How One Couple Used AI to Navigate a Brain Tumor Diagnosis (Without Replacing Their Doctors)
When my girlfriend’s prolactinoma—a benign but stubborn pituitary brain tumor—kept coming back, every MRI felt like a cliffhanger we hadn’t agreed to. The medications worked, then stopped. Side effects piled up. Different specialists gave different plans. Somewhere between the waiting rooms and lab portals, I realized I had quietly lost trust in “the system.”
Out of that frustration, I did something that scared even me: I started using artificial intelligence to help fight her tumor—not instead of our doctors, but alongside them. This is a story about what AI can actually do in a complex medical crisis, where its limits are, and how any patient or caregiver can use these tools to ask better questions and make more informed decisions.
Before we go further, a clear boundary: AI is not a doctor. It cannot diagnose, treat, or replace an in-person medical team. But used thoughtfully, it can be a powerful ally in understanding a condition like a prolactinoma, organizing options, and preparing for high‑stakes conversations with specialists.
When a “Benign” Brain Tumor Hijacks Your Life
Amy had just turned 25, but her body was sending signals that didn’t match her age: crushing fatigue, months without a period, bone density slipping into the danger zone. We bounced between clinics. Allergies, stress, “just getting older”—we heard it all. Then an MRI revealed the true cause: a prolactinoma, a prolactin-secreting pituitary tumor pressing on the delicate circuitry at the base of her brain.
The first-line treatment—dopamine agonist medications like cabergoline or bromocriptine—is well established. They often shrink the tumor and normalize prolactin levels. For a while, the script seemed to be working. Her labs improved, symptoms eased. Then, slowly, things drifted off course again.
- Prolactin levels rebounded.
- Side effects worsened—nausea, mood swings, brain fog.
- Different endocrinologists disagreed on dose, duration, and next steps.
That’s when the gap between “guideline” medicine and our lived reality started to widen. On paper, prolactinomas are among the “good” brain tumors—usually benign, often responsive to drugs, surgically resectable if needed. In real life, every new plan carried unknowns: future fertility, long‑term heart valve risks from medication, the danger of under‑treating or over‑treating.
“Patients with pituitary tumors often fall into a gray area where the choices aren’t clearly right or wrong. It’s about understanding trade‑offs, values, and uncertainties.”
— Neuroendocrinologist, tertiary care center (paraphrased from clinical interviews)
Why I Turned to AI When the System Felt Fragmented
I’m not a doctor. I’m an engineer. My instinct when something is broken is to debug it. But healthcare doesn’t come with source code. We had:
- PDFs of MRI and lab reports full of jargon.
- Conflicting recommendations from multiple specialists.
- Scattered notes from hours of late‑night Googling.
Traditional web search gave us fragments: outdated forum threads, dense paywalled studies, and SEO‑driven wellness pages with dubious claims. I wanted something that could:
- Read and summarize complex documents quickly.
- Cross‑reference guidelines and recent research.
- Help us generate better questions for our doctors—not answers to replace them.
Modern large language models (LLMs) trained on medical literature and guidelines promised just that: an always‑on research assistant. Not an oracle, but a high‑bandwidth way to make sense of information we already had and to locate higher‑quality sources we could bring back to our clinicians.
How We Actually Used AI Against a Recurring Prolactinoma
In the article from The Free Press, the author describes gradually building a custom AI “copilot” around his girlfriend’s data. You don’t need to be a programmer to borrow the best parts of that workflow. Below is a breakdown of what he did—and how a typical patient or caregiver could approximate it with off‑the‑shelf tools.
1. Turning Raw Medical Data into Understandable Narratives
The first step was collecting everything:
- MRI and CT reports
- Lab values over time (prolactin, pituitary hormones, bone markers)
- Clinic visit summaries and discharge notes
- A daily symptom and side‑effect log
Using AI, he asked for:
- Plain‑language translations of each report (“Explain this MRI in terms a college student could understand”).
- Lists of key trends (“Summarize how prolactin levels changed over the last 18 months”).
- Clarification of common terms (“What does ‘microadenoma’ vs ‘macroadenoma’ mean?”).
“Once we could actually read her story in one place, the decisions felt less like random guesses and more like understandable trade‑offs.”
