Episode 66
Is the AI Doctor In?
Educators in Medicine,
In this newsletter, we continue our journey through the fundamentals of AI, its applications in medicine, and its transformative role in faculty development and education. Let’s dive into learning.
Let me tell you about a patient I saw recently, we’ll call her Maria. She came in already knowing her diagnosis. She’d typed her symptoms into something online, gotten a confident answer, and spent three days convinced she had something serious. She didn’t. But she’d also missed the part where her real issue, a medication side effect in plain sight. This would have been invisible to any algorithm that didn’t know her full history.
That story kept popping in my head as I explored two new AI health tools making waves: PranaDoc’s AI Doctor and Perplexity Health. Both are sophisticated. Both are well-intentioned.
🩺 PranaDoc’s AI Doctor — The Friendly Front Door
PranaDoc has built something thoughtful. Their AI Doctor is a free, 24/7 symptom checker that can spit out a differential, asks follow-up questions, personalizes its guidance, and helps users decide whether they actually need to see a physician. If they do, a board-certified doctor is available for a $39 video visit, with prescriptions sent directly to the pharmacy.
What I appreciate: it’s honest about its limits. The platform is clear that it’s “not a replacement for professional medical diagnosis,” positioning itself as a smart triage tool rather than a substitute for care. That framing matters to me. In a world where millions delay care due to cost or access, a free tool that helps someone decide whether and how urgently to seek help could be valuable.
What gives me pause: the same thing that always gives me pause. The AI doesn’t know Maria. It doesn’t know her medication list, her health literacy, her anxiety baseline, or the fact that she catastrophizes at 2am because her husband is ill and comes to her senses after some reassurance. Symptom checkers operate on what the patient tells them. We probably agree that patients, are notoriously unreliable at times.
Does this dilute the beauty that is at the heart primary care? There are many similar services. That is for you to decide, but I struggle with it.
🔬 Perplexity Health — When AI Meets Your Actual Data
Perplexity Health is playing a different game entirely, and it’s a fascinating one. This is a platform that connects to your real health data: Apple Health, electronic health records from over 1.7 million care providers, and wearables like Fitbit, Ultrahuman, and Withings (with Oura and Function coming soon). FWIW I do use an Apple Watch and just returned my Whoop. More to come on that if you’re interested, I’m happy to share.
When you ask Perplexity Health a question about your resting heart rate, it can pull your recent activity data, your cardiac history, and your latest bloodwork, all at once. It can generate personalized fitness plans, nutrition guidance, and sleep strategies through AI agents that have actually seen your numbers, not just your complaints.
They’ve also assembled a Health Advisory Board of physicians, researchers, and health tech leaders to pressure-test decisions and clinical safeguards. Health data is encrypted, never used to train AI models, and never sold. Currently rolling out to Pro/Max users in the US.
As a physician, and a data nerd, I find this exciting and complicated. The integration of wearable data with lab values and medical records is something we’ve talked about in medicine for years. Seeing it operationalized is remarkable. But the question I keep asking is: who interprets the interpretation? When a patient sees a trend in their biomarkers and an AI draws a conclusion, the cognitive leap from “here’s what the data shows” to “here’s what you should do” is enormous and that leap has required a physician. This may mean more work, like remote patient monitoring solutions.
💡 What This Means for Us as Educators
Both of these tools are going to walk into our exam rooms before our patients do. It is already happening. Our job as physician educators is to prepare the next generation of clinicians for a world where patients arrive pre-informed, sometimes misinformed, and perhaps attached to what an algorithm told them.
A few things I’d want every resident I train to know:
The AI-informed patient is not the enemy. They engaged with their health. That’s a gift. Our job is to meet them there, validate the curiosity, and place that information into the full context of who they are.
AI is excellent at finding patterns in data. Physicians are trained to know when the pattern doesn’t fit the patient.
Talk to your patients about what they’re using. Ask them. “Did you look this up before you came in?” is an invitation. You might be surprised what they found, and what it tells you about how they experience their own health.
The AI doctor may be in. But so are we.
💌 As always, thanks for reading. Get in touch and let me know your thoughts!
Thank you for joining us on this adventure. Stay tuned for more AI insights, best practices, and more future editions of AI+MedEd.
For education and innovation,
Karim
Share this with someone - have them sign up here.
What feels most important here is that the real issue is not simply whether AI gets the diagnosis right or wrong. It is that these tools begin shaping the patient’s interpretive starting point before the clinical encounter even begins.
By the time the patient reaches us, the symptom is no longer just a symptom. It has already been organized into a provisional story, with its own hierarchy of salience, urgency, and fear. That matters because clinical work often depends not only on correcting the conclusion, but on undoing the frame that made other possibilities harder to see.
AI may be very good at pattern extraction. But medicine still depends on recognizing when the pattern has become too cognitively powerful relative to the person in front of us.