AI at the Doctor's Office: The New Patient Advocate

There is little doubt that AI is dramatically changing the face of medicine, from early detection of active illnesses like liver fibrosis and cirrhosis to the ability to predict the onset of over a thousand diseases years before a patient shows any outward symptoms, long before a doctor would think to order a test. AI’s powerful pattern recognition, while no substitute for physician reasoning, acts as a detector that flags signals in patient data indicative of a particular disease.

But the real revolution may be happening in patients’ homes, where people are quietly using AI to sift through bloodwork and medical test results, homing in on the questions and issues that need to be raised at the doctor’s office. AI is becoming an indispensable patient advocate, giving patients and their families the ability to focus their doctors, encourage coordination between specialists, and reduce the frustration that so often accompanies medical visits and treatments.

Patient advocacy is the act of making sure a patient’s needs, rights, and preferences are understood and respected within the healthcare system. In its most basic form, it is what a parent does for a sick child at a doctor’s visit. It evolves into what a spouse, partner, or friend does for us during illness, and often what we, as children, do for our aging parents. It is the art of getting the right medical attention for someone who, in many cases, is not in a position to be fully objective or stay on top of every detail. Face it, even when we are in our peak years, it is hard to advocate for ourselves once we are sitting in that exam room.

Working with doctors is tricky. While I have noticed a trend over the years toward more openness about uncertainty in diagnoses and treatments, there is still often a definitive, and at times overly confident, tone that can get in the way. There is also the familiar “stay in my lane” dynamic, where a cardiologist may hesitate to weigh in on what an oncologist is doing. This can be a real problem for the patient, because no one is fully stepping back to look at the whole picture or consider how one treatment, or lack of treatment, affects another seemingly unrelated issue that is, in reality, deeply connected. You would hope a primary care physician is tying it all together, but that is often not the case.

Of course, it is hard, if not impossible, to keep up with highly trained and specialized physicians, which is part of what has made the patient advocacy role so difficult in the past. AI is changing that.

We can now upload our data, bloodwork, MRI reports, PSA tests, even genetic sequences, into foundation models and structure prompts to analyze it, run statistical comparisons, and ask targeted questions. Privacy advocates will understandably push back and say they do not want their data shared, and they have every right to make that choice. After all, it is their HIPAA-protected data and their own lifespan at stake.

As for me, I am all in. I have already gotten some valuable insights and have lined up a set of very specific questions for my next doctor’s visit.

First, a little background. I have a doctor through One Medical, a low-cost concierge service that gives you prioritized access to doctors, both online and in person. It was purchased by Amazon in 2023 for about $3.9 billion. My experience has been nothing short of fantastic. I have a great doctor who takes a holistic view of health, does not overprescribe medications, and is not shy about recommending lifestyle changes. Yes, he is extremely efficient and I am usually in and out in 20 minutes — it is Amazon after all — but I can message him with follow-up questions.

My current situation provides a good example of how I am using AI to prepare for my upcoming checkup. Like much of the population, I have had stubbornly high cholesterol that I have struggled with over the years. I have been on low doses of statins in the past, but when it started creeping up again a few years ago, despite being fit and generally watching my diet, my doctor was adamant about putting me on a fairly high dose of rosuvastatin. Heart attacks do not run in my family, so I never suspected a high genetic risk there, but strokes are another story. My grandfather had one, and although I suspect his lifestyle may not have been the healthiest, as an immigrant from the Netherlands in the 1920s, I was not taking any chances.

Being on a high dose of rosuvastatin, along with making sure I get enough cardio exercise and watching my diet even more carefully (e.g. little to no red meat), brought my cholesterol way down. At the same time, my A1C started to rise, putting me into the “prediabetic” category. That’s not great.

I started digging into it, with help from AI, and quickly found that high doses of rosuvastatin are associated with insulin resistance, which is a contributing factor to diabetes. That is something you could uncover with a Google search and a few research papers, but of course there is more to the story than that.

By uploading my bloodwork from the past 24 months, which I have collected every three to six months, along with my rosuvastatin dosage and some carefully structured prompts, I was able to generate a detailed analysis. It suggested that the medication is likely a partial contributor to the rise in my A1C, while still indicating that my overall diabetes risk remains low to moderate. More importantly, it framed the situation for what it really is: a set of tradeoffs. The reduction in cholesterol from the statin is likely the right call given the cardiovascular risk, even with some upward pressure on A1C. The analysis also pointed to a possible alternative, combining a reduced statin dose with a drug like ezetimibe, which works by inhibiting cholesterol absorption.

Do I know precisely what the right path forward is? Certainly not. I have a PhD in electrical engineering and like to be referred to as Dr. Eldering, but I am not a physician. As my wife’s bio on one of her social media accounts says, “married to a fake doctor,” so there you have it. I do not know what the right treatment is, and it may very well be no change from the current treatment, but I will know exactly what to ask next week at my physical.

I have used the same approach on more serious issues, including cancer diagnoses. I have never used the results to contradict or challenge a physician, but I have used them to guide conversations and ask more pointed questions. In every case, the additional context, and at times fairly detailed statistical analysis, has proved invaluable. AI has become a powerful patient advocate, and unless you happen to be married to or otherwise related to a rock star physician, a real doctor, it may soon become your advocate as well.

What is changing is not just the technology, it is the dynamic in the room. Patients are walking in better prepared, asking better questions, and understanding the tradeoffs in a way that simply was not possible before. AI is not replacing doctors, but it is quietly leveling the playing field. Doctors are no longer the only ones in the room who have done the homework.