— Caregiver, reflecting on using AI to summarize records
2. Stress‑Testing Treatment Plans with Evidence, Not Hunches
Next, he used AI to map proposed treatments against evidence‑based guidelines and major studies. For example:
- Medication decisions: Asking AI to pull and summarize reputable sources (e.g., Endocrine Society guidelines, pituitary tumor center publications) on cabergoline dosing, duration, and relapse rates after stopping therapy.
- Surgery vs. medication: Comparing risks and benefits of transsphenoidal surgery versus long‑term dopamine agonist use, especially for women considering pregnancy.
- Fertility planning: Presenting questions about future IVF, miscarriage risk, and tumor behavior during pregnancy to AI and then cross‑checking cited studies with their doctors.
3. Designing Better Questions for Specialist Visits
Rather than walking into appointments with a vague sense of dread, he used AI to draft structured question lists:
- “Based on these MRI trends, what are the realistic options we should ask about?”
- “What are early warning signs that current treatment is failing?”
- “What lab or imaging schedule is typical for a tumor like this?”
Those lists went into a shared document on his phone. Each visit, they’d open it, prioritize 3–5 key questions, and take notes directly under each one.
4. Monitoring for Red Flags Without Obsessing
With a recurring tumor, it’s easy to spiral: is every headache a sign of growth? Is every bad mood a medication side effect? They asked AI to help create:
- A symptom tracker template emphasizing patterns over isolated events.
- A color‑coded list of “call the doctor now” vs. “mention at next visit” symptoms, based on reputable patient education materials.
- A plain‑language explanation of when vision changes, severe headaches, or hormonal crises require urgent care.
This didn’t eliminate anxiety, but it gave them a shared, structured way to distinguish between normal ups and downs and real emergencies.
What AI Did Well—And Where It Nearly Misled Us
The Free Press story is unusually candid about both the wins and the near‑misses of using AI for something as serious as a brain tumor. That honesty is essential if these tools are going to help more than they hurt.
Where AI Truly Helped
- Speed: Summarizing 20‑page guideline documents into digestible overviews, then letting them dig into specific sections during follow‑up questions.
- Organization: Turning scattered labs and notes into timelines, tables, and bullet‑point histories they could share with new specialists.
- Context: Explaining relative risk (“This complication is rare; this one is more common”) without scaremongering.
- Emotional buffer: Allowing them to process scary information in private first, then approach their doctors with clearer heads.
Where AI Was Weak or Risky
- Hallucinations: On several occasions, AI “cited” clinical trials that did not exist or misinterpreted small case reports as definitive evidence.
- Over‑confidence: It sometimes framed controversial topics (e.g., long‑term cabergoline safety at high doses) as more settled than they are.
- Individual nuance: It could not account for Amy’s unique risk factors, values, or life plans the way a human specialist could.
“Any AI‑generated medical suggestion should be treated as a starting point for discussion, not a conclusion. If it sounds surprising, verify the source and ask your clinician.”
— Adapted from 2024 guidance by major medical informatics societies
A Step‑by‑Step Playbook: Using AI to Navigate a Brain Tumor Diagnosis
Whether you’re facing a prolactinoma or another complex condition, you can borrow this framework and adapt it to your own situation. It assumes you already have a diagnosis from a qualified clinician.
Step 1: Gather and Digitize Your Records
- Request copies of all imaging reports, lab results, and clinic letters.
- Save them as PDFs or clear photos (with personal identifiers stored securely).
- Create a private folder in a secure cloud service or encrypted drive.
Step 2: Ask AI for Plain‑Language Summaries
For each document, you might prompt:
- “Summarize this MRI report in 5 bullet points for a non‑medical reader.”
- “Highlight any findings that usually need close follow‑up.”
- “List terms I should ask my endocrinologist to clarify.”
Step 3: Build a One‑Page Medical Summary
Ask AI to help you condense your story onto one page:
- Diagnosis and date.
- Key imaging and lab trends.
- Treatments tried and how you responded.
- Major side effects or complications.
- Your priorities (e.g., fertility, cognitive function, career).
This is incredibly useful when you seek a second opinion, change providers, or visit a pituitary center of excellence.
Step 4: Use AI to Prepare for Each Appointment
- Share your updated labs or imaging reports (with identifying info redacted if needed).
- Ask AI to suggest 5–10 relevant questions.
