How to Make STEM Documents ADA Compliant (When Standard Tools Can't)
Making STEM Documents ADA Compliant
If you've ever tried to run a chemistry syllabus or a physics problem set through an accessibility checker, you already know the pain.
Standard remediation tools were built for text-heavy documents — memos, reports, HR policies. They can tag headings, flag missing alt text, and set a language attribute. But hand them a PDF full of integrals, molecular structures, or handwritten lecture notes and they fall apart completely.
The result? STEM departments are sitting on thousands of documents that are effectively invisible to students using screen readers — and the ADA Title II deadline is less than two months away.
Why STEM Is the Hardest Accessibility Problem
A typical English department syllabus has paragraphs, headings, maybe a table or two. Remediation is straightforward. But STEM documents contain content that doesn't fit neatly into the standard accessibility framework:
Equations and formulas. A screenshot of a quadratic formula is meaningless to a screen reader. For it to be accessible, the equation needs to be encoded as MathML — a structured markup language that lets assistive technology read math aloud. Most accessibility tools don't even attempt this.
Chemical notation. Reaction mechanisms, Lewis structures, and molecular diagrams require specialized markup. Copying them as images and slapping on "image of chemical equation" as alt text doesn't meet the standard.
Handwritten content. Lecture notes, whiteboard photos, and handwritten problem sets are completely opaque to assistive technology. They need to be converted to real, structured text before any accessibility work can even begin.
Tables with mixed content. STEM tables frequently contain formulas, subscripts, superscripts, and special characters inside cells. Standard table-tagging tools choke on this.
What WCAG 2.1 AA Actually Requires
The standard that ADA Title II references — WCAG 2.1 Level AA — has several criteria that are particularly difficult for STEM content:
1.1.1 Non-text Content. Every image, diagram, and equation needs a text alternative that serves the same purpose. For a complex organic chemistry mechanism, that alternative needs to convey the actual chemistry, not just "diagram."
1.3.1 Info and Relationships. Document structure (headings, tables, lists) must be programmatically determined, not just visually implied. A bold line that looks like a heading but isn't tagged as one fails this criterion.
1.4.5 Images of Text. If text is embedded as an image — which describes most scanned PDFs and all handwritten notes — it must be converted to real text. This is where OCR comes in, but standard OCR mangles mathematical notation.
3.1.1 Language of Page. The document's language must be declared. Simple, but almost universally missing from PDFs exported from LaTeX.
What Actually Works
After months of working with university disability services offices, we've found that the only reliable approach for STEM documents is a pipeline that combines AI-powered OCR with structured document conversion:
Step 1: Extract everything. Use an OCR engine trained on STEM notation — not just general text. This means recognizing LaTeX-style equations, chemical formulas, table structures, and even messy handwriting.
Step 2: Structure it. Convert the raw extraction into properly structured content with tagged headings, MathML equations, table headers, and reading order.
Step 3: Output an accessible format. Generate a Word document (not a PDF) with all accessibility features built in. Word's accessibility checker can then validate the output, and instructors can save as tagged PDF if needed.
We Built a Free Tool for This
We got tired of hearing from disability services directors that they had "no solution for STEM." So we built one.
Our free accessibility converter handles the full pipeline: upload a PDF, image, or photo of handwritten notes, and get back a WCAG 2.1 AA compliant Word document with MathML markup, tagged headings, structured tables, and alt text — in under 30 seconds.
No account required. No file limits during early access. Just upload and download.
It won't solve every edge case — complex multi-page lab manuals with dozens of diagrams will still benefit from manual review — but for the everyday STEM documents that make up the bulk of course materials, it's the fastest path to compliance we've found.
The Clock Is Ticking
Institutions serving 50,000+ people must be WCAG 2.1 AA compliant by April 24, 2026. Smaller institutions have until April 2027. The documents that are hardest to remediate — STEM content — are exactly the ones most likely to be left until last.
If your department is staring at a stack of non-compliant PDFs and doesn't know where to start, start with one document. Upload it. See what comes back. Then decide how to scale from there.
Try the free tool at labnotes.ai/remediate
— The LabNotes.ai Team February 2026