|

Introducing Live Indexing: Fully Automated Medical Record Indexing

Introducing Live Indexing: Fully Automated Medical Record Indexing

Medical records do not arrive as a clean record set.

Sometimes they arrive as one giant flattened PDF. Sometimes they arrive as hundreds of separate PDFs from different custodians, portals, firms, and providers.

Either way, the problem is the same: the documents inside are rarely clean, deduplicated, ordered, or ready to review.

Three thousand pages. No reliable boundaries. No consistent naming. No single chronology.

The same ER visit appears four times because four custodians produced it.

A lab report ends halfway down page 841, and a different provider’s progress note starts on the same sheet.

A portal export breaks one visit into separate files.

Fax covers, blank pages, and proof-of-service forms sit between the documents that matter.

Somewhere in there is the treatment history a claim decision depends on.

Before anyone can review the medical story, someone has to reconstruct the record set by hand.

Today we’re launching Live Indexing, which does that reconstruction automatically, as the production arrives, before any review or analysis begins.

It is live in InQuery today.

Every reviewer pays for the same broken layer

Reconstructing a record set is unglamorous but necessary work, and the whole ecosystem does it.

It is the layer everyone works around and nobody owns.

Nurse reviewers spend the first day of a file paginating instead of reviewing.

IME and peer-review physicians burn scarce hours re-reading duplicates.

Adjusters work demand packages where the bills and underlying records do not line up.

Defense counsel bills record organization to carriers who audit every line.

The pattern repeats on every desk:

  • The reviewer’s real skill is judgment, not pagination.
  • The first hours of every file go to sorting, not judging.
  • The most expensive people in the workflow do the least skilled part of it.

We’ve written before about how the decisive fact is usually buried, not missing.

Disorganization is how it stays buried.

In California, the problem has a price

California workers’ comp put a number on disorganization.

$3.00 per page of record review past the included count

California medical-legal fee schedule, 8 CCR § 9795. Duplicates are not exempt.

Under the state’s medical-legal fee schedule, every re-faxed copy of the same ER visit bills like new evidence.

The parties sending records attest to page counts under penalty of perjury.

The evaluator verifies the pages reviewed the same way.

Disorganization there is not friction. It is an invoice, and someone swears to it.

A 3,000-page production carrying 800 duplicate pages is real money on every single evaluation.

We covered the mechanics in our post on AI chronologies for workers’ comp cases.

Most tools skip the layer that matters

The market’s answer to messy records has been tools that read faster.

Summarizers like EvenUp and Supio.

Chronology builders like CaseFleet and Wisedocs.

AI review platforms of every flavor, a category Clio has tracked growing quickly across personal injury practice.

Nearly all of them assume the record set is already organized.

So they inherit whatever chaos they are given.

A summary built on a file with four copies of the same visit is a confident summary of an event that happened once.

A chronology built on loose pages cannot tell you which provider authored the entry on a shared sheet.

The blind spot is not the reading. It is the record set.

Live Indexing fixes the layer everything else is built on.

What Live Indexing does

Live Indexing reconstructs the documents inside a medical production as it arrives, before any downstream review or analysis begins.

Five capabilities do the work.

Document boundary detection

A multimodal segmentation model reads every page: text, layout, letterhead, headers and footers, signatures, and clinical structure.

It marks where each document begins and ends, including boundaries that fall mid-page, which is common in faxed and scanned productions.

It works without page numbers, filenames, or bookmarks, because real productions rarely have clean ones.

It reads the page the way a trained reviewer would, and decides where one record stops and the next starts.

Classification and tagging

Every reconstructed document is enriched with structured metadata: record type, provider, facility, and date of service.

Progress notes, operative reports, imaging, labs, bills, correspondence, and other document types are tagged from the document itself.

Mislabeled and unlabeled productions still index correctly, because the tags come from the content, not the filename.

This is the difference between real sorting and indexing and a Bates stamp with better marketing.

Deduplication at the document level

Duplicates collapse at the document level, not just the page level.

That catches the copies page-hash tools often miss:

  • The re-faxed version with a new cover sheet
  • The second custodian’s copy with different stamps
  • The partial duplicate where 12 of 15 pages reappear inside a larger record

The original stays in the set.

Duplicates are flagged and filtered, not deleted, so the production remains defensible.

Sorting and filtering

Indexing starts on its own the moment a production arrives and runs unattended. That is the live part: no kickoff, no queue, no one driving it.

A few hours later, the set is fully sortable and filterable.

Sort chronologically or by provider.

Filter to every physical therapy note from one clinic across 3,000 pages in seconds.

Pull every document touching the date of injury.

Jump between documents instead of scrolling between pages.

Verified page accounting

Live Indexing counts total pages, unique pages, duplicate pages, and noise pages, including covers, blanks, and separator sheets.

Counts are available by document and by production.

For California med-legal work, that means a page count you would be comfortable attesting to, backed by a duplicate log that shows the work.

From loose pages to a record set

Before: raw productionAfter: Live Indexing
Unit of reviewPage 1 of 3,000Document, with provider, date, and type
DuplicatesRead again, billed againFlagged and filtered, with the original kept
Mid-page boundariesInvisibleDetected and split
Finding one provider’s notesManual scanOne filter
Page countAn estimateTotal, unique, duplicate, and noise, per production

What it looks like in practice

Live Indexing rebuilding a production: boundaries detected, duplicates flagged, every document tagged automatically

A production lands in the morning, and indexing starts on its own. No kickoff, no ticket, no one assigned to sort it.

