diagnoses, recognize when AI was overreaching or fabricating conclusions, and verify information through credible medical sources.

Without that skill set, outcomes are far less reliable and potentially dangerous. If you cannot critically evaluate what AI tells you, you are at serious risk of following harmful advice.

Potentially helpful uses of AI by MC: patients:

Pattern recognition across symptoms: AI can help you identify connections between your GI symptoms and other issues like joint pain, fatigue, skin problems, or cognitive difficulties that might suggest related conditions like mast cell activation or autoimmune overlap.

Trigger identification: You can describe complex symptom patterns after eating certain foods or taking specific medications, and AI can help generate hypotheses about mechanisms (although you must verify these through elimination trials and medical testing, not assume they're correct).

Treatment option exploration: AI can provide information about medications, supplements, or dietary approaches you haven't tried, giving you informed questions to bring to your physician rather than instructions to follow independently.

Lab interpretation context: If you have unusual lab results your doctor hasn't fully explained, AI can suggest possible interpretations (but these should be discussed with your physician, not acted upon alone).

Preparing for appointments: AI can help you organize your symptom timeline, prioritize questions, and anticipate what information your doctor might need, making appointments more productive.

AI cannot and should not:

Replace diagnostic testing: Suggesting you might have bile acid diarrhea doesn't mean you do—you need a SeHCAT scan or therapeutic trial with cholestyramine under medical supervision.

Provide personalized treatment plans: What works for "most MC patients" may be dangerous for you specifically based on your medications, comorbidities, or individual physiology.

Interpret complex or borderline lab results: A magnesium level that's "technically normal" might still be inadequate for someone with chronic diarrhea and absorption issues. But this requires clinical judgment, not algorithm output.

Override medical advice: If your gastroenterologist says not to stop budesonide yet and AI suggests you can, your doctor's recommendation should prevail unless you get a second opinion from another qualified physician.

Diagnose serious conditions: If AI suggests you might have colon cancer, inflammatory bowel disease overlap, or a cardiac arrhythmia, you need immediate medical evaluation, not more AI research.

How to use AI safely if you choose to use it:

If you decide to use AI for medical information despite these risks, follow these essential safety guidelines:

• Never act on AI medical advice without physician consultation. Use AI to generate questions and hypotheses, then verify everything with your doctor before making changes to treatment, diet, or medications.
• Cross-check every claim. If AI cites a study, find and read the actual study. If it mentions a treatment, research it through credible medical sources like PubMed, UpToDate, or medical society guidelines.
• Be skeptical of confidence. AI often sounds very certain about things that are actually controversial, uncertain, or context-dependent in medicine.
• Never use AI for emergencies. Severe dehydration, blood in stool, chest pain, severe abdominal pain, or sudden weakness require immediate medical care, not AI consultation.
• Recognize your knowledge limits. If you cannot critically evaluate what AI tells you, you shouldn't be using it for medical decisions.
• Document what AI suggests. Keep records of AI-generated hypotheses so you can discuss them systematically with your doctor rather than acting on them piecemeal.
• Use AI to empower medical conversations, not replace them. The goal is becoming a more informed patient who can advocate effectively, not becoming your own doctor.

When you absolutely need professional medical care:

Regardless of how frustrated you are with the healthcare system, certain situations require immediate physician involvement:

• Any significant change in symptoms, especially sudden worsening of diarrhea, new blood in stool, or severe abdominal pain
• Electrolyte imbalances, especially affecting potassium or magnesium
• Cardiac symptoms including palpitations, chest pain, or syncope
• Significant unintended weight loss
• Signs of sever dehydration
• New neurological symptoms
• Consideration of any new medication or supplement that could interact with your current regimen
• Any condition AI suggests that requires diagnostic testing or prescription treatment

A Balanced Perspective:

The growing reliance on AI in healthcare reflects two simultaneous truths.

• The technology is imperfect and potentially dangerous — it can mislead, it lacks accountability, and it depends heavily on user skill and critical thinking.
• The healthcare system is failing a subset of patients, especially those with complex, poorly understood conditions, particularly when symptoms cross traditional specialty boundaries.

What we are witnessing is not the replacement of doctors but a workaround for structural gaps in healthcare. AI is being used to generate hypotheses, connect symptoms across systems, identify overlooked conditions, and prepare patients for more informed discussions with physicians. But it is absolutely not a substitute for clinical judgment, diagnostic testing, safe treatment planning, or the irreplaceable value of a physician who knows your complete medical history.

Moving Forward:

If you're considering using AI because conventional care hasn't resolved your MC symptoms, first exhaust appropriate medical options:
​

• Seek a second opinion from a gastroenterologist with specific MC expertise, ideally at an academic medical center.
• Request comprehensive evaluation for conditions that commonly overlap with MC: bile acid diarrhea, mast cell activation, small intestinal bacterial overgrowth, medication side effects, and food sensitivities.
• Work with a dietitian experienced in elimination diets and inflammatory bowel conditions to systematically identify triggers.
• Ask for evaluation by other specialists if you have symptoms suggesting systemic involvement; for example, rheumatology for autoimmune overlap, endocrinology for metabolic issues, immunology/allergy for mast cell concerns.
• Consider clinical trials through ClinicalTrials.gov if standard treatment has failed. Researchers studying MC may have access to newer approaches.
• ​If you've genuinely exhausted these options and still choose to use AI, do so with eyes wide open to both the potential benefits and very real dangers. Use it as a tool to become a better-informed patient and advocate, not as a replacement for medical care. Verify everything, question everything, and never act on AI suggestions without consulting your physician.

It's interesting to note that according to a recent article on the Medscape website, in 2023, about 38% of physicians were using AI. Now, in 2026, that number has jumped to 81% (Whyte, 2026, March 26).2

The Bottom Line:

The rise of AI-assisted medical information seeking is not primarily a story about technology. It's a story about unmet needs. MC patients are not turning to AI because they prefer it to physicians — they're turning to it because their symptoms remain unexplained, their conditions are fragmented across specialties that don't communicate, standard treatments have failed, and their questions remain unanswered.

AI, for all its flaws, offers something that some patients cannot find elsewhere — a system that attempts to synthesize information across boundaries and look at the whole picture. Whether that system reaches accurate or dangerous conclusions is still an open question that depends heavily on how it's used.
​​
But the reason people are using it is painfully clear — when the healthcare system repeatedly fails to help you, you will look for help elsewhere. The solution isn't uncritical embrace of AI, it's fixing the systemic problems that drive desperate patients to seek answers from algorithms in the first place.

References:

1. Astor, M. (2026, April 2). Doctors Couldn’t Help Them. They Rolled the Dice With A.I. New York Times, Retrieved from https://www.nytimes.com/2026/04/02/well/live/ai-illness-claude-chatgpt.html?unlocked_article_code=1.ZFA.tWdj.SkN38Nt5nrCA&smid=url-share

2. Whyte, J. (2026, March 26). Physician AI Adoption Is Surging: We Must Lead Its Integration. Medscape, Retrieved from https://www.medscape.com/viewarticle/physician-ai-adoption-surging-we-must-lead-its-integration-2026a100095u