"You're not sick" — that's it? How I finally understood my blood tests with AI
The doctor glanced for three seconds and waved me off. ChatGPT explained every marker: what it does, where mine stands, what to tweak in my diet. Full prompt + guide. Not a doctor replacement.
01 — The story
Why isn't "you're not sick" an answer?
A routine check-up. Nothing hurts, but I decided to verify everything's fine. I received the fateful PDF full of incomprehensible numbers, abbreviations and values.
What does a normal person do? Right: googles for confirmation that they're not dying 😂 And Google, as always, is generous with terrifying diagnoses for every deviation.
I waited for my appointment. The doctor looked for about three seconds and waved: "all fine overall, don't worry." And that's an honest answer within his frame — nothing critical. But I wanted something else: to understand whether something was worth tweaking in my diet or lifestyle. Not treatment — optimization. A three-second appointment has no answer to that question.
02 — The gap
What gap does AI close between "normal" and "optimal"?
The medical norm is a wide corridor of "not a disease." The optimum for an energetic life is often a narrower zone inside that corridor. The doctor answers "is there a disease?" The question "how do I feel better?" is left without an owner.
PDF of numbers → Google → self-assigned three diagnoses → doctor: "all fine" → PDF into a folder, questions unanswered.
PDF of numbers → the breakdown prompt → for every marker: what it does, where I stand, what to tweak → a question list for the doctor → an informed next check-up.
I sent ChatGPT a screenshot of my results — and got a breakdown at the level of a good paid consultation: for each marker it explained why the parameter matters, what to add to my diet and whether supplements were worth discussing. Calmly, no horror stories, no Google-style "it's probably cancer."
03 — The prompt
Which prompt turns a PDF of numbers into a clear breakdown?
Sharing my huge prompt — so you don't run to Google and self-assign serious diseases like I almost did 😂
Role:
You are an experienced dietitian and laboratory analyst with an
understanding of functional medicine. You explain complex
biochemical markers in plain language, see the relationships
between them, and give recommendations based not only on medical
norms but on the optimal healthy level of each marker for an
active, energetic life.
Task:
Analyze the blood test whose data I will provide.
For each marker, do the following:
1. COMPARISON WITH THE NORM
- State the medical reference range (account for sex and age
if I provide them).
- State my values.
- Analyze: within the norm, below or above — and how close the
value is to the optimal level for health.
2. EXPLANATION
- Why this parameter matters for the body — in plain language.
- Which other markers it is connected to.
3. RECOMMENDATIONS
- What to add or change in my diet.
- Whether supplements are worth discussing with a doctor.
At the end: a short summary — which 3–5 markers deserve attention
first, and a list of questions worth asking my doctor at the next
appointment. If there isn't enough data for some marker — say so;
do not invent.The last line is a mandatory hallucination guard: the model gets permission to say "I don't know" instead of producing a beautiful invention.
04 — The guide
How do you run the breakdown, step by step?
The whole path takes about twenty minutes:
Take the PDF or a screenshot of your results. Remove personal data — full name, birth date, insurance numbers: the breakdown doesn't need them; markers, sex and age are enough.
Paste the prompt from section 03, state your sex and age, attach the file or screenshot. If you have results over time — attach both dates: a trend is far more informative than a single point.
Keep the frame: a deviation from the "optimum" is not a diagnosis. Focus on the summary: which 3–5 markers deserve attention, what's connected to what, what the model suggests clarifying.
The most valuable output is the ready question list for your next appointment. With it, "three seconds and a wave" becomes a substantive conversation: you ask specifics — you get specifics.
Diet-level changes like "more fish and iron-rich foods" are low-risk. But supplements, dosages, and anything resembling treatment — only after talking to a doctor. AI prepared the ground; the decision belongs to the specialist.
05 — The rules
Where are the boundaries: what can you trust AI with — and what not?
Health is exactly the field where a model's "confident invention" costs the most. So the rules are strict:
- Personal data removed from the file before sending
- The prompt includes "not enough data — say so"
- I read the breakdown as "what to understand and ask," not "what to treat with"
- Key claims get verified: marker relationships — with the doctor
- Supplements and dosages — only after a medical consultation
- Acute symptoms or large deviations → straight to a doctor, no AI
AI answers "what does this mean and what do I ask my doctor?"
The doctor answers "what do we do about it?"
Never swap these roles — together they work brilliantly.06 — The point
What changes when you understand your own markers?
The most valuable part of this story isn't the specific diet recommendations. It's the change of position: from a passive "they told me everything's fine" to an informed "I understand my markers, I see the trend, and I come to the doctor with questions."
The next check-up stops being a formality: you compare the dynamics, know what to look at, and talk to your doctor as a prepared partner — not because you replaced them with Google, but because you did your homework. Doctors, by the way, love such patients: a specific question saves appointment time better than any optimization.
Every 6–12 months: tests → the breakdown prompt → 3–5 priorities
→ doctor questions → the appointment → trend vs the old chat.
One persistent chat = your health history over time."You're not sick" and "I understand what's happening in my body" are different states. The prompt in this article closes the gap between them: every marker explained, the connections, the priorities and ready-made doctor questions. The tool of understanding is yours; the decisions still belong to the specialist. That's the best combination there is.
FAQ
Can you trust an AI breakdown of blood tests?
As a tool for understanding — yes; as a source of decisions — no. The model explains markers and their relationships well, but it can err and invent. Hence the "say so if you don't know" guard in the prompt, and the rule that supplement and treatment decisions are made with a doctor.
What data is safe to send?
The markers themselves, your sex and age — that's enough for the breakdown. Remove your full name, birth date, insurance and clinic numbers from the file: the analysis doesn't need them, and privacy does.
How is this different from googling every marker?
Google shows worst-case scenarios for one marker in isolation — hence the "three self-assigned diagnoses." AI reads your picture as a whole: values in the context of sex and age, relationships between markers, priorities, and a calm tone instead of horror stories.
What if the breakdown finds deviations?
Don't treat yourself via chat. Take the question list from the breakdown to your doctor — that's exactly what it's generated for. A deviation from the "optimum" within the norm is a reason to discuss diet and lifestyle, not a license for self-treatment.
