LifeFrom
Sign in
Research

Reading lab reports correctly

Last updated November 18, 2026

What we found

Before anything can explain a blood test, it has to read it: every number, unit and range, exactly. We built an open benchmark that measures this, ran our own reader and several AI models through it, and fixed what it found in LifeFrom.

The strongest models we tested made no mistakes; the smaller ones failed silently, and always at the same step: unit conversion (168 of 168 wrong values on text). The number looked plausible and nothing marked it. So in LifeFrom, exact code does the reading and the arithmetic, and AI explains.

On this page
  1. 01 · Why reading comes first
  2. 02 · How we measured it
  3. 03 · Results
  4. 04 · What we changed in LifeFrom
  5. 05 · Limits
  6. 06 · Open

01Why reading comes first

An explanation, a trend or an alert built on a misread value is wrong however well it is written. A German lab prints “2,240” where a US lab prints 2.24; a UK lab reports magnesium in mmol/L, a US lab in mg/dL. Reading a report means getting all of that right, every time, including the results a lab would phone your doctor about.

02How we measured it

We generate synthetic lab reports from results we choose, then print them the way real labs do: US, UK, German, Spanish and Romanian layouts, US or SI units, decimal points or commas, as text, spreadsheets, electronic records (FHIR), PDFs, page images and phone-style photos. Because we made each report, the correct reading of every value is known exactly; no person or AI grades anything. No real patient data is used.

We also print the same results five different ways, to see whether a reader gives the same answer when only the layout changes. The hypotheses and analysis were written down and timestamped before any AI model read the test reports.

03Results

Text reports, printed five ways each: 300 reports and 4,950 values (60 critical) for the first five readers, the first 150 for the last two. “Wrong” counts values read as a different number, with nothing to signal it; “stable” is the share read right in all five layouts of the same results.

ReaderCorrectWrongCritical caughtStable
LifeFrom’s
reader
99.9%060/6099.6%
Claude
Haiku 5.5
99.9%460/6099.5%
GPT-5.6
Luna
99.6%1860/6098.2%
Llama 4
Maverick
97.5%12060/6088.5%
DeepSeek
V4 Flash
86.4%2644/6048.6%
Claude
Sonnet 5.5
100%045/45100%
Gemini
3.8 Flash
100%045/45100%

On phone photos the vision models read as well as on clean pages (Gemini 3.8 Flash: every value right), while reading a photo with on-device text recognition and then our rules dropped from 80.0% to 75.4% correct. Full tables, confidence intervals and every model reply are in the repository.

04What we changed in LifeFrom

  • The reader now handles decimal commas per report (a German creatinine of “2,240 mg/dl” had been read as 2240), more flag styles, about 30 more units and about 60 more Spanish, German and Romanian test names. Across about 236,000 synthetic test values, it went from 78% to 99.9% of values read correctly, with no wrong values.
  • Photos: text recognition read “µmol/L” as “pmol/L”, which left a creatinine unconverted and raised a false critical alert. Fixed: wrong values on clean page images fell from 2.7% to 0.6%. Photos remain the weakest path, and we say so.
  • The design rule that follows: exact code reads and converts; AI explains and never does the unit arithmetic.

05Limits

The reports are synthetic and the generator was written by the same team as the reader, so a high score shows our reader handles every layout we thought of, not every layout that exists. The samples are modest (dozens of critical values per test), and the largest frontier models have not been tested yet. Reading a report correctly is also not the same as interpreting it: LifeFrom informs and does not diagnose; talk to your doctor about your results.

06Open

The generator, the scoring, the preregistration and every result are public under open licences: github.com/caraulani/lab-reading-bench, archived with a citable DOI: 10.5281/zenodo.24341108. Anyone can run their own reader or model through it.

LifeFrom

Sign inHow it worksDownload the appStart on the webAffiliatesResearchLifeFrom Data

Goals

Body performanceMind performanceLive longer, betterSleep deeperProtect your heartLose fat, keep muscleFor testosteroneTrying to conceiveThrough menopause

Read

Blood testUrine testChest X-rayCT scanMRIAll readers

Supplements

Build your protocolMagnesiumCreatineAshwagandhaOmega 3Vitamin B12BerberineAll 59 compounds

Library

BiomarkersSymptomsConditionsGuidesCompoundsBlogThe whole library

Calculators

A1CLDLeGFRTDEEProteinAll 36 calculators

Compare

LifeFrom vs SuperpowerLifeFrom vs Function HealthSuperpower vs Function HealthCompare all three

Legal

TermsTerms of salePrivacyMedical disclaimerAI useResearch program termsConsumer Health Data Privacy

© 2026 LifeFrom, operated by Harvest Sterling LLC, 30 N Gould St, STE R, Sheridan, WY 82801. All rights reserved.

Free on the App Store