Over the next few hours, documents take shape in the set: an ER visit here, an operative report there, each one tagged as it resolves.

By early afternoon, the work is done.

The four copies of that ER visit have collapsed into one document with three flagged duplicates.

The lab report that ended halfway down page 841 is its own record now, and so is the progress note that started beneath it.

The blanks, fax covers, and proof-of-service forms are counted and set aside as noise.

Total, unique, duplicate, and noise page counts are final.

The reviewer opens the file, filters to orthopedic records, sorts by date of service, and starts reading.

Nobody paginated anything.

What it changes downstream

Once a production is rebuilt as documents instead of loose pages, everything after it becomes more reliable.

Chronologies build from real encounters

A chronology built on a deduplicated set counts each encounter once, no matter how many custodians produced a copy of it.

Our summarization and chronology work starts from the indexed set, so every entry traces back to one document with one source page.

Absences become visible

Missing-record detection also gets sharper once the set has structure.

Think of a referenced consult that never appears, a bill with no clinical note behind it, or a treatment gap that was easy to miss while the file was unordered.

We’ve covered why missing records decide outcomes and how a proper gap analysis works.

The short version: once you know exactly what arrived, the gaps are much easier to spot.

Review time shifts from organizing to judging

Sorting pages was never a good use of a nurse, physician, adjuster, or attorney. With the organization handled, their time starts at the actual review.

Live Indexing organizes the record. Your experts still read it and make every call.

Who Live Indexing is for

Live Indexing is for teams whose work depends on large inbound medical productions.

TeamThe daily problemWhat changes
WC claims organizations and defense counselPage counts and duplicates are billable, regulated obligationsDefensible counts, duplicates flagged before billing
IME, QME, and peer-review organizationsPhysician hours are the business, and duplicates burn themExaminers start from an organized set
Carriers and TPAsDemand packages where bills and records do not line upRecords and bills reconcile against one indexed set
Legal nurse consultants and record-review teamsThe first day of every file goes to sortingReview starts on day one
Litigation support, MSA/MSP, and record retrievalDeliverables are only as good as the organization layerIndexed output as the deliverable
Plaintiff and defense firmsRecord-heavy dockets with per-page vendor pricingOne organized set feeding every downstream document

Two of these audiences deserve a closer look.

California workers’ compensation teams

The fee schedule math makes this the most direct win.

Each duplicate page flagged is $3.00 off the invoice, and the page count comes with the audit trail the attestation requires.

For these teams, page counts and duplicate handling directly affect cost, compliance, and defensibility.

Physician review organizations

IME and peer-review businesses sell expert hours.

Outsourced services like MOS Medical Record Review exist precisely because those hours are too expensive to spend on sorting.

Retrieval vendors like Record Grabber have long told firms to organize records before building work product.

Live Indexing makes that step automatic, so the physician opens a file that is already organized.

It runs ahead of everything else in InQuery

Live Indexing now runs before every other feature we offer, because chronologies, gap flags, and structured case facts are only as good as the record set underneath them.

Our medical record indexing service delivers the organized set itself: boundaries, tags, duplicate log, and verified page counts.

Everything else builds on top of it.

That is the whole point of medical record intelligence: decision-ready outputs start with a record set you can actually trust.

Send us your worst production

If your team reviews medical records for a living, you have a file in mind right now. Probably the one with four custodians that nobody wanted to open.

Send us that one. We’ll run it through Live Indexing and walk you through the organized set.

Frequently Asked Questions

What is Live Indexing?

Live Indexing is an InQuery capability that reads every page of a medical record production as it arrives, detects where each document begins and ends, tags each document with provider, date, facility, and record type, and collapses duplicates.

The result is a sortable, filterable, document-level record set, ready within hours and without anyone on your team sorting a page.

How is this different from Bates numbering or page-level indexing?

Bates numbering labels pages without knowing what they are.

Live Indexing reconstructs the documents themselves, including boundaries that fall mid-page in faxed and scanned productions.

So instead of just knowing where page 841 is, you know which document it belongs to, who wrote it, and when.

Are duplicate records deleted?

No. Duplicates are flagged and filtered, never deleted.

Every copy stays in the production, linked to its original and fully auditable, so you can always show exactly what was produced and what was consolidated.

That matters anywhere page counts carry legal weight.

Does it work on faxed and scanned records?

Yes. The segmentation model is multimodal and was built for the productions reviewers actually receive.

That includes faxes with cover sheets, skewed scans, portal exports that split one visit into several files, and sheets where one document ends and another begins.

Does the AI make decisions about my claim?

No. Live Indexing only organizes the record set.

The nurse, physician, adjuster, or attorney still reads the records and makes every call, the same as before.

How do I try it?

Send us a production, ideally the messiest one on your desk.

We’ll run it through Live Indexing and walk your team through the results: boundaries, tags, duplicate flags, and the page accounting.

Get started here and we’ll set it up with your team.

Erick Enriquez

Erick Enriquez

CEO & Co-Founder at InQuery

Share this article