ChatGPT can read an uploaded lab report PDF or photo and explain it in plain language. What it does not do by itself is keep a normalized, source-linked history across labs and years. That is the part Doctano adds, and the connector lets ChatGPT query it without you uploading the same files again.
What ChatGPT does well with a report
- Plain-language summary. It can turn a page of terms into a few sentences you can follow.
- Definitions. It can explain what a marker measures and what an unfamiliar abbreviation means.
- Flag review. It can point at the values marked high or low and list them in one place.
- Question drafting. It can help you write down what to ask at an appointment.
- Two-report comparison. If you upload both, it can compare them, within the limits of its context window.
Where a single-report chat breaks down
- Context is whatever you upload. If the 2023 report is not in the chat, the assistant cannot see the trend.
- No normalization. It may not treat SGOT and AST as the same marker, or convert units consistently across two reports.
- No source link per value. You cannot click a number back to the page it came from.
- Inconsistency. The same question can produce different answers on different days.
- A different privacy boundary. The report now sits in that tool's environment, under its terms and data controls.
- No family structure. Two people become two prompts, not two profiles.
A better prompt is not the fix
You can ask the model to be careful, to cite the report, or to compare only like with like. Those instructions help, but they do not supply the missing structure. The model can only reason over the values, units, dates, and ranges that reach it. With one report, that is one snapshot.
What the question needs is a small database: every result, with its unit, its range, its date, and a link to its source. Then the assistant can answer across years instead of across a single page.
Two ways to give ChatGPT your history
| Approach | What you do | What it gets right | What it misses |
|---|---|---|---|
| Upload each time | Attach one or more PDFs to a chat. | Fast, no setup, works with any assistant. | Manual every time, context limits, no per-value source links. |
| Connect structured data | Authorize Doctano through the MCP connector. | Any-lab history, normalized names and units, links back to reports. | Needs a Doctano account and an MCP-capable client. |
How the Doctano connector works
- Add the connector. Open the connector settings in ChatGPT and add Doctano. The exact label depends on your client and plan.
- Sign in and approve access. Doctano uses OAuth. You choose read access, or read and upload access.
- Ask in plain language. The assistant can list your profiles, read biomarkers and reports, and check the status of a report.
- Upload if you allowed it. With upload access, the assistant can send a PDF or photo to a profile you choose.
- Revoke when you want. Access can be withdrawn without deleting your Doctano account.
With read access, the assistant can see your latest values, changes over time, out-of-range results, and report text. With upload access, it can add a report but cannot edit or delete reports, manage profiles, or change account settings.
What about ChatGPT's own health features?
ChatGPT has added health-specific surfaces and connections to records and wellness data. Availability depends on your plan, your country, and the provider, and it changes. Check OpenAI's current documentation for what your account can connect to.
The structural point does not change with a feature launch: a general assistant answers from the context it has at that moment. A lab history from several labs and several years has to be assembled and normalized before it can be queried well. A report you have as a PDF or a photo still needs to become structured data.
Before you paste or connect
- Know which tool holds the data, and what its retention and deletion settings are.
- Give the narrowest access that answers your question. Read-only is enough for most questions.
- Verify important values against the original report. Extraction and AI output can be wrong.
- Do not use any assistant for an emergency. Contact a doctor or local emergency services.
- Remember that an AI explanation is information, not a diagnosis.
The honest recommendation
Use ChatGPT alone if
- You have one report and want a plain summary.
- You want definitions or help drafting questions.
- You do not need to keep a history.
Connect it to Doctano if
- Your reports span several labs and years.
- You want the assistant to see the whole history.
- You want a source link behind each value.
- You manage reports for more than one person.