Patient Engagement, AI, Patient Voice
What happens when the insight is wrong?
In April, someone in one of the IBD groups I run got a letter from their healthcare professional. It told them their biologic treatment was switching from a pre-filled syringe to a pen. Same treatment, different device. There was no conversation beforehand, just the letter.
The switch caused a delay. Only a few days, but when you are injecting yourself on a schedule to keep a disease like Crohn's under control, a few days is not nothing. They posted about it in the group. A handful of people replied and said that this had also happened to them.
That is how most of what I know actually surfaces. Not through anything official. Through someone posting and other people recognising it.
It happens with shortages as well. There was a post about bile acid sequestrants being switched because of a shortage. People commented that the same thing had happened to them. One person said they had been left without any medication at all, and that they did not even feel able to leave the house. That one got sorted, eventually, because they called their GP and asked to switch treatment. But for a while, that was someone's life, trapped at home, because they did not have a drug they needed.
Smaller things come through the same way. Someone asked in a group recently why the packaging for their bile acid medication had changed from one pot of 180 pills to two pots of 90. It's no big deal, but these changes happen and patients are left wondering why. In a completely different condition area, someone with ADHD posted this:
"I'm currently taking ADHD meds in a 30mg capsule. Can someone tell me WHY the pharmacy provided me with 20mg capsules AND 10mg capsules and charged me individually!"
None of that is written down anywhere except inside a Facebook group. I honestly don't know if any of it ever catches up. There are hundreds of IBD related journal articles published in just a few months, and I don't have the time or energy to track every new publication. Keeping track of whether something my community has been saying for months ever turns into a formal report or a safety signal would be a full time job on its own. I am about as plugged into this as a patient can be, and I still can't answer that question.
Say a team puts this to an AI model: how are patients responding to being switched from a syringe to a pen, and what does that mean for adherence. The model will most likely read that as a clinical question, about whether the drug still works the same. It can only answer from what has been published, and what has been published says efficacy is equivalent across devices.
What it cannot see is what actually decides whether someone keeps taking the drug. The healthcare team tells the patient it works exactly the same. The patient's own experience tells them something different. That disagreement either goes back into a conversation with an IBD nurse, which never reaches the company that makes the drug, or it goes nowhere and just sits there.
An AI will tell you the drug works the same either way. It cannot tell you that the patient doesn't believe it, and that disbelief is what decides whether they keep taking it.
Here is roughly what that feels like from the inside, from someone living with Crohn's who went through exactly this kind of switch:
"For most people, that might not sound like a big deal. But for those of us living with Crohn's disease, this medication is what keeps us functioning, out of hospital, and able to live some kind of normal life. Missing even one dose isn't just a delay, it can trigger a serious flare, pain, exhaustion, hospital visits, everything we work so hard to keep under control can come crashing back. We shouldn't have to worry about access to the treatment that keeps us well."
A patient who isn't involved in a switch doesn't just risk a side effect. They risk deciding quietly, on their own, not to take the drug at all. Nobody finds out until they are back in, flaring, needing new diagnostics just to confirm there is an active flare before they can even restart treatment.
That's the cost of trusting a confident answer built only from what was written down. My job, and the reason patient experts sit in these conversations at all, is to be the person you can go back to and ask what actually happened. A model can't do that. It wasn't there, and it isn't current.
If you want the fuller argument, including four more gaps like this one, Merakoi's white paper on this has just come together. Worth requesting if your team is wrestling with the same question.
The white paper
Lived Experience Can’t Be Prompted
The full paper: the patient accounts in a patient expert's own words, the automate-or-never table, the references and the method.
✕