- Pick the 3 most important ones to bring to your visit.
Over time, this habit can dramatically improve the quality of your short, precious appointment windows.
Step 5: Double‑Check Everything with Humans
After the visit, you can:
- Ask AI to rephrase your doctor’s plan so you’re sure you understand it.
- Draft a follow‑up message to your clinic portal if you still have questions.
- Keep notes on how you feel on new treatments so your care team has concrete feedback.
Common Obstacles—and How to Avoid AI Backfiring
Using AI in a health crisis can be empowering, but it also introduces new pitfalls. Here’s how to navigate them without burning yourself out or alienating your care team.
Obstacle 1: Overwhelm and Health Anxiety
Constantly querying AI can turn into a form of high‑tech doomscrolling. The more you ask, the more rare complications you learn about.
What helps:
- Set time limits (e.g., 30 minutes twice a week for research).
- Focus on upcoming decisions, not every hypothetical.
- Ask AI for risk ranges and absolute numbers, not just scary names.
Obstacle 2: Tension with Clinicians
Some specialists reasonably worry that AI‑driven self‑research will lead to misinformation or adversarial conversations.
What helps:
- Frame AI as a note‑taking and translation tool, not a rival expert.
- Share your one‑page summary and ask, “What would you correct or add?”
- Invite collaboration: “AI suggested this study—does it apply to my case?”
Obstacle 3: Privacy and Data Security
Uploading sensitive medical documents to any cloud‑based AI tool carries privacy risks. As of 2026, many consumer tools are not HIPAA‑compliant.
What helps:
- Use tools that explicitly state how they handle and store data.
- Redact identifying information when feasible.
- Consider locally running models or hospital‑provided tools where available.
What the Science Says About AI, Brain Tumors, and Decision‑Making
Research on AI in neuroendocrine and brain tumor care is still young, but several themes are emerging from 2023–2025 studies:
- Imaging analysis: AI algorithms can sometimes match or exceed radiologists at identifying and segmenting certain brain tumors on MRI, but these tools are largely experimental and used behind the scenes in research settings, not for patient self‑diagnosis.
- Clinical decision support: Early systems can suggest guideline‑concordant treatment options for complex cases, but they are meant to assist specialists, not replace clinical reasoning.
- Patient education: Studies in oncology and endocrinology show that AI‑generated summaries can improve perceived understanding, but clinicians emphasize the need to verify accuracy and avoid over‑simplification.
For up‑to‑date, trustworthy guidance on prolactinomas and pituitary tumors, consider:
- Endocrine Society
- Pituitary Network Association
- Major academic centers with pituitary programs (e.g., Mass General, Mayo Clinic, UCSF, UCLA)
Before and After AI: What Actually Changed for Us
The tumor did not vanish because of AI. There was no miracle protocol hiding in the data. The outcome for Amy, as of the article’s publication, was the product of standard medicine: expert endocrinologists and neurosurgeons, carefully titrated medications, and regular imaging.
What did change—with AI in the loop—was their experience of the journey:
- They wasted less emotional energy decoding basic terminology.
- They were able to spot inconsistencies in care plans and politely ask for clarification.
- They felt less at the mercy of opaque systems and more like active partners in her care.
“AI didn’t cure her tumor. It changed our posture from ‘Please fix us’ to ‘Help us understand and choose.’ That shift, psychologically, was enormous.”
— Caregiver, reflecting on the role of AI in their journey
Moving Forward: Using AI Without Losing Yourself
If you or someone you love is staring down a diagnosis like a prolactinoma, it’s completely understandable to feel betrayed by your own body—and, at times, by the healthcare system around you. AI won’t fix all of that. But it can give you something rare in medicine: a sense of coherence.
By turning scattered records into a story, dense research into choices, and vague dread into specific questions, AI can help you reclaim a measure of control without pretending to be a doctor. The system may still be imperfect, but you don’t have to walk through it blind.
If you’re ready to start, you might:
- Collect your key records into a single digital folder.
- Ask an AI tool to summarize your diagnosis and current plan in everyday language.
- Use that summary to draft three questions for your next appointment.
None of this guarantees a smooth road. But it can turn you from a passive recipient of care into an informed collaborator—one who uses every tool available, including AI, to protect what matters most: a life that feels like your own, even in the shadow of a brain tumor.
