Could AI allow for a faster medical diagnosis of rare conditions?

Artificial intelligence (AI) in healthcare? You’ve got to be kidding me! What a ridiculous idea! That sounds horrifying. We don’t want computers diagnosing us; we want trained, experienced professionals who can think for themselves.

As someone with hemophilia, I felt this way. Strongly. Until last week.

As I’ve mentioned in recent columns, I’ve been struggling to receive a definitive diagnosis for a gastrointestinal condition that’s been causing me significant issues for a while. Last week, I had an interesting experience involving assistance from AI.

Recommended Reading

May 10, 2024 News by Margarida Maia, PhD

AI-assisted ultrasounds may better detect joint bleeding, inflammation

To set the stage, I need to give you a bit of background. A while back, I had a CT scan to help diagnose the condition. The scan was read at an emergency room and then by my current health system. The latter read the scan as “normal” but didn’t take the specific measurements that would help diagnose my suspected rare condition.

Last week, I took a day off from work and flew to a medical facility in another state to get two more tests done. The first was normal. The second was also read as normal, but I disagreed. As with the CT scan, I knew something might have been missed on this second test. This can often happen with rare conditions that medical professionals don’t see every day.

My suspected condition — superior mesenteric artery syndrome — affects an estimated 0.1%-0.3% of the population, so a doctor may encounter only one or two cases in their entire career. It makes sense that the issue could be missed, especially if a provider doesn’t know what to look for. Still, I was frustrated.

How do I prove the scans aren’t normal?

I planned to write up something to show the radiologists that the scan readings weren’t accurate, but this was intimidating and made me extremely uncomfortable.

Not having any clue where to begin, I decided I would experiment with AI. I uploaded my scans, the radiologist’s report, and a list of my symptoms. Then I asked the program, “Is the radiology report right for this patient’s scans and symptoms?”

The AI program said, “This radiology report is not accurate for the imaging findings shown. The report completely misses the clear pathology demonstrated in these studies.” It then detailed findings that should have been taken into account and suggested a specific diagnosis, even calling my symptoms a “textbook case.”

I wasn’t entirely convinced, so I ran the same question through a different AI program. This one stated, “The radiology report is inaccurate because it appears to have overlooked the most important findings from the studies.” It went on to suggest the same diagnosis, calling my scans and symptoms a “classic clinical presentation.”

Because I like to be certain, I ran the question through a third AI program. It said, “The official report is not accurate for this patient’s imaging. These missed findings directly explain the patient’s symptoms and clinical decline.” The program suggested the same diagnosis as the other two.

None of the AI programs read the scans as normal.

I now see the value of AI in healthcare

This experience changed my opinion of AI in healthcare. I realized it can be a valuable tool, particularly for those of us with rare conditions. AI can help recognize disorders with unique presentations that aren’t widely understood or on most doctors’ radars.

Armed with the knowledge I gathered from AI, I wrote to the hospital system and requested a reread of the scans that included the specific measurements needed to diagnose my suspected syndrome. A patient advocate is helping me navigate this process. It is so important to get the correct diagnosis to obtain the appropriate treatment, which, in my case, will likely involve surgery.

I included the AI summaries in my letter to the hospital system. The programs did a fantastic job analyzing the scans and explaining what was wrong with the original report. They also provided more accurate diagnostic options backed up with solid data.

I do not think AI can replace doctors; it is crucial we continue to connect with caring, empathetic providers who can help us navigate our care. However, I do think AI can be a powerful tool, particularly in hard-to-solve cases. It may help to expedite the diagnosis of rare conditions.

Note: Hemophilia News Today is strictly a news and information website about the disease. It does not provide medical advice, diagnosis, or treatment. This content is not intended to be a substitute for professional medical advice, diagnosis, or treatment. Always seek the advice of your physician or another qualified health provider with any questions you may have regarding a medical condition. Never disregard professional medical advice or delay in seeking it because of something you have read on this website. The opinions expressed in this column are not those of Hemophilia News Today or its parent company, Bionews, and are intended to spark discussion about issues pertaining to hemophilia.