About readmyscans.com

readmyscans.com is built for two kinds of people: patients trying to make sense of a stack of lab reports, and doctors — residents especially — who need to scan a patient's history quickly without flipping through page after page of PDFs.

Upload a case — lab reports, discharge summaries, imaging reports, prescriptions, progress notes, as PDFs or photos. Everything is read together, in one pass, so the same person can be recognised across documents and a value can be followed across admissions.

Nothing about the clinical content is fixed in advance. There is no built-in list of tests this tool knows about: the model decides what the documents contain and how it should be grouped — a renal picture, an antibiotic course, a culture-and-sensitivity grid, an imaging timeline — and names those groupings itself.

The model decides what matters, how to group it, what to call it, and which view suits it. What it does not do is draw the picture. It hands over typed data — this column is a date, this one is a number in mg/dL with a normal range of 0.6–1.1 — and ordinary code plots it.

That split matters because a chart drawn by a language model can put 13.5 where 3.5 belongs and nothing would catch it. Before anything is plotted, the columns are checked against what the chart needs; if they don't hold up, you get a table instead of a wrong chart.

Because the model decides what is worth highlighting, it necessarily leaves things out — and you would have no way of telling “this wasn't in the file” from “this wasn't judged relevant.” So alongside the highlighted views, every upload also produces an Everything found section holding every value read from your documents, including the normal and unremarkable ones.

Values are kept exactly as printed. A result of “<0.01”, a blood pressure of “150/90”, or a “2+” stays as written rather than being forced into a number.

Medical documents are sensitive, so the simplest way to keep them safe is to never keep them at all. Uploaded files are processed in memory to generate your results and then discarded — there is no database of uploaded documents, and no account is required. See the privacy policy for details.

Automatic reading of a scanned document is not perfect. A smudged number, an unusual layout, or a poor photo can all produce a wrong value. What gets highlighted is also a machine's judgement of what matters, not a clinician's — it can rank the wrong thing first. Always check what you see here against the original documents, and talk to a clinician about what any of it means.